The Wenchuan earthquake recovery: civil society, institutions, and planning

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
THE WENCHUAN EARTHQUAKE RECOVERY:
CIVIL SOCIETY, INSTITUTIONS, AND PLANNING

By
Jia Lu
University of Southern California
Sol Price School of Public Policy

ii

Dedication

To my mother Fuying Hao and my father Yongjun Lu, whose never-failing love gave me
the strength to keep moving forward making this work possible;
To the people in Sichuan, whose passionate pursuit in making a difference in their lives
through the darkest moments in their lives gave me the courage and faith in striving to
make this work complete.

iii

Acknowledgements

This piece of work could not have been completed without the support and
encouragement of my dissertation committee members—Profs. Tridib Banerjee, Terry
Cooper, and Dan Lynch. First of all, I am deeply grateful to have Prof. Tridib Banerjee as
my dissertation Chair. The depth and breadth of his knowledge and understanding,
particularly in the human and social development side of the planning field has always
encouraged me to think, explore, and take on my own path for professional and personal
growth. I am also deeply thankful for the tremendous amount of support and care from
Prof. Terry Cooper throughout these years of my doctoral study. During some of the most
frustrating and difficult times of my dissertation writing and completion, he has always
been there to provide moral and professional support. It is through Prof. Tridib Banerjee’s
and Prof. Terry Cooper’s patience and trust in me and my work over the years that I learned
the virtue of being persistent and never giving up. My parents and I shared many
memorable experiences during Profs. Tridib Banerjee and Terry Cooper’s visits to China
and each time has been a joyful one. So I want thank the two professors for all their care
for me and my parents. I deeply respect their consistent interest in the Chinese society and
the genuine concern for the betterment of people’s lives.

iv

During the initial stage of my dissertation research, Prof. Robert Stallings offered
his valuable times listening to the development of my ideas related to civil society and the
2008 Wenchuan Earthquake. His in-depth knowledge related to crises and disasters
expanded my understanding of the social and human aspects of how people respond to
calamities and extreme stressful life events. Special thanks to Prof. Rob Olshansky in
sharing my interest in the civil society aspect of disaster recovery, especially during our
field trip to the city center of Dujiangyan just two months after the Wenchuan earthquake.
He has always been willing to offer guidance in helping me explore the various directions
of disaster research related to urban planning. One other person in the disaster research
community that I would like to express my thanks to is Prof. Louise Comfort at the
University of Pittsburgh. Since we have met when I was a master level student at the
University of Pittsburgh, she has been a mentor to me especially in helping me navigate
the most important literatures in disaster research since I developed my interest in the
Wenchuan earthquake recovery.
I am also grateful to have taken Prof. Adrianna Kezar’s qualitative research
methods course from the School of Education at USC. She and her class probed me further
into exploring the possible and the most appropriate research designs to reveal the life
experiences of the people impacted by the Wenchuan earthquake. Her guidance also

v

brought me to rethink of my role as a researcher in the field. The final research method
design of my dissertation could not have come into being without the help of Prof. Julie
Cederbaum from the School of Social Work at USC. I had the opportunity to learn the
research tools from her course to investigate the various designs combining quantitative
and qualitative studies. What I have learned moved my dissertation research forward in
such ways that could not have been imaginable otherwise. Her sense of clarity and focus
every time I saw her not only gave the confidence in my own research, but also set up as a
role model for me to develop as a scholar.
Talking about research methods being implemented in my dissertation study, I owe
Prof. Tom Valente from the Keck School of Medicine at USC a debt of gratitude for his
continuous support for me to learn and grasp social network theory and its analytical tools
together as an important approach to study social processes. Because of his consistent
guidance and help, I was able to analyze and document the actions of Chinese civil society
at a critical historical point of time after the 2008 Wenchuan earthquake. His advice has
always been timely and to the point, which in turn developed my ability to closely follow
through and eventually solving the problems related to social network analysis emerged
throughout the different phases of my study. Special thanks also to the people in the RSiena
online community in collaboratively helping each other out to develop tools in

vi

understanding the dynamics of social networks. While navigating through the SIENA
program to examine my longitudinal network data, I am especially grateful to receive the
help of Kayla de la Haye, a behavioral scientist from the RAND Corporation. I would also
like to thank Prof. Tom A.B. Snijders at the University of Oxford in providing valuable in-
person and online advices in developing the longitudinal analysis of my network data.
During the writing stages of my dissertation, Prof. Tess Cruz from Keck School of
Medicine also offered her moral support for me to overcome some of the writing hurdles.
I am grateful to have the opportunities to help her with health promotion related courses,
which also had an impact on how I understand and explore health issues within the context
of my research.
I am also grateful to have my long time high school friends in China, Liu Wei and
Li Min, to have kept staying in contact with me, supporting me throughout these many
years. Every time we got together, either China or online, we always got to share some
precious moments of joy that could light up our hearts. Thanks also to my dear colleagues
at USC for being there in cheering me on. Over the years, we listened to each other’s
frustrations and shared our thoughts on things that only those who have been through the
process could understand.

vii

The source of my inspiration for my dissertation, of course, originated from the
people who took the action and empowered themselves through relentless formation of
grassroots groups and organizations for the social development of China after the
Wenchuan earthquake. Their collective vision and persistence in developing a vibrant
Chinese civil society have also spurred me on during the times of frustration in completing
this dissertation. In the moments when I was lost in direction and energy to push forward,
I would remind myself of what all of this is for. It is a piece of work by the people and for
the people who took the actions on the ground despite of the extreme catastrophic
circumstances in their lives. This inspiration comes from discovering the potential of what
people together are capable of achieving, through sustained interactions with the purpose
of serving each other.
Lastly and most importantly, I could not have begun and finished this dissertation
without the unfading confidence, encouragement, and support from my parents. It is them
who made it all possible. This piece of work is truly dedicated to my mother, Fuying Hao
and father, Yongjun Lu. My thankfulness and love for them are beyond words. Over these
years, they have been there for me through every single one of my ups and downs. They
shared my every moment of happiness and disappointment. It is because of their love that
I am able to present this dissertation. So THANK YOU mom and dad!

viii

Table of Contents
Dedication

ii
Acknowledgements

iii
List of Tables

xiii
List of Figures

xx
Abstract xxvii

Chapter 1: Introduction 1
The Planning Context of the Wenchuan Earthquake Recovery 1
Research Background and Questions 1
Purpose of the Study 12
Significance of the Study 14

Chapter 2: Theoretical Reflection 16
Disaster Literature 17
The Conceptualization of Disaster and Resilience 17
Disaster Recovery and the Role of Civil Society 21
Civil Society Theory 25
The Conceptualization of Civil Society 25
Three Domains of Planning Theory: Civil Society, the State, and
the Market
29
Institutional Theory 33
The Conceptualization of Institutions 33
Institutional Structuration and Transformation 35
Missing Links: Institutions and Social Change 37
A Capability Approach to Institutional Change in Civil Society
Domain
38
Social Network Theory 40

Chapter 3: Methodology 45
Philosophical Assumptions 45
Research Statement 46
Methodology 51
Research Methods 51
Rationale for Mixed Methods Research 54
Nature of the Data 55
Assumptions 55
Actors, Relations, Networks 59
Data Collection Procedures 62
Network Data Collection 62
Sampling Strategy and Description of Setting 62

ix

Relational Contents Specification 68
Level of Analysis 70
The Survey Questionnaire Method 73
Qualitative Data Collection 77
In-depth Interviews of Group/Organizational Informants 79
Ethnographic Field Observations 84
Documents and Audiovisual Recordings 86
Methods of Data Analysis 88
Sequential Data Analysis 88
Quantitative Data Analysis 90
Analysis of Descriptive Statistics using UCINET 90
The Handling of Relational Data 90
Sociograms and Graph Theory 92
Multi-level Network Analysis 94
Longitudinal Modeling of Network Dynamics using RSIENA 99
Qualitative Data Analysis 102
Rationale for Sequential Data Analysis 102
Coding and Thematic Categorization Using ATLAS.ti 103
Interpretation and Presentation 107

Chapter 4: Tracing Actions and Processes 110
Overview of Actor Characteristics 110
One-mode Network Data Description 110
Two-mode Network Data Description 112
Perceived Institutional Factors for Resilience-building 115
Overview of Network Structure Characteristics 117
Communication Network 117
Collaboration Network 130
Agency Freedom Initiation and Capability Formation 141
Initiation of Agency Action (Out-degree) 141
Pre-earthquake Actions 144
Post-earthquake Actions (Emergency Response) 166
Post-earthquake Actions (Recovery) 195
Summary 211
Incoming Nominations and Status of Prominence (In-degree) 213
Overview 213
Treatment of Actors inside the State and the Market Domains 214
Pre-Earthquake Incoming Nomination Action 216
Power of Influence 216
Communicator and Facilitator 220
“Home-grown” Communicator 229
“Home-grown” Sources of Information 231
Post-earthquake Status of Influence (Emergency Response) 239
Empowered “New-comer” 241
The Rise of Domestic Civil Society Organizations 267
Expansion of Empowerment 269

x

Post-earthquake Status of Influence (Recovery) 272
Capability Formation 274
“Late-comer” Emergence 276
Summary 280

Chapter 5: Enduring Civil Society: Sustainability of Actions 283
Part I. Communication and Persistence of Agency Action 283
Cohesion 283
Structuration of Solidarity 283
Creation of Efficient Communication Channels 286
Emerging Network Boundary 294
Capability Set Formation 296
Strength Formation 306
Structural Formation from Action to Persistence 313
Reciprocity 314
Transitivity 319
Clustering 330
Registration Status Group-external and Group-internal
Ties
339
Structural Foundations of Agency Action 378
Top-down Approach 380
Component 380
K-core Analysis 386
Community Structuration in Information Exchange 394
Bottom-up Approach 406
Cliques 406

Part II. Collaboration and Sustainability of Agency Structures 443
Formation of the Sustenance Structure 443
Reciprocity 444
Transitivity 450
Clustering 456
Registration Status Group-external and Group-internal
Ties
469
Structural Foundations of Sustenance and Support 499
Top-down Approach 499
Component 499
K-core Analysis 505
Community Structuring in Project Collaboration 522
Bottom-up Approach: Structural Transformation of Civil
Society
535
Overview 535
Two-clan Analysis 538

Chapter 6: The Autopoietic Civil Society: Rules of Network 550
Longitudinal Modeling of Network Dynamics 550

xi

Motivation 550
Research Setting and Design 552
Stochastic Actor-based Models (SAB Models) 552
The SIENA Methods in R Statistical System 558
Data Treatment Procedures 559
Some Descriptive Measures 576
Testing and Model Specification 580
Issues with Data Requirement in SAB Models 587
Uniplex Results 593
Communication Network Evolution 593
Properties of Communication Networks 594
Emergency Response 594
Disaster Recovery 601
Collaboration Network Evolution 605
Properties of Collaboration Networks 605
Emergency Response 605
Disaster Recovery 609
General Model for the Role of Civil Society 619
Communication Network Evolution 622
Collaboration Network Evolution 625
Institutional Formality and Registration Similarity in
Reciprocity
628
Multiplex Results 643
Co-evolution of Communication and Collaboration Networks 643
Properties of Emergency Response 644
Properties of Disaster Recovery 646
General Model of Network Co-evolution 649

Chapter 7: Civil Society, State, and Market System 654
Formation and Expansion of Available Capability Set 654
Pre-earthquake Stage 655
Emergency Response 657
Long-term Recovery 670
Clustering across Civil Society, State, and Market Domains 681
Registration Institutional Status and Cross-sector Dynamics 689
Cross-Sector Communication Networks 689
Cross-sector Collaboration Networks 693
Cross-sector Institutional Development in Action: The Case of Actor
#51
702

Chapter 8: Institution-building and Role Formation 734
Solidarity Formation 734
Process of Inclusion 737
Strength Formation 751
Local Centrality 752
Degree Centrality 753

xii

Global Centrality 761
Communication Network Betweenness Centrality 765
Collaboration Network Betweenness Centrality 774
Emerging Resilience Structure: Empowering, Strengthening and
Sustaining
782
Resilience as a Learning Process, Risk Adaptation, and Transformation 794
Role Formation of Chinese Civil Society 832
The Structural Equivalence of Actors 834
Communication Role Structure 837
Collaboration Role Structure 861

Chapter 9: Conclusion 879
Overview 879
Resilience Formation: Chinese Civil Society in Times of Crisis 882
Sources of Power: Agency Freedom and Action 882
Voluntary Coordination and Self-organization Process 885
Institutional Climate: Enablers and Obstacles 887
Capability Formation: Communication and Collaboration 891
Process of Inclusion 892
Emergence of Structural Resilience 896
Self-evolution 901
Civil Society, State, and Market 904
Civil Society Institutions: Role Formation and Diversification 908
An Emerging Theoretical Framework for Civil Society Action in Times
of Crisis
909
Implications of the Study for Planning and Policy-making 917
Limitations of the Study 923
Directions for Future Research 928

Bibliography 934

Appendices
Appendix I. Chapter 1 Appendices 962
Appendix II. Chapter 3 Appendices 975
Appendix III. Chapter 4 Appendices 1050
Appendix IV. Chapter 5 (Part I) Appendices 1074
Appendix V. Chapter 5 (Part II) Appendices 1110
Appendix VI. Chapter 6 Appendices 1140
Appendix VII. Chapter 7 Appendices 1165
Appendix VIII. Chapter 8 Appendices 1198
Appendix IX. Chapter 9 Appendices 1215

xiii

List of Tables
Table 2.1. Use of Social Network Theory in Studying the Wenchuan Earthquake
Recovery
42

Table 4.1. Attribute Data Description 110

Table 4.2. Civil Society Participation in Earthquake Recovery Activities
(70 actors)
113

Table 4.3. Perceived Importance of Institutional Factors for Resilience-building
(70 Actors)
115

Table 4.4. Comprehensive Communication Network Characteristic Measures
Comparing Three Time Periods (138 actors)
118

Table 4.5. Communication Network Three-Period Comparison of Local
Centrality Measures (138 actors)
122

Table 4.6. Comprehensive Collaboration Network Characteristic Measures
Comparing Three Time Periods (138 actors)
131

Table 4.7. Collaboration Network Three-Period Comparison of Local Centrality
Measures (138 actors)
132

Table 4.8. Ranking of Initiation of Communication Agency Action Intensity
(Pre-earthquake)
145

Table 4.9. Traits of Key Actors with High Tie-initiation Action (Pre-Earthquake) 165

Table 4.10. Ranking of Initiation of Agency Action Intensity
(Emergency Response)
168

Table 4.11. Communication Structural Foundation (Pre-Earthquake Incoming
Nomination Action)
216

Table 4.12. Communication Structural Change (Emergency Response Incoming
Nomination Action)
240

Table 4.13. Communication Structural Endurance (Recovery Incoming
Nomination Action)
274

Table 5.1.1. Efficient Communication Paths Distribution (Pre-earthquake) 288

Table 5.1.2. Efficient Communication Paths Distribution (Emergency Response) 290

xiv

Table 5.1.3. Efficient Communication Paths Distribution (Long-term Recovery) 292

Table 5.1.4. Combined Geodesic Distance Results (Communication Network) 293

Table 5.1.5. Reciprocity Measures (Communication Network) 316

Table 5.1.6. Triadic Relationships in Communication Networks
(Transitivity Measures)
322

Table 5.1.7. Clustering Coefficient (Communication Network)-Motivational
Network
331

Table 5.1.8. Communication Network Pre-Earthquake Within-Group and Cross-
Group Density Measures based on Registration Actor Attribute
342

Table 5.1.9. Communication Network Pre-Earthquake Whole Network Results of
Group Internal and Group External Ties Based on Registration
Actor Attribute
343

Table 5.1.10. Communication Network Pre-Earthquake Group level E-I Index
Based on Registration Actor Attribute
344

Table 5.1.11. Communication Network Pre-Earthquake Ranking of Variability
across Actors with Group Trait Based on Registration Actor
Attribute
345

Table 5.1.12. Communication Network Emergency Response Within-Group and
Cross-Group Density Measures based on Registration Actor
Attribute
348

Table 5.1.13. Communication Network Long-term Recovery Within-Group and
Cross-Group Density Measures based on Registration Actor
Attribute
348

Table 5.1.14. Communication Network Emergency Response Whole Network
Results of Group Internal and Group External Ties Based on
Registration Actor Attribute
350

Table 5.1.15. Communication Network Long-term Recovery Whole Network
Results of Group Internal and Group External Ties Based on
Registration Actor Attribute
350

Table 5.1.16. Communication Network Emergency Response Group level E-I
Index Based on Registration Actor Attribute
354

xv

Table 5.1.17. Communication Network Long-term Recovery Group level E-I
Index Based on Registration Actor Attribute
354

Table 5.1.18. Communication Network Pre-Earthquake Ranking of Variability
across Individual Actors with Group Trait Based on Registration
Actor Attribute
355

Table 5.1.19. Communication Network Emergency Response Ranking of
Variability across Individual Actors with Group Trait Based on
Registration Actor Attribute
356

Table 5.1.20. Communication Network Long-term Recovery Ranking of
Variability across Individual Actors with Group Trait Based on
Registration Actor Attribute
356

Table 5.1.21. Development Stages of Weak and Strong Components in
Communication Networks
382

Table 5.2.1. Comparison of Reciprocity Measures before and after the Wenchuan
Earthquake (Communication and Collaboration)
444

Table 5.2.2. Triadic Relationships in Communication and Collaboration
Networks (Transitivity Measures)
450

Table 5.2.3. Cross-Network Comparison of Clustering Coefficient 456

Table 5.2.4. Communication and Collaboration Network Pre-Earthquake Within-
Group and Cross-Group Density Measures (Registration Actor
Attribute)
472

Table 5.2.5. Communication and Collaboration Network Emergency Response
Within-Group and Cross-Group Density Measures (Registration
Actor Attribute)
473

Table 5.2.6. Communication and Collaboration Network Recovery Within-
Group and Cross-Group Density Measures (Registration Actor
Attribute)
474

Table 5.2.7. Collaboration Network Pre-earthquake Whole Network Results of
Group Internal and Group External Ties Based on Registration
Actor Attribute
475

Table 5.2.8. Collaboration Network Emergency Response Whole Network
Results of Group Internal and Group External Ties Based on
Registration Actor Attribute
475

xvi

Table 5.2.9. Collaboration Network Recovery Whole Network Results of Group
Internal and Group External Ties Based on Registration Actor
Attribute
475

Table 5.2.10. Collaboration Network Re-scaled E-I Index 477

Table 5.2.11. Collaboration Network Pre-Earthquake Group level E-I Index
Based on Registration Actor Attribute
479

Table 5.2.12. Collaboration Network Emergency Response Period Group level
E-I Index Based on Registration Actor Attribute
480

Table 5.2.13. Collaboration Network Recovery Period Group level E-I Index
Based on Registration Actor Attribute
481

Table 5.2.14. Collaboration Network Ranking of Variability across Individual
Actors (Pre-earthquake)
483

Table 5.2.15. Collaboration Network Ranking of Variability across Individual
Actors (Emergency Response)
484

Table 5.2.16. Collaboration Network Ranking of Variability across Individual
Actors (Recovery)
484

Table 5.2.17. Development Stages of Weak and Strong Components in
Collaboration Networks
499

Table 6.1.1. Pre-earthquake Stage QAP Correlations/P-values (with Registration
Attribute)
563

Table 6.1.2. Pre-earthquake Stage QAP correlations/P-values (with Geographic
Location Attribute)
563

Table 6.2.1. Emergency Response Stage QAP correlations/P-values (with
Registration Attribute)
565

Table 6.2.2. Emergency Response QAP correlations/P-values (with Geographic
Location Attribute)
565

Table 6.3.1. Recovery Stage QAP correlations/P-values (with Registration
Attribute)

568
Table 6.3.2. Recovery Stage QAP correlations/P-values (with Geographic
Location Attribute)
569

Table 6.4. Communication Network-Core Class Membership 572

xvii

Table 6.5. Collaboration Network-Core Class Membership 572

Table 6.6. Core-periphery Density Table (Communication Network) 574

Table 6.7. Core-periphery Density Table (Collaboration Network) 574

Table 6.8. Three-period Network Measures Comparison (Communication and
Collaboration Networks)
576

Table 6.9. QAP Correlations between Communication and Collaboration
Networks (70 Actors)
579

Table 6.10.1. Communication Network Density Indicators 588

Table 6.10.2. Communication Network Tie Changes between Subsequent
Observational Periods
588

Table 6.11.1. Collaboration Network Density Indicators 589

Table 6.11.2. Collaboration Network Tie Changes between Subsequent
Observational Periods
589

Table 6.12. Rules Governing Emergency Response and Recovery Periods
Communication Network Evolution
595

Table 6.13. Rules Governing Emergency Response and Recovery Periods
Collaboration Network Evolution
605

Table 6.14. General Rules Governing Communication and Collaboration
Network Evolution
621

Table 6.15. Co-evolution of Communication and Collaboration during
Emergency Response Period (t1-t2)
644

Table 6.16. Co-evolution of Communication and Collaboration during Recovery
Period (t2-t3)
647

Table 6.17. Co-evolution of Communication and Collaboration (t1-t3) 649

Table 7.1. Communication Network Pre-Earthquake Inter-sector relationships 689

Table 7.2. Communication Network Emergency Response Inter-sector
relationships
691

Table 7.3. Communication Network Recovery Inter-sector relationships 691

xviii

Table 7.4. Collaboration Network Pre-earthquake Inter-sector relationships 694

Table 7.5. Collaboration Network Emergency Response Inter-sector
relationship

694
Table 7.6. Collaboration Network Recovery Inter-sector relationships 694

Table 8.1. Density Measures of Communication and Collaboration Networks
(t1-t3)
738

Table 8.2. Communication Network Pre-earthquake Period Degree Centrality
Measures
754

Table 8.3. Communication Network Emergency Response Period Degree
Centrality Measures
757

Table 8.4. Communication Network Recovery Period Degree Centrality
Measures
760

Table 8.5. Actor-level Betweenness Descriptive Measures for Communication
and Collaboration Networks (t1-t3)
781

Table 8.6. Communication and Collaboration Whole Network Betweenness
Centralization Measures (t1-t3)
784

Table 8.7. Communication and Collaboration Network Local Centrality
(Out-Degree Centralization): Emerging Resilient Structuration
788

Table 8.8. Communication and Collaboration Network Local Centrality
(In-Degree Centralization): Empowerment
788

Table 8.9. Communication and Collaboration Network Global Centrality
(Betweenness Centralization): Strengthening and Sustaining
788

Table 8.10. Conceptual Framework of Institutional Change in Civil Society
Domain
831

Table 8.11. Pre-earthquake Communication Network Density Matrix (CONCOR
Analysis)
838

Table 8.12. Pre-earthquake Communication Network Block Image of CONCOR
Analysis
839

Table 8.13. Emergency Response Communication Network Density Matrix
(CONCOR Analysis)
845

xix

Table 8.14. Emergency Response Communication Network Block Image of
CONCOR Analysis
846

Table 8.15. Recovery Period Communication Network Density Matrix
(CONCOR Analysis)
853

Table 8.16. Recovery Period Communication Network Block Image of
CONCOR Analysis
854

Table 8.17. Pre-earthquake Collaboration Network Density Matrix
(CONCOR Analysis)
863

Table 8.18. Pre-earthquake Collaboration Network Block Image of CONCOR
Analysis
864

Table 8.19. Emergency Response Collaboration Network Density Matrix
(CONCOR Analysis)
867

Table 8.20. Emergency Response Collaboration Network Block Image of
CONCOR Analysis
868

Table 8.21. Recovery Period Collaboration Network Density Matrix
(CONCOR Analysis)
873

Table 8.22. Recovery Period Collaboration Network Block Image of CONCOR
Analysis
873

Table 9.1. Network Theory Guideline 881

Table 9.2.1. Degree Measurements Illustrating the Development of Civil Society
Structural Resilience
898

Table 9.2.2. Conceptual Summary the Construction of Civil Society Structural
Resilience
899

Table 9.3. Communication and Collaboration Network Clustering Activities
Before and After the Earthquake
901

Table 9.4. Haberma’s (1989) Schema of Social Realms 908

xx

List of Figures
Figure 1.1. Near-collapsed Buildings in Dujiangyan 3

Figure 1.2. Destruction of a Residential Building in Dujiangyan 4

Figure 1.3. Damage of the Earthquake to a Local Hotel in Dujiangyan 5

Figure 1.4. Dujiangyan Resident Sitting Outside of a Completely Destructed
Residential Site Located in the City Center Area
6

Figure 1.5. Earthquake Damage to a Local Historical Site in Dujiangyan 7

Figure 1.6. Post-earthquake Reconstructed Housing in Dujiangyan-01 10

Figure 1.7. Post-earthquake Reconstructed Housing in Dujiangyan-02 11

Figure 2.1. Theoretical Mapping of Literature Review 17

Figure 3.1. Levels of Institutional Analysis 51

Figure 3.2. Research Design 53

Figure 3.3. Mixed Methods Design 55

Figure 3.4. Concurrent Data Collection and Sequential Data Analysis Procedures 88

Figure 4.1. Recovery Activities of Civil Society Actors 113

Figure 4.2. Perceived Resilient-building Factors 116

Figure 4.3A. Communication Network (Pre-Earthquake) 124

Figure 4.3B. Communication Network-Pre-Earthquake (Circular Layout) 125

Figure 4.4A. Communication Network (Emergency Response) 126

Figure 4.4B. Communication Network-Emergency Response (Circular Layout) 127

Figure 4.5A. Communication Network - Recovery 128

Figure 4.5B. Communication Network-Recovery (Circular Layout) 129

Figure 4.6 A. Collaboration Network (Pre-Earthquake) 135

Figure 4.6B. Collaboration Network-Pre-Earthquake (Circular Layout) 136

xxi

Figure 4.7A. Collaboration Network (Emergency Response) 137

Figure 4.7B. Collaboration Network-Emergency Response (Circular Layout) 138

Figure 4.8A. Collaboration Network - Recovery 139

Figure 4.8B. Collaboration Network-Recovery (Circular Layout) 140

Figure 4.8.1A. Banner Placed Outside Actor #51 Chengdu Field Office Showing
Civil Society Groups and Organizations in its “Incubation”
Program
152

Figure 4.8.1B. Banner Placed Outside Actor #51 Chengdu Field Office Showing
Civil Society Groups and Organizations Completed its
“Incubation” Program in Shanghai, Beijing, and Chengdu
153

Figure 4.8.2A. The birds-eye view of the newly reconstructed post-earthquake
housing of Yangping village, the primary location where one of
the rural recovery programs managed by Actor #61 was
implemented
157

Figure 4.8.2B. Another birds-eye view of the Yangping village 158

Figure 5.1.1 Illustration of Communication Channels 286

Figure 5.1.2. Illustration of Efficient Communication Channels 298

Figure 5.1.3 Triad Representations as in Transitive Triplet and Three-cycle
Relationships
320

Figure 5.1.4. Communication Network with Selected Actor Traits (Pre-
earthquake)
333

Figure 5.1.5. Communication Network Pre-earthquake Weak Component
Structure
381

Figure 5.1.6. Communication Network Emergency Response Weak Component
Structure
383

Figure 5.1.7. Communication Long-term Recovery Weak Component Structure 383

Figure 5.1.8. Communication Network Pre-earthquake K-core Structure 389

Figure 5.1.9. Communication Network Emergency Response K-core Structure 391

xxii

Figure 5.1.10. Communication Network Long-term Recovery K-core Structure

392

Figure 5.1.11. Pre-earthquake Girvan and Newman Community Structure-
Communication Network (Highest Q Modularity Value=0.136; 4
Communities Found)
396

Figure 5.1.12. Emergency Response Girvan and Newman Community Structure-
Communication Network (Highest Q Modularity Value=-0.000; 5
Communities Found)
398

Figure 5.1.13. Long-term Recovery Girvan and Newman Community Structure-
Communication Network (Highest Q Modularity Value=-0.000; 2
Communities Found)
399

Figure 5.1.14. Pre-earthquake

Community Structure Detection-Communication
Network (Detection Method by Blondel, et al., 2008)
403

Figure 5.1.15. Emergency Response Community Structure Detection-
Communication Network (Detection Method by Blondel, et al.,
2008)
404

Figure 5.1.16. Recovery Community Structure Detection-Communication
Network (Detection Method by Blondel, et al., 2008)
405

Figure 5.1.17. Combined Attributes of Communication Network (Pre-
earthquake)
412

Figure 5.1.18. Pre-earthquake Communication Network (Out-degree and
Communities)
414

Figure 5.1.19. Emergency Response Communication Network (Combined Actor
Attributes of Registration, Community, and Geographic Location)
421

Figure 5.1.20. Emergency Response Communication Network (Out-degree and
Communities)
423

Figure 5.1.21. Long-term Recovery Communication Network (Out-degree and
Communities)
429

Figure 5.1.22. Long-term Recovery Communication Network (Combined Actor
Attributes of Registration, Community, and Geographic Location)
430

Figure 5.1.23. Types of Communication Platforms (Long-term Recovery) 440

xxiii

Figure 5.2.1. Pre-earthquake Collaboration Network (Combined Actor Attributes
of Date of Establishment, Geographic Location, and Registration
Status)
460

Figure 5.2.2. Comparison of Pre-earthquake Communication and Collaboration
Networks (Combined Actor Attributes of Date of Establishment,
Geographic Location, and Registration Status)
462

Figure 5.2.3. Comparison of Emergency Response Communication and
Collaboration Networks (Combined Actor Attributes of Date of
Establishment, Geographic Location, and Registration Status)
464

Figure 5.2.4. Comparison of Recovery Communication and Collaboration
Networks (Combined Actor Attributes of Date of Establishment,
Geographic Location, and Registration Status)
467

Figure 5.2.5. Collaboration Network Pre-earthquake Weak Component Structure 500

Figure 5.2.6. Collaboration Network Emergency Response Weak Component
Structure
502

Figure 5.2.7. Collaboration Network Recovery Weak Component Structure 503

Figure 5.2.8. Collaboration Network Pre-earthquake K-core Structure 506

Figure 5.2.9. Collaboration Network Emergency Response K-core Structure 508

Figure 5.2.10. Collaboration Network Recovery K-core Structure 516

Figure 5.2.11. Pre-earthquake Collaboration Network Community Structure
(Partition=8, Highest Q Modularity Value=0.391)
523

Figure 5.2.12. Emergency Response Collaboration Network Community
Structure (Partition=10, Highest Q Modularity Value=0.247)
525

Figure 5.2.13. Long-term Recovery Collaboration Network Community Structure
(Partition=9, Highest Q Modularity Value=0.130)
528

Figure 5.2.14. Collaboration Network Pre-earthquake Community Structure
(Blondel et al., 2008)
532

Figure 5.2.15. Collaboration Network Emergency Response Community
Structure (Blondel et al., 2008)
533

Figure 5.2.16. Collaboration Network Recovery Community Structure
(Blondel et al., 2008)
534

xxiv

Figure 5.2.17. Pre-earthquake Cross-sector Structure (Collaboration Network
Dynamics)
542

Figure 5.2.18. Emergency Response Cross-sector Structural Transformation
(Collaboration Network Dynamics)
545

Figure 5.2.19. Recovery Cross-sector Structural Transformation (Collaboration
Network Dynamics)
548

Figure 6.1. Communication Network Evolution (70 actors) 577

Figure 6.2. Civil Society Domain Institutional Formality Structures (Post-
earthquake China)
618

Figure 7.1. Pre-earthquake Communication Network (Actor Attributes in Date of
Establishment, Geographic Location, and Registration Status)
685

Figure 7.2. Emergency Response Communication Network (Actor Attributes in
Date of Establishment, Geographic Location, and Registration Status)
687

Figure 7.3. Long-term Recovery Communication Network (Actor Attributes in
Date of Establishment, Geographic Location, and Registration Status)
688

Figure 7.4. Process of Civil Society Role Identification (Case of #51) 729

Figure 8.1. Process of Inclusion in Civil Society Action Structure before the
Earthquake
741

Figure 8.2. Process of Inclusion in Civil Society Action Structure during
Emergency Response Stage
742

Figure 8.3. Process of Inclusion in Civil Society Action Structure during
Recovery Stage
744

Figure 8.4. Pre-earthquake Communication Network Local Centrality
(Out-degree)
762

Figure 8.5. Pre-earthquake Communication Network Global Centrality
(Betweenness)
763

Figure 8.6. Emergency Response Communication Network Global Centrality
(Betweenness)
769

Figure 8.7. Recovery Period Communication Network Global Centrality
(Betweenness)
772

xxv

Figure 8.8. Pre-earthquake Period Collaboration Network Global Centrality
(Betweenness)
775

Figure 8.9. Emergency Response Period Collaboration Network Global
Centrality (Betweenness)
778

Figure 8.10. Recovery Period Collaboration Network Global Centrality
(Betweenness)
781

Figure 8.10.1. Newly Constructed Urban Communities Post-earthquake in
Dujiangyan-1
810

Figure 8.10.2. Newly Constructed Urban Communities Post-earthquake in
Dujiangyan-2
811

Figure 8.10.3. Newly Constructed Urban Communities Post-earthquake in
Dujiangyan-3
812

Figure 8.10.4. Newly Constructed Community Hospital Serving the Residents
Living in Reconstructed Housing in the Surrounding Areas
813

Figure 8.10.5. A Community Center Established by Actor #49 Providing Food
and Refreshments to Community Residents Nearby
819

Figure 8.10.6. Community Center Established by Actor #49 Providing Residents
a Place to Gather and Relax with Services such as Food,
Entertainment, and Newspapers
820

Figure 8.10.7. Books provided by Actor #49 as Part of the Community Center
Service
821

Figure 8.11. Pre-Earthquake Communication Role Structure (Block Diagram of
CONCOR)
841

Figure 8.12. Emergency Response Communication Role Structure (Block
Diagram of CONCOR)
848

Figure 8.13. Recovery Period Communication Role Structure (Block Diagram of
CONCOR)
855

Figure 8.14. Pre-Earthquake Collaboration Role Structure (Block Diagram of
CONCOR)
865

Figure 8.15. Emergency Response Collaboration Role Structure (Block Diagram
of CONCOR)
869

xxvi

Figure 8.16. Recovery Period Collaboration Role Structure (Block Diagram of
CONCOR)
875

Figure 9.1. Research Framework (Data Analysis Sequence) 882

Figure 9.2. Meaning and Role of Civil Society in Driving the Processes of
Institutional Change (Civil Society in Wenchuan Earthquake
Recovery-Social Groups and NGOs)
884

Figure 9.3. Communication Network Structure Evolution 894

Figure 9.4. Collaboration Network Structure Evolution 895

Figure 9.5. Strengthening Mechanism of Civil Society Institutions 900

Figure 9.6. Post-Wenchuan Earthquake Chinese Civil Society and the State 906

Figure 9.7. Resilience Formation: Chinese Civil Society in Times of Crisis 910

Figure 9.8. A Conceptual Framework for Civil Society Action in Times of Crisis 915

xxvii

Abstract

The importance and the theoretical significance of the civil society construct in the
public sphere and its involvement in the policy decision-making process have long been
emphasized by scholars in policy and planning. The theory itself has yet to deal with the
role of a particular set of actors in civil society through a process of social change. My
research approaches this piece of the social justice issue by defining a set of foundational
problems called the “theoretical paradoxes of action”: 1) If the institutions of planning exist
to reduce uncertainty in our lives and thus provide social order, how do they deal with
unexpected change? 2) If by definition, institutions exist to provide stability and meaning
to social life, to what extent can they contribute to the ability of society to learn, adapt, and
reorganize to meet urban challenges? This dissertation tackles this problem set from the
perspective of civil society actors through a procedural-action-oriented approach, while
taking into consideration of the diversity of planning cultures across countries.
I investigated the role of civil society in developing long-term collaborative efforts
for urban settlements to cope with risks and uncertainties associated with catastrophic
disasters. Using emerging citizen groups and non-governmental organizations (NGOs) as
the main unit of analysis, the primary intention of the study is to examine their role in

xxviii

forming communication and collaboration governance networks during the post-
earthquake response and recovery period. It seeks to explore the experiences of social
groups and organizations’ participation during the initial three-year recovery process after
the 2008 May12 Wenchuan earthquake in Sichuan province, China. I adopted a case study
methodology with a mixed-methods research design. The quantitative method utilizing
social network analysis to look at the emergence and evolution of institutions inside the
domain of civil society, their relational arrangements among each other, as well as to the
state and the market domains at different points of time. The supplemental qualitative study
of key participants representing group/organizational actors focused on in-depth
understanding of the meaning and the driving forces of institutional change inside the civil
society domain. Two types of contextual environments—communication and collaboration
network structures—were being investigated along with a longitudinal study of their
evolution over three time periods: before the 2008 Wenchuan earthquake, immediate
emergency response and short-term recovery period, and the longer-term (up to three years)
recovery stage after the earthquake.
Using UCINET network analysis software program, I focused on understanding the
macro and micro structural characteristics of how social groups and organizations built
their communication and collaboration networks over three distinct periods of time. Each

xxix

of the six structural environments (three communication networks and three collaboration
networks) was investigated separately to look at how actors were connected and embedded
within their local and global network environments. Accompanied by qualitative data
collected from in-depth interviews and ethnographic field observations, the analysis
demonstrated primary evidences in understanding the formation, persistence, and the
sustenance of the action structures for both communication and collaboration network
environments before and after the Wenchuan earthquake.
I further developed longitudinal network models in discovering the rules that
governed the dynamic network behavior over the specified three periods of time. I utilized
the SIENA program implemented in the R statistical system to longitudinally investigate:
1) Whether the institutional status in terms of actor registration had an effect on
communication and collaboration behavior; 2) Whether there were structural tendencies
that would affect the specific formation patterns of the communication and collaboration
network development; 3) Whether the types of recovery activities that actors engaged in
had an effect on the structural dynamics of the two types of networks; 4) Whether there
were tendencies for the cross-mediation between communication and collaboration
structures that facilitated the creation and maintenance of the two types of networks. With
the supporting evidences from the qualitative data, the findings demonstrated the formation

xxx

of a type of proactive coping style through which newly emerged group and organizational
actors took the primary role in overcoming their differences in institutional status and in
re-constructing a social structural environment that nurtured the long term social capacity
in dealing with extreme distress or uncertainty. Throughout the different stages of network
development, the group behavior self-generated a kind of change dynamics that prompted
its own evolution, thus showing signs of endurance and transformation.
At last, I proposed for a preliminary conceptual framework to understand civil society
actions in times of crisis. It depicted the development of a resilient social structure as a
growth process and elaborated on the role of civil society in building resilient social
structures as a coping and adapting mechanism when facing extreme uncertainties and
changes. The dissertation contributed to the understanding of the meaning and role of civil
society in driving the processes of institutional change as a learning process for social
capacity-building. My study also added to Amartya Sen’s (1992) conceptual development
of agency freedom by adopting a relational approach to understand the capability formation
of social actors, particularly identifying the different types of risk-coping and adaptation
behaviors and their links to defining a set of social-political conditions that prompted the
self-emergence and evolution of network structures. This perspective emphasizes on the

xxxi

empowerment issue of civil society actors, particularly when institutions are generally
perceived as constraints on human behavior.

1

Chapter 1
Introduction

The Planning Context of the Wenchuan Earthquake Recovery

Research Background and Questions
The event under study is the recovery of Sichuan Earthquake, which struck 92km
northwest of the Sichuan provincial capital, Chengdu, on May 12, 2008, with a
magnitude of M8.0 on the Richter scale (UNDP). It was the most devastating natural
disaster ever to occur since the founding of the People’s Republic of China, both in
seismic scale and in terms of impact on the Chinese people. The earthquake affected
more than 100,000 square miles and about 30 million people; 69, 226 deaths were
attributed to the disaster (EERI 2008). Dujiangyan, a city with national historic and
economic importance, is located not too far from the Sichuan capital Chengdu (see Map
1.1 to Map 1.9 in Appendix I.). The city has an area of 1,208 square kilometers and had a
population of 60,000 in 2003 (xzqh Forum, 2008). The city was one of the 39 hardest-hit
disaster affected areas in Sichuan Province and was subject to the Post Earthquake

2
Restoration and Reconstruction Planning implemented by the central government
(General Recovery Plan 2008).
My interest in the 2008 Wenchuan Earthquake recovery and the set of related
research questions did not come out of a blank slate. I visited the heart-breaking sites of
destructions in the city of Dujiangyan two months after the May 12 earthquake. I was
struck by the level of physical damage that the earthquake had done to the city (see figs.
1.1, 1.2, 1.3, 1.4, and 1.5).

3

Figure 1.1. Near-collapsed Buildings in Dujiangyan
(Source: Photo taken by Jia Lu, 2008)

4

Figure 1.2. Destruction of a Residential Building in Dujiangyan
(Source: Photo taken by Jia Lu, 2008)

5

Figure 1.3. Damage of the Earthquake to a Local Hotel in Dujiangyan
(Source: Photo taken by Jia Lu, 2008)

6

Figure 1.4. Dujiangyan Resident Sitting Outside of a Completely Destructed
Residential Site Located in the City Center Area
(Source: Photo taken by Jia Lu, 2008)

7

Figure 1.5. Earthquake Damage to a Local Historical Site in Dujiangyan
(Source: Photo taken by Jia Lu, 2008)
With a group of planning scholars from the United States who were invited to a
recovery planning debriefing hosted by the city mayor of Chengdu, I first encountered the
Chinese government’s comprehensive plan for the entire Chengdu region and particularly
the intention of incorporating citizen participation as part of the reconstruction process.
The importance of “self-help initiation and enthusiasm” on the part of Chinese people as
a “persistent motivation” (General Recovery Plan 2008, 5) to rebuild homes together was
also emphasized consistently throughout the government mandated General Recovery
Plan.

8
From further reviewing the state’s disaster recovery plan, I became intrigued by
how several of the following aspects that appeared in the original Plan were implemented
on the ground and how the active role of the citizens would be revealed over time, if any.
First of all, among the many basic principles that were being mandated in the Plan, was a
clear welcoming gesture on the participatory “mechanism innovations” among “the state,
private enterprises, social organizations, and individual citizens” (General Recovery Plan
2008, 7). This means that the state recognized and welcomed such cross-sector initiatives
within the disaster recovery context. But to what extent could these efforts be brought
forth through an emerging process and possibly being enabled to institutionalize would
require in-depth investigation. Secondly, the “three-year recovery completion goal” (8)
was clearly targeted from the state’s perception in terms of what constitutes sustainable
economic and social development. If this was indeed the time-frame that was sufficient
for such long term development incorporating social components, how has the process
been experienced or lived through by those who took an active part in disaster recovery?
In other words, an investigation looking at the role of citizens and how the possible
collaboration mechanisms unfolded throughout this time frame will provide a
documentation of the catastrophic disaster recovery experiences of the Chinese case from
the society’s point of view.

9
Looking at the Plan through a social recovery point of view would build up the
inventory for future disaster mitigation and preparedness practices not only for China but
also internationally. Although the General Plan did stress the importance of capacity-
building for the purpose of constructing a more comprehensive disaster mitigation system,
the foundational principle was emphasized on “central coordination 统筹兼顾” (General
Recovery Plan 2008, 6), thus being silent on the role of social domain as an active part
of constructing recovery. If increasing capacity for conducting emergency response and
recovery are indeed part of the goal for disaster mitigation and preparedness as mandated
in the Plan, the question then becomes how should public policy encourage or enable
“social participation” (46) throughout the various kinds of social recovery activities, such
as providing “educational support, disabled and elderly, psychological counseling, etc.”
(46).
As I was pondering on these different aspects of the recovery Plan, what
immediately came to my mind were these questions: How would the Chinese citizens
participate in the process? How would the citizen initiatives be enabled? Is there anything
about the Chinese society that would make this process unique? Would the earthquake
change the way planning is conducted in the longer term? Two years later, in the summer
of 2010, I went back to Dujiangyan and was surprised to see the ways how people could

10
enjoy their lives and the physical progresses they made in housing reconstruction (see
figs. 1.6 and 1.7).
Figure 1.6. Post-earthquake Reconstructed Housing in Dujiangyan-01
(Source: Photo taken by Jia Lu, 2010)

11

Figure 1.7. Post-earthquake Reconstructed Housing in Dujiangyan-02
(Source: Photo taken by Jia Lu, 2010)
What was still left in my mind is the question how people themselves would feel
and describe their experiences since the earthquake in 2008. What would the collective
memories say about the Chinese society and its possible transformations? For these
purposes, I find that a mixed methods inquiry combining both quantitative and qualitative
research will be the most appropriate in providing answers to the questions asked.

12
Purpose of the Study
The purpose of this study is to first examine the role of civil society in post-
earthquake recovery, mitigation, and preparedness efforts for urban settlements. Then I
investigated whether these efforts were collaborative and sustainable in the long term.
Specifically, I seek to explore the recovery experiences of Chinese citizens after the
May12 Wenchuan earthquake in the year 2008.
The impetus for the study originated not just from the devastating impact of the
earthquake in China’s Sichuan Province in 2008 (EERI 2008) but also from the signs of
an emerging civil society both during the immediate response (Wang 2009; Zhang 2009;
Teets 2009) and in the recovery period after the earthquake. In the numerous studies that
have focused on in-depth investigations of disaster response in different cultural contexts,
one of the recurring themes is the relationship between civil society and the state. Aseem
Inam’s (2005) comparative study of recovery planning in Mexico City and Los Angeles
argued that institutions within the state and civil society are both important in building
‘resilience’ for megacities. Emel Ganapati (2005) examined the construction of social
capital in the context of a ‘weak’ state and its role of saving lives from the rubble after a
1999 earthquake in Turkey. In the U.S. context, Judith Steele (2006) investigated the case
of Oakland fire in California in 1991. Divya Chandrasekhar (2010) examined the

13
participation by different ‘stakeholder groups’ within the state and civil society by
looking at the recovery of South India from the 2004 Indian Ocean Tsunami. There were
also voluminous studies that touch upon the social aspect of the recent recovery of
Hurricane Katrina in New Orleans (Arnold 2006; Boettke et al. 2007; Chamlee-Wright
2007; Chamlee-Wright and Storr 2009). However, the role of civil society and its
relationship with the state and market forces are less understood within the Chinese
context, particularly in the aftermath of catastrophic disasters. The examination is
expected to make a contribution to the understanding of planning cultures, which is the
key in creating a global conversation about the role of planning in social change (Sanyal
2005). The term “planning culture” is defined as “the collective ethos and dominant
attitude of professional planners in different nations toward the appropriate roles of the
state, market forces, and civil society in urban, regional, and national development”
(Sanyal 2005, 3). When planning cultures are in constant flux with social change (Sanyal
2005), examining the role of civil society and its relationships with the state, and market
forces constitutes a crucial step in understanding how changes occur in planning practice.
A “prostrate civil society” (Scott 1998, 5), within which “unmarked citizens” have no
ability to contribute their own values, original ideas, and personalities to the planning
enterprise, has been regarded as a “breeding ground” for authoritarian states to execute

14
high-modernist plans with no resistance from the ordinary people. Scott (1998) argues
that disastrous events, such as war, revolution, and economic collapse can either weaken
civil society or make societies receptive to new orders of arrangements. But how a
“prostrate civil society” can be transformed into a “standing-up” civil society with
citizens realizing their skills, intelligence, and experiences to actually act upon their
opinions and desires is less well understood.
Virtually no research has addressed the ‘planning culture’ in relation to China’s
disaster recovery response. If this is a dynamic concept that evolves with social, political,
and economic influences (Sanyal 2005), then two essential questions emerge: 1) Can
catastrophic disasters “jump start” the emergence of a civil society? 2) Can the
institutional changes being generated in the domain of civil society have longer term
effects on the general planning culture? This study will attempt to answer these questions.

Significance of the Study
The potential findings of the proposed study are expected to be significant to the
field of urban planning in a number of ways. First, for the theory of planning,
understanding the concept of civil society in China provides a setting in which planning
culture can be examined and which may ultimately contribute to the creation of a “global

15
conversation about the role of planning in social change” (Sanyal 2005, 24). In particular,
the term “civil society” was non-existent throughout Chinese history until the beginning
of the country’s modernization process towards the end of the Qing Dynasty (Wu and
Gong 2008; Xiao 1993). How this concept has been understood and constructed socially
in China is expected to contribute to the foundation that builds up the inquiries into the
theory of planning.
Secondly, the actual practices of civil society’s involvement in decision-making
brought about further normative debates among theorists on how planning processes
should be conducted. One approach argues for a participatory decision-making process in
which planners, politicians, developers, and the public forge working agreements
together, thus keeping power relations in balance (Forester 2007; Healey 1997; Innes
1995). Another approach argues for the direct opposition of civil society to the state in
public decision making (Friedmann 1987) while realizing the limits of either in isolation
in providing social justice (Fainstein 2000). But in a situation where citizens initiated the
emergence of civil society alongside with the functioning of the state, a close
examination revealing such a process and the meaning of civil society in this particular
cultural context renders in-depth focus of attention.

16
Chapter 2
Theoretical Reflection

The study is based on a critical review of the following three bodies of literatures:
disaster, civil society theory, and institutional theory. The existing gaps in the planning
and disaster literatures point to further investigation in the institutional theory literature.
The debate related to agency freedom and institutional change demand close examination
of the civil society concept. I first discuss the existing debates and gaps in the literature.
Then, I will present a graph to clarify the theoretical mapping that guides the
development of this research (see Figure 2.1).

17

Figure 2.1. Theoretical Mapping of Literature Review

Disaster Literature
Conceptualization of Disaster and Resilience
Three main concepts in the disaster literature were essential in exploring the role of
Chinese civil society in the 2008 Wenchuan earthquake recovery in Sichuan province:
“disasters”, “sustainability”, and “resilience”. None of these, however, thus far has a
definitional consensus. First of all, the term “disaster” itself is a contested concept (Oliver-
Smith 1998). Scholars have tried to reach an agreement over the years about its sociological
meaning (Fritz 1961; Quarantelli and Dynes 1977; Dubin 1978; Kreps 1985; Barton 1989 in
Disaster
Institutions
Civil Society
Civil
Society
State Market
Disaster
-Response
-Recovery
-Mitigation
Domain of Relevant Literature
The Institutional Domain
of
Civil Society Theory
Policy
Application
Planning

18
Kreps 1998; Kreps 1998; Oliver-Smith 1998; Quarantelli 2001). Some examined “how social
systems react to physical harm and social disruptions after an event has occurred” while
others examined “what social systems do to increase or mitigate the risk of physical harm and
social disruption before an event has occurred” (Kreps 1998, 33). An emerging paradigm in
looking at the key dimensions of disasters is “disaster as uncertainty” has come into being in
recent years (Gilbert 1998). This concept of disaster is closely related to the notion of
community as the location of social action (Dynes 1998, 113). Therefore, disaster can be
linked to uncertainty in three ways (Gilbert 1998): 1) when a danger threatens a community
and neither the cause nor the effect of the danger can be clearly defined or identified, 2) the
growing complexity of modern communities have rendered uncertainty to be mainly a
product of community organization but not of external factors, 3) when there is a loss of
capacity to define situations “through traditional understandings and symbolic parameters”,
disasters will mean a loss of “key standpoints in common sense” for a community and
“difficulties of understanding reality through ordinary mental frameworks” (Gilbert 1998, 17).
Paralleling such “uncertainty” and “system of meaning” approach, other researchers have
focused on returning to a general sociological theory concerning the maintenance of social
order in the face of uncertainty—“including uncertainties created by the destructive effects of
disasters” (Stallings 1998, 132) as well as Oliver-Smith’s “political ecology” approach to

19
disaster studies by focusing on the “relationships between people, the environment, and the
sociopolitical structures that characterize the society of which the people are members
(Campbell 1996 in Oliver-Smith 1998, 189). In a word, the impact of disasters on people as
members of the society, their relations with one another, with the society and the state have
been some of the sustained interests in disaster research. However, how individuals or
communities can take up the responsibilities to cope with the risks and uncertainties
associated with catastrophic disasters requires the introduction of two other crucial concepts:
“sustainability” and “resilience”.
While “sustainability” and “resilience” are two emerging concepts that intrigued
diverse discussions among scholars in disaster research (Berke 1995; Mileti 1999; Berke
2002; Mileti and Gailus 2005; Aguirre et al. 2005; Berke et al. 2008) as well as those in the
field of planning or resilience (Pickett et al. 2004; Vale and Campanella 2005; Campanella
2006; Gleeson 2008; Wallace and Wallace 2008), few have actually explicitly defined what
is meant by “sustainability” and “resilience” as well as their relationships in disaster contexts.
The definition of “sustainability” has owed its origins in the study of “new ecology” within
social-ecological systems (Scoones 1999; Berkes et al. 2003), where it is defined as “the
capacity to create, test, and maintain adaptive capacity” (Holling et al. 2002, 76). In this
approach, sustainability is considered a process rather than an end product and as “a dynamic

20
process that requires adaptive capacity for societies to deal with change” (Berkes et al. 2003,
2). Another aspect of looking at this process is to reverse our current conceptualization in
relation to stability and change: “rather than assuming stability and explaining change, as
often done, one needs to assume change and explain stability (van der Leeuw 2000 in Berkes
et al. 2003, 2).
“Resilience”, on the other hand, is the “buffer capacity or the ability of a system to
absorb perturbations” (Holling et al. 1995 in Berkes and Folk 1998, 6) such as natural
disturbance (earthquakes) and human activities (resource use and pollution). It is a tool to
analyze adaptive change towards sustainability because it provides a way to look at “how to
maintain stability in the face of change” (Berkes et al. 2003, 15). Building resilience, is thus a
process to enhance the adaptive capacity for social systems to deal with change or surprises
towards sustainability, which is “inherently unpredictable” and “cannot be planned in a
rational fashion” (Berkes 2003, 15).

21
Disaster Recovery and the Role of Civil Society
Following these conceptual clarifications, I now explore the part of disaster
literature that particularly dealt with the role of civil society in developing the social
adaptive capacity when facing extreme perturbations, such as earthquakes.
The “associational life” that Young (1999) and Habermas (1996) have
emphasized in their discussions regarding the unit of composition of civil society has
been one of the main focuses in disaster research. It is relevant in resilience-building in
urban settlements after catastrophic natural disasters. Wallace and Wallace (2008) have
studied urban neighborhoods as consisting of social networks and the interlocking of
layers of social networks in community efficacy in building resilience. Campanella (2006)
emphasized the role of citizenry as the source of city’s resilience after Hurricane Katrina.
According to his account of post-Katrina New Orleans, the grassroots mobilization led to
“lasting political reforms” and “commitment to building affordable housing” (143), but
the lack of coordinated grassroots activism also made it difficult for a city to have the
capacity to rebound.
Earlier disaster scholars were also aware of the interactions among the public
sector, the private sector, and “emergent citizen groups” (Stallings and Quarantelli 1985)
at different “phases” of a disaster cycle, such as preparedness, emergency response,

22
recovery, and mitigation (Neal 1997). Until more recently, the relationships among the
interactions of civil society, the state, and the market, particularly the proactive role of
civil society actors during disaster recovery are being argued to be critical for areas of
risk reduction and development (Arnold 2006). In his examination of World Bank
disaster projects implemented in a cross-country context, Arnold (2006) concluded that
the empowerment and capacity-building at the local community level are keys to
“effective risk management” when taking a developmental approach to disaster recovery”
(279). The right incentives need to be found not only in the U.S. domestic but also in the
international context. After investigating the long term recovery and redevelopment
efforts of eight catastrophic disaster events from the attacks of September 11, 2001, to
Hurricane Katrina in 2005 and the Haiti Earthquake in 2010, Garnett and Moore (2010)
found that both the bottom-up approach involving local people and the top-down
approach incorporating a long term development vision from the state are all needed for
disaster recovery. In particular, “local empowerment, organization and leadership, and
planning for sustainability” (1) are integral aspects of recovery planning and risk
management. While these principles of “participation”, “empowerment”, and
“collaboration” are important in disaster recovery at the local community level for
countries recovering from the 2004 Indian Ocean Tsunami (Rowland and Tan 2008), the

23
social-political conditions under which the actions of civil society occurred indeed varies
from other countries’ experiences. For the case of Turkey, disasters such as the Golcuk
earthquake in 1999, the politics between the society and the state was being re-negotiated
(Pelling and Dill 2010) with a “shattered” role of the state (Ganapati 2005). The role of
civil society, in the form of the “emergence of civic networks” (Ganapati 2005, 289),
arose out of the context of a weak state. An active role of civil society in promoting
“disaster mitigation and prevention strategies” was also being noted in the recovery
process from Hurricane Mitch in El Salvador in 1998 (Wisner 2001, 252). A “dogmatic”
neoliberal state accompanied by a lack of capacity for local municipal governments to re-
build after the disaster was found to serve as the social-political backdrop in the El
Savadorian case. Some examples of the municipal government’s lack of capacity
included problems in “planning, programming, budgeting, management, and litigation,
etc.” (262). The investigation for cases in Japan’s disaster recoveries showed another side
of the social-political landscape. Archival research reviewing Japanese disaster recovery
processes showed greater role of the state in building-up the physical infrastructures as
compared to the cases investigated by Wisner (2001) and Ganapati (2005). While less
attention was being paid to social infrastructure reconstruction on the government side,
there was an active role of civil society in bringing residents back into the disaster-

24
damaged cities and thus keeping to accelerate the disaster recovery process (Aldriech
2008).
However, little has been said about how institutions in the Chinese context
respond to natural disasters over the long term development planning. The relationship
between the state and civil society under China’s market reform has been widely
contested (Hui 2004; Harvey 2005; Nonini 2008; Wu 2008) and the debate has
intensified when examining the country’s urbanization process (Lin and Ho 2005; Chan
2007; Lin 2007). How catastrophic disaster recovery relates to the institutional
complexities of urban development in China remains unclear. Quarantelli (1999) has
suggested a linkage between disaster planning and the long term regional and national
developmental planning, because “every decision about residential land use, plant siting,
and indeed every industrial and economic policy or program, carries with it some
consequences for risk and hazard” (15). Bates and Peacock (1987) further defined the
term “development” in the disaster context as “a process by which a population improves
its level of adaptation to an environment and through such improvements raises the level
at which it satisfies human needs and wants, and at the same time lowers its levels of
vulnerability to disruptions” (Bates and Peacock 1987 in Quarantelli 1999, 15). However,
the question of how developmental goals and disaster recovery can be carried out by

25
different institutional actors has not been substantially discussed through empirical
studies that contribute to the recovery and development processes in the developing
countries’ context, particularly China.

Civil Society Theory
The Conceptualization of Civil Society
The review of the disaster literature calls for further attention to two other lines of
theories to help understand the case of Wenchuan earthquake recovery. One is the “civil
society theory”, and the other is the “institutional theory”.
The term “civil society”, in the Western history, has been greatly contested.
Distinct arguments have been put forward by theorists who made their observations
“within a particular society or to explore a particular dimension at one moment in time”
(Hall 2005, 3). From a concept central to the public sphere (Habermas 1989) in late
seventeenth and eighteenth century Europe, to the current debate of its nature with the
rise of the market economies, the concept of “civil society” has lacked a consensual
definition (Edwards and Gaventa 2001; Edwards 2009). The term has always been in flux
with social realities. On the one hand, one could argue that the concept essentially implies
a contested nature at certain historical moments (Hall 2005). On the other hand, others

26
could argue for a “global civil society” with its “self-directing or non-governmental
institutions and ways of life” stretching across borders (Keane 2003 in Hall 2005, 287).
Whether one sides with one argument or the other, it is how civil society as an aspect of
the public sphere interacts with the state and the economy that remains the crux of the
theoretical contention.
This sphere within which civil society interacts with the state in a market
economy is often referred to as the public sphere. The relationship between the public
sphere and civil society has been subject to debates. Young (1999) argues that public
sphere corresponds to the activities of political and some civic associations. These
associations further belong to those within civil society whose voluntary associational life
is usefully distinguished from the state and the economy. The latter two domains are
systemic in the sense that people are “conditioned by system imperatives of bureaucratic
routines or profit making” (144); 2). Rawls (2001), however, regard the ways of
reasoning of associations is “nonpublic” (92) with respect to political society and so to
citizens generally. For Habermas (1995), a public sphere needs to be characterized by
having cumulative experience, a political character, a provision of justice to reality and
the need to put forward a public use of reason, as well as a relatively homogeneous public
composed of private citizens engaged in rational-critical debate.

27
Among the different schools of thoughts on the concept of civil society, I chose to
understand civil society in such a way that “shows its development through the stages by
which we have attained it” (Cohen and Arato 1992, 605) and thus include the possibilities
of further theoretical adjustments to the term itself. From this historical perspective, civil
society is “a sphere of social interaction between economy and state, composed above all
of the intimate sphere (especially the family), the sphere of associations (especially
voluntary associations), social movements, and forms of public communication” (ix). The
term also refers to “the structures of socialization, association, and organized forms of
communication of the life world to the extent that these are institutionalized or are in the
process of being institutionalized” (x). Cohen and Arato (1992) also argued for a
‘mediating’ rather than antagonistic relationship among the spheres of civil society, the
state, and the economy. If this “three-part” model (Cohen and Arato 1992) is at the center
of the theoretical debate about civil society, then the institutional process necessary for
the emergence of such a model deserves further examination.
As the evolving conceptual history of civil society has struggled with the nature
of the concept and left many questions unanswered, the recent theoretical endeavors have
ventured beyond the European continent and America, particularly into the vast
developing worlds. Much has been debated about the existence of Chinese civil society

28
among Western scholars (Strand 1990; Metzger 2001 in Kaviraj and Khilnani 2001;
Howell and Pearce 2001, Heberer and Sausmikat 2005) and Chinese scholars (Li 2004;
Zhang 2005; Tao 2009; Yuan 2009; Zhang 2009). The center of the contention revolves
around the relationship between the state and civil society in an emerging market
economy with the concept of civil society defined in a variety of ways. Since the Sichuan
Earthquake in May 2008, the massive number of self-motivated volunteer efforts by
Chinese citizens and the establishment of related disaster-relief non-governmental
organizations (NGOs) have led some scholars in China (Chang and Fu 2009; Xiao et al.
2009) to believe we are witnessing the “birth” of Chinese civil society after the
catastrophic disaster. Two to three months after the earthquake, Teets (2009) tried to
verify these claims by studying the participation of “civil society groups” participation in
short-term relief efforts. However, in looking at the emergence of civil society in the
Chinese context, nothing has been written to examine the process of interaction among
the institutions of the state, civil society, and market economy over the long-term
recovery period. I expect that my study will contribute to the understanding of this
process.

29
Three Domains of Planning Theory: Civil Society, the State, and the
Market
The planning literature, especially theories in planning, has undergone dramatic
paradigm shifts. It started from the “top-down” rational planning approach with planners
using their scientific and technical knowledge to manage spatial change. “The rational
man”, particularly among economists, relies heavily on classical economic theory to
justify the role of state in managing the economy “in the public interest”. Incrementalism
(Dahl and Lindblom 1953) and advocacy planning (Davidoff 1962) responded to the
shortcomings of rationality-based models by acknowledging the various group interests
in the state’s decision-making process. This is followed by the approach of
“communicative action” (Innes 1995; Healey 1999) in which planners become facilitators
in the negotiation process among politicians, developers, and the public. The “bottom-up”
paradigm, however, emphasizes the direct collective action “from below” and the
mobilization against ruling elites situated in the domains of the state and market
(Friedmann 1987). Some recent literature suggests that the state, market forces, and civil
society are all important in achieving social justice (Fainstein 2000). Little has been done
to investigate the actual meaning of civil society and the institutional processes of
interactions among the domains of the state, civil society, and market forces.

30
For theories of planning, there are three domains throughout the history of
modernity and arguably, within the conditions of “post-modernity” (Berman 1998;
Giddens 1998): the state, the market, and the civil society. Each of these components has
been argued to have its own linkages to the planning tradition. For its connection to the
market, proponents of the neo-classical economics have argued for the superiority of
market mechanisms such as privatization, deregulation, and price-based approaches to
planning interventions. The role of planners, according to this tradition, is supposed to be
strengthening or even creating markets. Consequently, the contribution of the civil
society to planning has largely been ignored. The connection between the state and
planning can be traced to traditions such as “social reform” and “policy analysis”. The
primary emphasis in this model is to focus on how planning can serve the state and thus
making actions by the state more effective. The appropriate role or engagement of the
civil society in planning has been regarded with minimal tolerance since the general
public is often assumed not possessing the adequate scientific knowledge to inform the
decision making process. The direct connection between the civil society and planning
can be related to tradition of social movement as well as what was later referred to as
“radical planning” (Friedmann 1987, 389). Civil society, when “organized for a life in
common” (344), will formulate part of the “political community” which directly

31
challenges the existing power structure domination. Planning, within such tradition, must
be grounded in assisting the civil society to stand against the repressive state and planners
must be able to stand up against the hegemonic power and “put their work in the service
of emancipatory values and a strong political community” (315).
Although the three lines of inquiry are important in the history of planning
thoughts, one critical under-developed area in relation to these traditions in planning
theory is the extent of the institutional context and structure within which the planning
processes are being or can be transformed through the interactions among actors of the
state, the market, and the civil society. On the one hand, the dialectic forces of our human
experience of modernity have re-bonded the interactions among the market, the state, and
the civil society. On the other hand, with the on-going process of globalization, where
uncertainties, contingencies, and fluidity prevail with dissolution of distinctions and
differences, the three components are put in a relationship of permeating spheres where
they become increasingly fused with each other. The field of planning, then, resides in a
public sphere of the modern society. According to Habermas (1995), this sphere needs a
political character of communication based on critical debates, a provision of justice to
reality, and putting forward a public use of reason. Such public sphere within which
planning operates can thus be expanded to an overall context of social justice. If this is so,

32
a subsequent set of questions that are left unanswered within the theoretical terrain are:
What is the role of planning in constructing such a public sphere? What are the
institutions and the institutional arrangements that could sustain the role of civil society
in planning and its interactions with actors in the state and the market domains?
Apart from the aspects of substantive planning theory, the procedural aspect of
planning is another integral part of reaching an outcome of social justice. Tracking back
to the theoretical assessment of justice in Sen’s (1992, 1999) capability approach and
John Rawls’ (2001) “Justice as Fairness”, there has been a consistent gap between what
earlier scholars have envisioned for and the institutional processes to achieve them. For
Sen (1999), both the capability approach and the need of plurality of institutions for
promoting people’s overall freedom are inextricably linked. For Rawls, the argument
focused on the importance of the basic institutional structure of the society and its role in
serving as the “primary subject of political and social justice” (10). However, what has
been left unanswered in this line of inquiry is the operational side of the picture: How to
achieve a “plurality of institutions” so that people’s opportunities for functioning and
freedom can be maximized? And how to build a basic institutional structure to provide
opportunities for citizens to achieve the kind of life they value, which include their “aims,
aspirations, and character” (Rawls, 2001, 10)? Such are the themes dealing with the

33
procedural aspect of planning institutions. An important concept to consider with respect
to processes is “human agency” (Giddens 1984) and its connections to
“institutionalization” (Jepperson 1991 in Powell and Dimaggio 1991) and “institutional
change” (Powell and Dimaggio 1991; Scott 2008), given the concept of institutions and
the importance of it properly understood. However, if the role of institutions is played out
by reducing uncertainty, making our expectations more reliable (Verma 2007), and
providing stability and meaning to social life (Scott 2008), the question to consider then,
is how planning can be founded in Sen’s political conception of justice with the provision
of human capabilities representing the freedom enjoyed by each person “to choose the
lives that they have reason to value” (1992 p81)? The possible convergence of the
institutional dimension and the capability approach of planning remains to be
investigated and such endeavor may lead to an important contribution to an institutional
perspective of planning theory.

Institutional Theory
The Conceptualization of Institutions
The institutional dimension of civil society theory thus offers other threads to look
further into institutional theory for some necessary conceptual clarifications.

34
There are three widely used approaches to understanding the concept of
institutions. One approach underscores the institutional constraints and the regularization
of behavior (Scott 2008), and prominence is given to explicit regulatory processes—
“rule-setting, monitoring, and sanctioning activities”. Social order is believed to be based
on rules. New institutional economics scholars such as transaction cost theorists (Coase
1937, 1960; Williamson 1988), economic historians (North 1990), as well as political
scientists in the rational choice/game-theoretic tradition (Ostrom 1990, 2005) have all
emphasized this regulative pillar of institutions. Another approach gives prominence to
“binding expectations” as the basis of social order. Organizational theorists such as
March and Olsen (1989) embraced such normative roles of institutions in which the basis
of compliance is “social obligation” compared to “expedience” in the regulative pillar.
Both the regulative and the normative approaches contend that institutions can impose
constraints on social behavior as well as empower and enable social actors (Scott 2008).
Little has been said about the extent to which individuals and groups can be enabled by
regulatory and normative pillars of institutions, have the capabilities to choose to exercise
their “agency freedom” (Sen 1992), thus taking action to perform responsibilities, duties,
and accepting mandates.

35
The third approach argues for a cognitive and cultural explanation of institutions
(DiMaggio and Powell, 1991). The cultural-cognitive explanation emphasizes the
“internal” interpretive processes that are shaped by “external” cultural frameworks over
an extensive period of time. Institutions are created as a social process through repetitive
actions (Berger and Luckmann 1967) and as an entrenchment of an “intellectual process”
through people’s minds (Douglass 1986). However, what is left to be understood is how
institutional change can occur if the cognitive roots of institutions are formulated through
long period of sedimentation and historical tradition.

Institutional Structuration and Transformation
If institutions can be understood as “a conventional to standardized interaction
sequences” that has “attained certain state or property” (Jepperson 1991, 145), then,
institutionalization represents a “process of such attainment” (145). In the institutional
theory literature, the paradox of action and structure (Giddens 1984) has commonly been
understood as a state within which institutions “simultaneously empowers and control”
(Jepperson 1991, 146) as representing a “constraint and freedom duality” (Fararo and
Skcoretz 1986 in Jepperson 1991, 146). Then, a key question is how to articulate the
institutionalization process amidst such an action/structure co-production dynamic.

36
Current discussion in institutional theory has yet to develop an explicit set of metrics that
can represent the various “degrees of institutionalization” (150) in different topical
domains of the society. One example would be social resiliency in response to crisis and
catastrophic disasters, particularly the structural institutionalization among civil society
actors in various cultural contexts. This inevitably brings about the concept of
institutional change, which can exhibit four stages of characteristics: institutional
formation, institutional development, deinstitutionalization, and reinstitutionalization
(Jepperson 1991). In the context of this study, I focus on the first two processes: a)
institutional formation denoted as “reproductive patterns based upon ‘action’”; b)
institutional development as a process of maintaining and elaborating the formation
momentum (152).
Essentially, there are two levels of unit of analysis when considering institutional
change. One focuses on the generation of institutional forms such as groups and
organizations themselves. The other one focuses on the “institutional environment” (Scott
1991, 165) within which groups and organizations function and conduct their activities.
Along the latter line of research, Powell (1991) also argued that in order to explain the
different stages of change, the structural sources of various types of institutional

37
environments requires further investigation. He particularly pointed out the relevancy of
such processes being exemplified in the crisis context:
Actions taken to respond to challenges and crises often lead to the establishment
of new institutional powers and precedents. Yet at this point we know relatively
little about organizational fields change their structure and content (Powell 1991,
201).

Missing Links: Institutions and Social Change
In summary of the previous literature review, if the definition of institutions is to be:
Institutions are comprised of regulative, normative, and cultural-cognitive
elements that, together with associated activities and resources, provide stability
and meaning to social life (Scott 2008, 48).
One issue that cut across both institutional and planning theory is how planning
institutions deal with uncertainties and risks in general. The challenge is that it questions
the underlying assumption of the institutional analysis that institutions are important.
How important are they in dealing with social change in diverse cultural contexts? And
under what conditions are they to be associated with social justice?
Planning and policy-making are embedded within institutions, be it rules, norms,
or cultural practices, which have inevitably become part of our daily lives. When the goal
for planning is to make social change toward social justice ends, such as within the
growing concerns of sustainability and resilience-building, particularly at the level of

38
groups, organizations, and communities, the origin of institutions and their formation
process must become an integral part of the intellectual endeavor in order to look at the
source of the making that change.
Along with scholars in social justice who addressed the possible constructs of
civil society in the public sphere, the importance of involvement of civil society in the
decision-making process has been argued through different approaches in the planning
literature (Forester 2007; Healey 1997; Innes 1995; Friedmann 1987; Fainstein 2000). On
the one hand, these studies have generally taken a theoretical approach from which ideas
and arguments have been conceptualized and debated in a descriptive nature. On the
other hand, the theory itself has yet to deal with the role of a particular set of actors in
civil society through a process of social change, particularly in cultures where a
“prostrate civil society” (Scott 1998, 5), within which “unmarked citizens” (Scott 1998, 5)
have no ability to contribute their own values, original ideas, and personalities to the
planning enterprise.

A Capability Approach to Institutional Change in Civil Society Domain
My research approached this piece of social justice issue by defining a set of
problems of what I call the “theoretical paradoxes of action”. And they were exemplified

39
by institutional theory and the theory of civil society in the following ways. 1) If by
definition, institutions exist to provide stability and meaning to social life, to what extent
can they contribute to the ability of society to learn, adapt, and reorganize to meet urban
challenges? If the institutions of planning exist to reduce uncertainty in our lives and thus
provide social order, how do they deal with unexpected change? 2) If civil society is
indeed important in the planning and policy decision-making process, to what is its role
in the process of social change, in different cultural contexts?
I attempted to approach this problem set through a procedural-action-oriented
approach emphasizing on the perspective of civil society actors. The theoretical aspect of
such a stance originated from Sen’s (1992) definition of “agency freedom” and his
“capability” approach to social justice. The concept of “agency freedom” was
distinguished from the “well-being freedom” as the former reveals a person’s “freedom to
bring about the achievements one values and which one attempts to produce, while the
latter is one’s freedom to achieve those things that are constitutive of one’s well-being”
(57). The primary focus of agency freedom relies on the person’s own initiative in taking
agency actions in bringing about the achievements. And such freedom is further reflected
by the person’s capability in bringing forth the different kinds of lives that he or she has a
reason to value. Following Sen’s arguments on the importance of agency freedom, the

40
purpose of my study is to discover the source and extent of such freedom. And it is
conducted through examining the origin and development of the capability set that actors
inside the civil society domain can be enabled with their own agency actions in times of
change and crisis.
The uniqueness of this approach is that it allows for explicit observations of the
interactions of civil society actors over time. It also offers new insights from
understanding their experiences and developing possible scenarios and dynamic models
that specify the rules governing processes of the source of social change. The findings in
this respect will be making a contribution to understand the process that agency freedom
can be developed when institutions are generally understood as constraints on human
behavior.

Social Network Theory
In order to implement the “procedural-action-oriented approach” emphasizing on
the civil society group/organizational actors in this research, I chose to utilize the social
network theory as a lens through which I investigate the role of civil society in the
Chinese disaster recovery context. This approach can be further breaking down into both
theoretical and methodological perspectives. In the following paragraphs, I will briefly

41
introduce the background literatures in social networks illustrating the uniqueness of such
a research approach in this study context, and leave the empirical application part of the
discussion to the later methodological section.
According to Borgatti (2009), theories of social networks can be developed
through different levels of perspectives and each one has its significance in linking the
theories discussed in the previous sections. First of all, the most basic level of approach is
looking at the “types of ties” (893), which in this research context would be the
theoretical backdrop of my structural analysis. Table 2.1 illustrates some of the examples
that are relevant to the context of this research study based on Borgatti’s original
categorizations. As we can see, the study of the Wenchuan earthquake recovery can be
further broken down into sub-areas. The investigation of patterns in institutional
formation and development, or measures of institutionalization (Zucker 1991), inside the
civil society domain can thus be specified with different combinations of characteristics
in each category. For example, one can look at the communication interactions among
those non-registered informal civil society actors who emerged in Sichuan both during
the response period and the recovery period. This way, indicators of the degree of
institutionalization for the focused structural environment can be generated.

42
Table 2.1. Use of Social Network Theory in Studying the Wenchuan Earthquake
Recovery
Similarities Social
Relations
Flows Interaction
Contexts
Interactions
Location Attribute Activities -Communication
-Collaboration
-
Information
-Resources
-Emotions
-Beliefs
-Before
Disaster
-Response
-Recovery
-Mitigation
-Preparedness
(Within
Sector)
-Civil Society

(Cross
Sector)
-Civil Society
-State
-Market
Geographical
proximity
-Registration
Status,
-
Establishment
Date
-Women,
-Children,
-Environment
-etc.

Source: Borgatti 2009; Adapted by Lu 2013
The second level of network theory development can be discovered at the
structure level. By “structure”, this research specifically refers to the institutional
environment that civil society actors enacted to shape and at the same time may be
constrained it. The graphical shapes of the different types of structures, either at the
whole network level or at the sub-structure level, reveal various levels of cohesiveness or
disintegrated-ness of the institutional structures. Properties of dyadic, triadic, and sub-
group properties illustrated through graph theory provide the key for examining how
actors chose to situate themselves across different time stages of the disaster recovery
process. When the focus is on each actor inside the structure, centrality measures can be
compared and the different characteristics of a set of actors can be identified based on the
context of disaster recovery. For example, in the communication network for each time

43
stage, a set of actors can be identified as more “powerful” because many others reached
out to them to seek earthquake related information or others must pass through them to
reach to the rest of the actors.
The third level in developing network theory regards finding out the factors that
can explain “the formation of network ties and, more generally, to predict a host of
network properties, such as the clusteredness of networks or the distribution of node
centrality…” (Borgatti 2009, 894). Besides this focus on formation, some recent
methodological development (Snijders et al. 2010, 2012) in network analysis also
allowed theories to look into the rules governing the institutional change process thus
exemplifying their prediction power in forward-looking models. And these models can be
looked at from two perspectives. One set of research questions can be developed by
seeing network structures as dependent variable. In the context of this study for example,
one can ask: what is the basis of communication or collaboration ties over time? What are
the endogenous structural and the exogenous attribute factors can help explain the
development of these structures? In other words, the premise is that actions and behaviors
shape the structure. The other side of the picture is looking at network structures as
independent variables and examining if they can predict certain outcomes related to
homogeneity and performance (Borgatti 2009). For example, one can ask whether the

44
development of the communication or the collaboration network predicts the joint
participation of response and recovery related activities such as building up livelihood for
local communities, environmental protection, taking care of disadvantaged populations,
and providing psychological counseling services, etc.
The divergent “mechanisms” (Borgatti 2009, 894) that can explain networks as
outcomes or the consequences of network variables may then be identified as another
level of network theory-building. One specific concern for network theory is related to
how the findings can used to inform the actors with the hope of “influencing the way
people see themselves and how they act” (895). In this research, the degree to which the
structural and the dynamic theoretical formulations that can be utilized by policy makers
and the civil society actors themselves to enhance the capacity of the coordinated disaster
response, recovery, mitigation, and preparedness remains to be seen. This would require
cross-cultural examinations to develop network theories in particular relevancy to disaster
recovery planning and risk management.

45
Chapter 3
Methodology

Philosophical Assumptions
One’s worldviews plays an important role in the formulation of research questions,
researcher’s role in the study, as well as the actual practice of research (Creswell 2007;
Padgett 2008). Therefore, I feel it is crucial for me to explain my philosophical
assumptions that influence the way I conduct my research. Social constructivism is one of
these assumptions. In this worldview, individuals seek to understand the world in which
they live and work. The meanings toward certain objects or things are varied and multiple,
leading the researcher to look for complexity of views rather than narrow the meanings
down into a few categories and ideas (Creswell 2007). I see the ways that meanings being
developed are negotiated socially and historically. Therefore, I allow myself to ask open-
ended questions and to listen carefully about what people say about their lives and work.
Interactions with others play a decisive role in constructing meaning of a situation.
Participatory is another one of my worldviews. Growing up in China, I have seen
some of the poorest rural areas where people still lived without electricity and have not
seen any kind of modern automobile. As in the developed countries today, even with

46
China’s more recent experiences of modernization that has brought luxurious ways of
living for some, the problems related to the quality and standard of living of the vast
majority remains. As a researcher, I not only want to study the issues at hand but also
would like my research to benefit the participants in a way that may change their lives.
Especially for the marginalized or simply those whose point of views have often been
ignored by the dominant structures, I hope to provide a voice for the participants, raising
their consciousness and improving their lives as the issues are studied and exposed.

Research Statement
The overarching central research question of this study is: What is the role of civil
society in developing long-term collaborative efforts at the level of local citizens to cope
with risks and uncertainties associated with catastrophic disasters?
I structure the examination of this question through three sub-components so that
the major concerns and complexities in the central question can be addressed and
resolved.

47
Component 1(Drivers of Agency Action and Institutional Change)
Can catastrophic disasters bring about actions for institutional change in Chinese civil
society? If so, how do actions and interactions inside the civil society domain emerge in
the Chinese disaster recovery context? What are the motivational factors driving the
emergence and development processes? How do the sources of the processes contribute
to the understanding of meaning of Chinese civil society?
By the term “institutional change”, I refer to two types of processes. One is the “internal
generation of institutionalized forms” (Zucker 1991, 165) within groups and
organizations. In other words, it is the study of actions that triggered the formation and
development of the informal social groups as well as formal NGOs that established field
offices in Sichuan Province during the disaster response and the recovery periods. The
other type of institutional change I examined was the emergence and evolution of the
institutional environments represented by the structure of interactions among actors
within civil society and across the civil society, the state, and the market sectors.

48
Component 2 (Communication and Collaboration: Structural Dynamics of Civil Society
in Action)
Do institutions within the domains of civil society, the state, and market system
“collaborate” in shaping recovery responses after catastrophic disasters? What are the
structural dynamics within the civil society domain and across civil society, the state, and
the market?
The structural changes within the domain of civil society were examined through the
development of social groups and NGOs that either emerged or established field offices
after the Sichuan earthquake. I then looked at the interactions among the institutions of
civil society, the state and the market system.
Rather than using the term “market”, I choose “market system” in my research
inquiry by defining it as “a system of society-wide coordination of human activities not
by central command but by mutual interactions in the form of transactions” (Lindblom
2001, 4). The use of terms such as “market” and “economy” does not differentiate
themselves from the activities of the state and civil society because they all make use of
and basically focus on the interchanges of certain goods and services. The “market
system”, however, emphasizes the activity of “social coordination by mutual adjustment”
(Lindblom 2001, 23) and encompasses society as a whole rather than an area of behavior

49
often referred by “market” or “economy”. This way, the market system serves as an
alternative schema, besides civil society and the state, to think and understand a society
(Lindblom 2001). Different types of coordination may be adopted in each domain and
may vary from one country to another. The scope of all three domains in this study—civil
society, state, and market system—is consistently society-wide.
Another critical component here is the idea of “collaboration”. While
collaborative public management scholars have argued for greater role of public and
citizens in participatory governance (Bingham et al. 2008), the conceptualization of the
term “collaboration” depicting the “antecedents, processes, and outcomes” (8) of actions
within and across sectors has yet been made clear and requires further research endeavor.

Component 3: (Resilience: Rules Governing Emergence and Evolution)
Are there rules governing the structural dynamics within civil society domain and across
civil society, the state, and the market? Do their interactions shape the institutional
change in civil society? How?
As the within and cross-sector interactions emerge and develop over time, it became
critical to examine the network structural characteristics and the actor attributes that can
help explain the persistence of institutional environment outcomes. The process from

50
institutional formation to institutional development can be accounted by the sub-structure
formation inside communication and collaboration networks. These different types of
sub-structures that contributed to the process can be further captured by longitudinal
social network models. It is therefore possible to depict the social rules and regularities
guiding the structural changes both within the civil society and across sector boundaries.
The evolution of these networks will use three points of references: before the 2008
Wenchuan earthquake, short-term response period, long term recovery period. The
persistence of both the internal institutionalized social structures inside the civil society
domain was a key source of social resilience after the catastrophic earthquake in the
Chinese case.
In summary, figure 3.1 below illustrates a map in developing the different levels of
analysis in relation to each component inside the research statement.

51

Figure 3.1. Levels of Institutional Analysis

Methodology
Research Methods
Figure 3.2 below shows a general map of the lay-out of my research plan. The
proposed research adopts a case study methodology for an in-depth description and
analysis of the role of Chinese civil society in the 2008 Sichuan Earthquake recovery
process. The role of civil society is examined through resolving three sub-components in
relation to the central research question: 1) Structural dynamics of civil society in Action,
Internal generation of
institutionalized forms
Among Civil Society Actors Civil Society Actor
Institutional Environment
inside civil society domain
Civil Society, State, Market
Cross-sector institutional
environment

52
2) Rules governing civil society emergence and evolution, 3) Sources and motivations of
agency action and institutional change. Each of these components also contributes to one
another in answering the overall research question. My case study research will use both
quantitative and qualitative approaches in which the investigator explores a bounded
system (a case) over time, through detailed, in-depth data collection involving multiple
sources of information (e.g., observations, interviews, audiovisual material, documents
and reports) (Creswell 2007, 73). Barney Glaser and Anselm Strauss (1967) emphasize
the importance of a qualitative method in order to achieve the ends of concept
specification for theory-generating purposes, particularly when data from one nation
involve “structural conditions, consequences, deviances, norms, processes, patterns, and
systems” (18). I strive to understand the actual recovery experiences of people and how
their lives had changed since the catastrophic disaster. A qualitative component within
the case study approach was designed to understand these issues in which processes and
connections are imperative but are incompletely comprehended (Peattie 1983).

53

Figure 3.2. Research Design
Role of Civil Society
in Disaster
Response/Recovery/
Mitigation
Structural Dynamics of
Civil Society in Action

Rules Governing the
Emergence and Evolution
of Civil Society Institutions
Drivers of Agency Action
& Institutional Change

2008 Wenchuan Earthquake Recovery
A Case Study Methodology
Quantitative
Method
Quantitative
Method
Qualitative
Method
-Survey Questionnaires
-Archival documents
-Electronic traces (Online Resources)

-Interviews
-semi-structured
-open-ended
-Online conversations
-Documents
-Audiovisual
-Participant Observation

54
Rationale for Mixed Methods Research
The proposed study will be using a mixed methods research design (Creswell and
Clark 2007) with quantitative and qualitative data each answering a different set of
research questions. One reason for combining qualitative and quantitative data in this
study is that they together provide a more complete picture of the earthquake recovery
process rather than adopting either approach alone. The qualitative data focuses on
understanding the meaning of the civil society concept in the Chinese context and seeks
to identify the motivational origins that ignited actions and processes of institutional
change in civil society. The quantitative data provides a structural outlook of the actions
and interactions taken inside the domain of civil society and the factors contributing to
the structural evolutionary processes over time. The different research components that
are answered by the two types of data complement one another in making contributions
to answering the central research question.
Secondly, this research uses an embedded design (see figure 3.3) (Creswell and
Clark 2007) within which the qualitative data play a supportive role in adding to the
understanding of the mechanisms found in the quantitative data set. The quantitative
method is designed to look at the evolutionary stages of how actions inside civil society
took place. The qualitative study builds upon and further expands on the social network

55
analysis of the quantitative data by helping to understanding the underlying process that
builds the foundation for relations to emerge.

Figure 3.3. Mixed Methods Design
Source: (Creswell and Clark, 2007) adapted by Lu, 2013

Nature of the Data
Assumptions
By saying ‘quantitative data’ in my research, I am referring to the type of data that
can depict the structural relations between a pair of actors. On the one hand, patterns of
relations came into being as a result of actions taken by “agents”, which can take the
form of individuals, groups, organizations, communities, or nation-states, etc. These

Interpretation
based on QUAN
(QUAL) results
QUALITATIVE Method:
-Interviews
-Participant observation
-documents
QUANTITATIVE Method:
-Survey Questionnaire
-Interviews
Supplemental and
Complementary

56
agents can also be called “actors” as they take the initiative to reach out to others to form
a relationship with other agents or actors. On the other hand, the regularities of relations
or the “overall structures” in turn influence the perceptions, beliefs, decisions, actions,
and behaviors (Knoke and Young 2008) of the actors themselves. This kind of data as
contacts, connections, and ties is primarily referred to as relational data and the methods
that are being commonly used to examine it is called network analysis (Scott 2000). In
essence, relations are not qualities or traits of the agents themselves but rather, they are
properties of the “system of agents” (Scott 2000, 3). From this lens of looking at data, the
identity or role of one actor does not solely depend on a set of self-claimed characteristics,
but most importantly, on how it relates to others in a larger system of actors.
The standard quantitative data in most planning research is in the nature of
attribute data, which reveals the properties, qualities, and characteristics of individuals or
groups. The corresponding method in examining this type of data is variable analysis
(Scott 2001). The current state of art in the field has yet to utilize the relational data as
well as the appropriate analyzing methods to bring about their capacities in understanding
processes of social change. In this research study, I investigate a central phenomenon of
the actions and interactions among grassroots civil society groups/organizations after the

57
Wenchuan earthquake. There are three key assumptions in the analysis of relational data
(Knoke and Young 2008). I will briefly illustrate them in the following paragraphs.
First of all, “structural relations are often more important for understanding
observed behaviors than are such attributes as age, gender, values, and ideology” (Knoke
and Young 2008, 4). This assumption has to do with the question related to action and
how it can be revealed explicitly to illustrate a possible set of regularities and patterns.
The action perspective is the lens chosen here to examine the role of civil society groups
and organizations in the Wenchuan earthquake disaster recovery. Although actors defined
as group/organizations or understood as “collective social actors” (Knoke and Young
2008, 7) might vary in size, registration status, or working locations, or other attribute-
related traits, many become “leaders” in providing assistance towards other actors. Others
became key intermediaries in passing information along during the periods of emergency
response and disaster recovery. When the group/organizational actors’ attributes remain
constant over time, how actors relate to one another provide critical insights in explaining
the emergence and development of behavioral patterns. In other words, a structural-
relational lens brings about a different aspect of theoretical and empirical explanations on
how phenomenon in various social contexts would occur and the possible sources of
social actions.

58
The second assumption in focusing on the patterns of relational data and their
effects are “social networks affect perceptions, beliefs, and actions through a variety of
structural mechanisms that are socially constructed by relations among entities” (Knoke
and Young 2008, 5). In other words, the structural relations that are being actively
constructed will in turn not only affect the behavior of actors but also their perception of
the social reality within which they are embedded. The formation of informational and
collaborative structural relationships among civil society actors in this research for
example, may also enhance a sense of perceived belonging and thus sustaining cohesion
inside the civil society domain over time. The key emphasis here is on the transient
nature of social structures. When working with relational data, the social structure itself
became a dynamic one as it is being constantly constructed and reconstructed by the
observed behavior of social actors. In turn, the actors’ decision-making process, their
perceptions or beliefs are also being shaped by the way they are being situated inside the
structural relationships.
The third assumption is that “the structural relations should be viewed as dynamic
processes” (Knoke and Young 2008, 6). The premise here is that relational structures not
only can impose constraints on actor’s behavior but also can bring about actions through
socially constructed perceptions and cultural understandings. In the context of this

59
research, the role of civil society in emergency response and disaster recovery after the
earthquake event is in itself a dynamic process. Relation structures among civil society
actors not only emerge but also go through systemic transformations as a result of
“combined preferences and purposeful actions” (6) of actors. Examining such structural
changes longitudinally can, on the one hand, provide an alternative conceptual tool to
link micro-level actor behavior after the disaster event to macro-level structural
transformations in the civil society domain. On the other hand, it is also a methodological
tool in the planning field to design measures in evaluating the progresses of social
recovery after catastrophic disasters.

Actors, Relations, Networks
Two of the fundamental network concepts when investigating relational data are
actors and relations. They are the two key components when studying social structure, as
it “consists of regularities in the patterns of relations among concrete entities” (White et
al. 1976, 733-734 in Knoke and Young 2008). The “concrete entities”, or actors, can be
represented by individuals, informal groups, and formal organizations, or nation-states,
etc. In this research, the actors that I am focusing on reflect the associational nature of the
“civil society” concept, and they are voluntary informal social groups and formal Non-

60
governmental Organizations (NGOs) that participated in the emergency response and
recovery stages after the Wenchuan earthquake in 2008. I call them “civil society actors”
throughout the remainder of this study. Considering the different components of my
research question, I also included two other types of actors in this study. One is the state
actor representing the aggregate of all of the government branches. The other one is the
market actor representing the aggregate of all the private enterprises or businesses
functioning in the market system. This mixed design of network actors will ensure that
the cross-sector dynamics as part of the structural mechanism that build up the role of
civil society be explored and illustrated explicitly in a disaster recovery setting.
The concept of Relations, on the other hand, are the connections and ties built
between a pair of actors. Relations can also have directions. A tie can be directed when
one actor is the initiator of a relationship and the other is the recipient. A tie can also be
non-directed “when mutuality occurs” such as one in conversations (Knoke and Young
2008). In this research context, I focus on directed relations because during the times
after the disaster event, civil society actors from across the country came to Sichuan
Province to participate in the emergency response and recovery efforts. Many of them did
not know the existence of others beforehand and some of the actors only established
themselves as entities, either formal or informal, after the disaster event. In these

61
incidences, the overture by one group/organizational actor towards the other will be a
significant piece of information regarding the actor’s decision-making process and its
patterns of behavior.
A social network is defined as “a structure composed of a set of actors, some of
whose members are connected by a set of one or more relations” (Knoke and Young
2008, 8). Actors and relations are the two fundamental components of any network
structure. This research examines two types of network relations. One is how actors
communicate and exchange information with each other. The other is project
collaboration. Empirically, this study intends to represent, in a descriptive manner, the
micro and macro-structures of both communication and collaboration networks for
periods before and after the earthquake. Theoretically, I first aim to explore the sources of
emergence of these two types of networks by tracing the cognitive and motivational
origins of relations or ties. Then, I examine the rules governing the evolution of the two
types of networks as they develop over time, thus providing explanations of actor
behaviors transitioning from before the earthquake to the emergency response period, and
further into the long term recovery period.

62
Data Collection Procedures
Network Data Collection
Sampling Strategy and Description of Setting
Approaches to Network Boundary Specification
Several design elements in the network analysis require clarification in the
discussion about the collection of network data. The first element involves the decision
on sampling units (Knoke and Young 2008), which comprises setting of the network
boundary and the inclusion of actors.
At the theoretical level of social action, there are two types of boundary
specification strategies regarding the task of collecting relational data. One is the realist
strategy in setting network boundary “by definition assumes the proposition that a social
entity exists as a collectively shared subjective awareness of all, or at least most, of the
actors who are members” (Burt and Minor 1983). The second approach is called the
nominalist strategy, as it is a way of drawing the network boundaries based on a
nominalist perspective on social reality (Burt and Minor 1983). This strategy is further
defined as:
…an analyst self-consciously imposes a conceptual framework constructed to
serve his own analytic purposes. Delineation of network boundaries is analytically
relative to the purposes of the investigator, and thus network closure has no

63
ontologically independent status. There is no assumption that reality itself will
naturally conform to the analyst’s distinction; the perception of reality is assumed
to be mediated by the conceptual apparatus of the analyst, be he (or she) an active
participant in the social scene under study or an outside observer (21-22).

The setting of the current empirical research along with the specific site and field
access issues required a nominalist strategy for defining the network closure. The
following paragraph illustrates how I made the decision through my field encounters.
During my second field visit to Chengdu for a pilot study in the summer of 2010,
two key informants were being identified when I personally visited the joint office
established by two of the civil society groups participating in disaster recovery at the time.
Both groups quickly formed through voluntary initiatives shortly after the earthquake
event and maintained their functioning beyond the short-term emergency response phase.
I first came across the names of the two groups when browsing through their online
webpages and began to initiate phone conversations with one informant shortly after
arriving at Chengdu for my second field visit. The office spaces that belonged to the two
groups were actually joined together on the same floor of a residential building in the city
center of Chengdu. And this location proximity facilitated the process of identifying the
second key informant who also provided me with his permission to further investigate the
group’s formation process in my future formal fieldwork studies.

64
During my third field trip in early 2011, when my formal dissertation field study
actually took place, I first resumed my contacts with the two informants upon arriving in
the city of Chengdu. From the first few of my preliminary interviews, the nature and the
functioning of one of the civil society groups gradually became clearer to me. As this
group of interest called itself a “service center” focusing its work on assisting other civil
society actors engaging in disaster recovery-related activities, I found out that its
functioning was actually based on the existence of other actors that the center intended to
provide service for. Since shortly after the earthquake, the “center” maintained contacts
with a set of civil society actors who had been engaged in earthquake response and
recovery works in Sichuan. Although no formal agreement has ever existed among them,
this group of actors not only emerged out of a voluntary self-organized process but also
sustained their general contacts among each other over time, with the “center”
functioning and being treated as part of the group member at the same time. Discovering
this key piece of information during the early phase of my informant interviews
facilitated my thought process in identifying a possible set of appropriate actors in my
network approach, which was originally designed to look at the institutional change
inside the civil society domain after the Wenchuan earthquake.

65
Since this set of actors did not belong to a formally-titled group where members
have “widely agreed-upon labels such as General Motors or the University of Chicago”
(Burt and Minor 1983), the identification process not only involved a nominalist network
closure strategy but also contained certain degree of my own conceptual delineation in
drawing the network boundary for the purpose of this particular study. There is also a
possibility that the actors themselves might have shared a type of “we-feeling” by
perceiving each other as camarades and a subjective awareness of a sense of belonging.
However, relying on this strategy alone would generate more uncertainty in the data
collection stage due to the informality and the fluid nature of the group (Burt and Minor
1983).

Inclusion of Actors
Moving on to the details in the relational data collection procedure, I consider two
components in order to further clarify the networks to be studied. First of all, it is the task
to formally delineate the network actors. I combined two available approaches to decide
upon how to include the actors. One is the positional approach and the other is the
reputational approach (Burt and Minor 1983). To execute the former approach, I
conducted a membership test to refer to the “presence or absence of some attribute, most

66
commonly the occupancy of a position in a formally constituted group” (Burt and Minor
1983, 23). The initial set of civil society actors that voluntarily came together to
participate in the emergency response and recovery process was identified as part of a
service platform being provided by the center
1
, which in itself was also part of the group
set. Although the platform itself cannot be seen as a formally constituted group with
clearly identified position and membership titles, a list of their names was recorded as
those who initiated contacts with the “center” after the earthquake event. This list,
provided to me with the permission of the center’s informants, was used to serve as the
member test to initially include those civil society actors that became part of the center’s
platform after the earthquake.
I also adopted the reputational approach in order to further delimit the boundary
of actors. It is a strategy that is sometimes being used in combination with the positional
approach (Laumann and Pappi 1976) and “utilizes the judgments of knowledgeable
informants in delimiting participant actors” (Burt and Minor 1983, 23). The logic for this
research to use both approaches originates from the issues that arose from the field.
Through my field interaction with one of the key informants at the center, I realized that
not all of the civil society actors in the initial list remained to be active or continued to be

1
The center here refers to the group actor #3.

67
recognized as part of the platform. At that point, I decided to utilize the best judgments of
the informant to narrow down the original list to only those that she recognized as active
participants throughout the emergency response and the disaster recovery stage. The
informant can be named as “knowledgeable” in this regard because she became one of
the first volunteers working for the center since its establishment just three days after the
earthquake event. She also remained to be the center’s only long term staff from 2008 to
2011, by the time of my fieldwork in Sichuan.
The second actor inclusion rule was based on defining events or activities within
which the selected actors all participated and became involved in. Recall that the central
phenomenon being investigated in this study was the role of civil society in the
Wenchuan earthquake recovery. Therefore, the determining factor for inclusion here is
that the actors thus defined not only had to participate in the emergency response stage
but also had to sustain their actions into the long term recovery period. On the one hand,
since the center’s platform itself came into being as a formal entity only shortly after the
earthquake, the original list that it provided necessarily included all of those that
participated in activities after the earthquake, both short-term and long-term. However,
throughout the three years of recovery period until 2011, it is possible that some civil
society groups/organizations might have been disintegrated or did not survive beyond the

68
emergency response stage. It was then important to distinguish those actors that were able
to maintain functioning all the way into the disaster recovery period. The field knowledge
of the informant was again being utilized to draw an initial boundary of those actors
remained in the field for recovery purposes. But the delineation process continued
throughout the preliminary stage of survey distribution stage. Some actors, upon my first
field contacts through email or phone conversations, confirmed that they had not been
actively participating or engaged in field activities during long term after the disaster, and
in these instances, the network boundary clarification remained an ongoing process
throughout the data collection stage.

Relational Content Specification
The relational data revealed two kinds of experiences about the actors being
investigated. One is the relational form representing the intensity, frequency, strength,
and direction between pairs of actors (Knoke and Yang 2008). This aspect of relations
reveals the outward features of a structural system of actors. For example, the
cohesiveness or solidarity of a network structure under investigation can be thought of as
a form with a tendency to occur (Burt 1983) naturally in a particular social or cultural
context. The other actor experience can be explored is the relational content, also called

69
type of tie, representing the structure’s “substance as reason for occurring”. In this
research context, I explore two kinds of relational contents: communication and
collaboration. In other words, it is to examine multiplex relations when pairs of actors are
engaged in more than one type of tie (ex. communication and collaboration). The
theoretical motivations to collect these two types of relational contents at the same time
originated from my intention to understand the discourse of the civil society domain
within the context of disaster recovery in China. Empirically, I examined both the action
and the institutionalization processes among the civil society actors over their course of
participation in disaster recovery.
I used the communication content to capture the essence of the action structures.
This was necessary because during both the emergency response and the recovery periods,
information seeking and exchange among civil society actors became a critical source of
action for them to locate the appropriate types of activities for engagement, especially for
those actors coming from outside of the Sichuan Province. As Laumann and Knoke (1987)
noted that in the national policy domain, “the greater the variety of information and the
more diverse the sources that a consequential actor can tap, the better situated the actor is
to anticipate and respond to policy events that can affect its interests” (13).

70
I also used the collaboration content to capture the dynamics of
institutionalization (Jepperson 1991) of actions over time. The concept of “collaboration”
has been subject to debates in the area of public management, especially regarding the
preconditions for its emergence, processes, as well as outcomes (Bingham et al. 2008). In
this research, the concept is investigated as a social pattern where actors self-organize and
are persistently involved in an emergent or voluntary process carrying out different
aspects of a project or a program that require the long term devotion for them to practice
in the field. Overall, the multiplexity investigation is designed to understand the
emergence of the civil society construct after the disaster and its evolutionary processes
as a result of the institutional structural changes occurring over time.

Level of Analysis
There are four alternative levels to observe the relational data. At this point, it is
critical to clarify what each one of these levels of studies entails so as to provide a
background for the network data collection procedures.
The first level is the ego-centric network, which consists of “one actor (ego) and
all other actors (alters) with which ego has direct relations, as well as the direct relations
among those alters” (Knoke and Young 2008, 13). Each ego can be described by the

71
number, intensity, and the types of ties that it has with its set of alters. For the
examination of civil society in the disaster context, the above measures can be used to
look at the immediate “neighborhood” of those in direct contacts with the ego actor of
interest. The changes in these measures over time illustrate the ego’s behavioral change
both before and after the earthquake event. The second level of analysis lies in dyadic
networks, which consists of pairs of actors. The main purpose in examining this type of
network is to explain the existence of a type of relation between a pair of actors and to
see whether such relation is formed as the result of actor characteristic variations. For
example, if two civil society actors in the study share the same type of registration status
or other attributes together, the examination of the dyadic relation will provide
clarifications in depicting whether a tie would exist between the two actors of similar
traits, and the direction, intensity, as well as the duration of the tie. The third level of
analysis involves the triadic relations, which consists of a set of actors that involved in
triples. One of the basic phenomena that the investigation of triadic structures examines is
the process through which “friends of friends became friends”. In the context of this
study, as new civil society actors emerge during the periods after the earthquake event,
how they chose to close a triadic relationship with two other actors will reveal the
balance of the structure at the micro-level. The patterns of changes that are found over

72
time may be used to describe the specific features of structural evolution in the Chinese
disaster response and recovery context. There is a set of 16 distinct triad types when
taking into consideration of “all possible combinations of present and absent choice
relations among the actors in a triple” and a “basic descriptive question for empirical
network analysis is the distribution of the observed triads among the 16 types, a summary
tabulation called the triad census” (Knoke and Yang 2008, 14). I will elaborate more on
this topic in the data analysis section and illustrate its incorporation in the modeling of
network dynamics.
The fourth level of analysis deals with the complete networks. It is a macro-level
analysis as compared to the previous three micro-level relations. Information about every
relation among all the actors are being collected “to represent and explain an entire
network’s structural relations” (Knoke and Yang 2008, 14). This research study collects
data on the complete networks emerged in the civil society domain after the 2008
Wenchuan earthquake. The purposes for focusing on collecting the complete network
data are two-fold. One is to represent the structural emergence and evolution of both the
communication and collaboration networks formed by civil society actors. Positions and
roles are to be identified as the complete networks undergo changes throughout the
periods before and after the earthquake event. The second purpose of looking at complete

73
networks is to investigate the hypotheses tracing the causes and consequences of the
structural variations.

The Survey Questionnaire Method
In order to collect the thus specified relational data on complete networks, I
adopted a survey and questionnaire method as part of my data collection procedure.
Network studies in general draw extensively on the survey and questionnaire methods
(Carrington et al. 2005). The first section of the survey questionnaire was designed in to
collect: 1) the attribute data regarding actors’ date of establishment, registration status,
and the types of activities the actor engaged during the disaster recovery stage; 2) the
opinions of the respondents representing their group/organization on factors contributing
to disaster preparedness
2
. Although the study’s primary focus is on examining relational
data, actors’ attributes are incorporated as an integral part of the research, especially
when tracing the causes of the structural variations in network dynamics.
The second section of the survey and questionnaire instrument compiles a roster
list
3
of all the actors that were being considered after going through the previously stated

2
Please see Appendix 3.4.1 for original survey questions.
3
Please see Appendix 3.4.2 for a sample of the original roster list of the survey.

74
boundary specification process in the field. The provision of the entire list of names will
provide the opportunity to allow “respondents to recognize rather than recall their
relationships” (Carrington et al. 2005, 10). Along with the roster list of actor names, the
response formats utilized the “binary judgments” or “sociometric choices” (11) design
within which respondents specify whether the groups/organizations that they represent
had and/or are having a particular relationship with each actor on the roster. Two general
categories of relational contents are being specified, one is the communication ties and
the other is collaboration ties. Within each category, the respondents were given three
sub-categories illustrating time phases of: 1) before the earthquake, 2) during emergency
response period, 3) during the disaster recovery period. Therefore, a total of six
categories were illustrated for the respondents to specify whether they engaged in
communication or collaboration relationships, as well as the duration of each of these two
types of ties before and after the disaster event.
It is possible that one type of uniplex relationship may not be conceptually
distinguished clearly with another (Burt 1983). One example is the respondents’
difficulty in drawing the line between communication and collaboration ties in this
survey. Therefore, a brief explanation of the nature of these two kinds of content domains

75
was being provided at the beginning of the roster
4
section in the survey. I clarified two
aspects of the relational contents for the respondents. First, “communication” was further
defined as those activities related to “information exchange”. And “collaboration” was
clarified as those related to building up field projects together. Such referral elicits the
long term and the persistent aspect of the collaborative relationships. Secondly, due to the
lack of clear definition of the time span of emergency response and the disaster recovery
periods, I restricted the former to be within a timeframe from seven days to one year after
the earthquake event, thus including both the immediate response and short term
emergency response. The recovery period was restricted to anywhere between 2-5 years
or longer. Thus, by the time that the respondents were filling out the survey in 2011, it
was actually three years after the disaster event and would fall within the recovery stage.
One last clarification I make in this section is to define the nature of the civil society
actors. One of the main research purposes in this study was to trace the origins and the
development of the newly emerged grassroots groups after the earthquake. These actors
can be both formal and informal in terms of their institutional status. For the informal

4
The first section of the survey was composed of questions gathering information of actor attributes. The second part
of the survey was composed of a list of names of all the actors being examined in the study, and each actor can name
however many others that they have had communication and collaboration relationships before and after the earthquake.
The complete list of names is called a roster.

76
ones, I name them as “teams”/“social groups”. And for the formal ones, I call them
“organizations”. In order to reflect this conceptual consideration in the survey, I used the
phrase “teams and organizations” to include both types of civil society actors.
Another dimension of the central research question is to explore the dynamics of
the interactions among the actors inside the civil society, the state, and market domains.
One issue I encountered in the field when designing the survey questionnaire was the
format through which the actors inside the state and the market domains can be properly
incorporated into the roster list without further burdening the respondents’ loads. During
the emergency response and the disaster recovery periods, the Chinese government from
central to the local branches, were all actively involved in the process. If listing these
actors one by one in the second section of the survey, the roster would quickly become
unmanageable considering the time and efforts of the respondents. The similar issue
occurred with listing the market actors such as private enterprises and businesses that
participated in the two periods after the earthquake. Upon reflecting the purpose of my
research study, it became clear to me that the actors inside the civil society domain are
the key focus of attention. I therefore created a design to incorporate two macro-level
actors each representing the aggregate of all the micro-level actors acting inside the state
and the market domains. This way, when a civil society actor named a relation towards

77
one of these two actors, the tie itself would demonstrate the existence of a connection
between the civil society actor and any of those actors inside the state/market. The
direction of these ties can only be going from the former to the latter and not vice versa.
Since no actors inside the state and the market sectors were designed to fill out the survey,
the ties that are being named towards them will necessarily be one-directional without
any reciprocity. Essentially, the information being captured here can be regarded as a
type of cognitive mapping of the perceived cross-sector structure from the civil society
actors’ perspective.
Overall, a total of 138 actors comprised the roster list. Among them, 136 are civil
society actors. The state and the market were treated as two separate actors and listed
separately in the roster. Each one of the actors were given a unique code number so that
the identities of the actors would remain anonymous and being protected, consistent with
IRB guidelines. The final revision of the survey questionnaire was completed in the field
5
.

Qualitative Data Collection
The main purpose of the qualitative method in this study is to trace the
motivational origin of the action taken inside the civil society domain as well as the

5
Please see Appendix 3.4 for my field notes regarding the finalizing process.

78
driving factors behind the institutional dynamics being maintained over the long term.
The quantitative relational data captures the structural forms, patterns, dynamics and the
rules governing the evolution of the structural changes over time. However, they do not
provide sufficient explanation on the initial emergence of relational structures taking into
consideration of the social and cultural contexts within which they arose in the first place.
In general, “network analysts rarely capture the complexity of naturally occurring
relations. Their concern is less the complexity of the relationship between pairs of actors
than it is the complexity of the structure of relations among many actors as a system”
(Burt 1983, 35). But in order to understand the role of civil society in the Wenchuan
earthquake recovery, deciphering the complexity of how relationship originated and the
motivational factors that prompted certain role formation are as important as the concerns
for the complexity of the structures themselves. Therefore, qualitative data in this
research serves two complementary purposes: 1) explaining quantitative network results;
2) expand and built on the relational data.
In this study, I collected my qualitative data concurrently with the network data.
Since the qualitative research questions inherently build on the collection of relational
data, the initial purposeful sampling of individuals for collecting qualitative data overlaps
with the boundary specification process in relational data collection. The identification of

79
civil society groups and organizations as network actors was based on a combination of
positional and reputational approaches focusing particularly on those who actively
participated in the Wenchuan earthquake response and recovery in 2008. I then collected
three main forms of qualitative data and they are illustrated in the following sections.
The fieldwork was conducted over a period of four months in the metropolitan
region of Chengdu, Sichuan province. My primary qualitative data collection method was
in-depth interviews and direct observations. They were supplemented by reviewing
archival sources as well as audio/visual documents. The main advantage of collecting
data from various sources is that they help build an information-rich case study—a “thick
description” to use Geertz’s (1973) terms and convey the depth of the case by allowing
multiple perspectives from the experiences of the participants to be revealed.

In-depth Interviews of Group/Organizational Informants
As the survey questionnaire for the collection of relational data were first being
sent out to respondents through mass email distribution, three contact strategies were
being utilized to schedule follow-up interviews with informants: 1) phone conversations;
2) QQ messenger conversations; 3) email correspondence. In my actual field encounters,
these follow-up contact strategies worked together not only to help me confirming

80
interviews but also worked as another round of reminder for respondents to fill out the
survey questionnaires.
Initially, a total of 136 civil society actors were being listed on the survey roster.
And they can be described by three sub-categories. The first set consisted of domestic
formal NGOs that already obtained formal registration status. The second comprised
domestic informal social groups. Most of them were established after the Wenchuan
earthquake and had a grassroots nature. The third type includes those foreign NGOs that
were formally registered and established field offices in China. Since the main research
emphasis for this study was on the emergence and institutional change of civil society
domain, I primarily focused on interviewing informants from the first and the second
category. Essentially, the informants who later agreed to be interviewed were also
included among the survey respondents who filled out the original questionnaire.
I initially hoped to conduct all of my interviews in a semi-structured fashion and
completely audiotape them. But my first field encounter experience at the local
community level took the data collection strategy for a surprising turn
6
. This is because in
the field, I realized that the original semi-structured design needed to be complemented
by un-structured and open-ended interview questions to bring about the motivational

6
Please refer to Appendix 3.2.1 for detailed explanations.

81
aspect of the informants’ experiences. At times, field encounters would also happen at
unexpected circumstances which rendered audiotaping not possible. Therefore, I also
took interview notes to jog down the main conversations and key words being mentioned
by the informants.
For the semi-structured face-to-face interviews, I used an interview protocol to
help guide the main structure of the flow of my questions. The table in Appendix 3.5
shows some of the preliminary areas of focus and theme concentrations. As the interview
process unfolded, the focus of questions in each theme had undergone some minor
adjustments according to the varying experiences of informants. Each of the face-to-face
formal interviews lasted about 1-2 hours. In between the scheduled formal interviews, I
also conducted computer-mediated ones using an online messenger tool called “QQ”. The
online chat tool such as QQ was a very popular communication medium among Chinese
people in general and played an important role in facilitating contacts among volunteer
groups and organizations after the earthquake. The functioning of the chatting domain on
QQ is similar to those provided by Yahoo or MSN Messengers. A user name can be
chosen and a unique QQ identification number will be generated once the software is
downloaded and registered without any charges. As soon as I was provided with and
being permitted to use the list of civil society actors by the key informants from the

82
“center”, I realized that the only medium for contacting some of the respondents was
through their QQ identification number. I therefore registered as a QQ user and started
utilizing it as one of my survey distribution medium. Along the way, as I initiated
chatting sessions with my respondents, I found that they were more comfortable in
opening up and carry on the conversation as compared to phone contacts. This happened
especially for those respondents from the smaller and more grassroots groups. At times,
they would start elaborating on their stories and revealed the emotional side of their
experiences in participating in the disaster response and recovery. In these conversations,
I used more unstructured and open-ended format of interviews in order to bring about the
motivational side of my respondents’ experiences. Either at the beginning or towards the
end of each of our QQ conversations, I always made sure to confirm two things: 1) they
have received my survey questionnaire correctly and to encourage them to fill it when
time allows; 2) ask for other forms of contacts such as work phone number or email
address so as to arrange for possible in-person follow-up interviews. I then recorded and
saved the online conversations as part of my field interview folder on my computer.
Lastly, some of the advantages of collecting the interview form of data are worth
mentioning. Providing a private setting for face-to-face interviews with my informants is
critical in the assurance of confidentiality, in gaining their trust, and in creating an

83
inviting environment for them to open-up comfortably. This also allowed me to re-shape
my interview questions that are tailored to individuals so that maximal use of time can be
ensured and unique perspectives of each interviewee can be drawn upon (Padgett 2008).
Overall, the primary purpose of conducting the semi-structured interviews along with
open-ended questions was to approach the qualitative part of the research question
through a set of themes. And these themes were weaved through stories of how
informants lived through and experienced the earthquake event itself as well as the
response and recovery processes.
In summary, the in-depth interviews were collected regarding two of the
following groups. In the first group, a total of 16 interviews with NGO informants were
conducted and recorded. The respondents of the survey questionnaires would also be the
group/organizational informants thus being interviewed. Therefore, during each of the
interview sessions, the informants would reflect on both their own personal lived
experiences since the earthquake event and the institutional development experiences of
the different types of NGO groups/organizations that they were part of. In the second
group, a total of 8 interviews were conducted with a set of local community residents
whose unique experiences provided a “consumer” perspective of the disaster recovery
process.

84
Ethnographic Field Observations
In order to capture the depth of the informants’ experiences through the
earthquake event and how they led to the emergence and evolution of civil society
institutions, I also used the ethnographic direct field observations to accompany the
interview data collection strategy. This is because the social and cultural contexts from
which the actions occurred in the wake of the event of a disaster were all shared together
by this group of actors in the civil society domain. Exploring the origin of their actions
cannot be separated from understanding their shared beliefs and interpretation of the
meaning of civil society manifested through concepts such as solidarity, power, risk, and
resilience.
Throughout my stay in the field, I directly observed some of the activities
organized by civil society actors and recorded my observations through photos,
audio/tape, and written field notes/memos. On-site participant observations had
significantly helped me getting to know the group of civil society actors that I intended to
investigate, particularly in gaining their trust by becoming part of their daily life.
Revealing my identity and the purposes of my study in the field actually enhanced a
sense of trust on the part of my informants. And upon knowing the purpose of my
research study and having been reminded that I would be remaining in the field for some

85
time, many of the informants became less guarded and even expressed their gratitude for
me in knowing that their works were indeed being cared about and recognized. Every
field encounter with my informants became a bonding experience between me as a
researcher and the group of civil society actors that I strove to understand. I slowly came
to realize their side of reality and the type of cultural experiences that they shared and
shaped by the disaster event. The more sites I visited and the more interviews I conducted,
informants started inviting me to their fieldtrip activities, meetings, and gatherings. And
by being a participant observant in the process, an understanding of the lived-experiences
of the informants gradually became clearer to me. These pieces of in-depth information
regarding the culture of the group setting cannot be replaced by any types of face-to-face
interviews, or through other interview mediums.
Lastly, this kind of immersion of in the field through the observation of the
informants’ lives and works on the ground also helped me staying motivated when my
own field conditions turned out to be severe. A few of the field visits with my informants
ended up with rather difficult traveling and living arrangements. But in those
circumstances, a self-reminder of the original purpose of my research and my goal in
providing an account of the setting and actions that would be as truthful as to those who
experienced the processes eventually kept me staying encouraged and motivated.

86
From a researcher’s point of view, my ethnographic field experiences provided
me with enhanced clarity of the reality lived by the local people through their lens of
seeing the Wenchuan Earthquake recovery. By initially following the original interview
and fieldwork plans (see Appendix 3.1, 3.2, 3.3), I quickly recognized a pattern of how
people on the ground had experienced and framed the earthquake event and their
participation in the recovery process. This allowed me to have the opportunities to make
revisions to my interview protocols so that the life histories of my informants can be
closely reflected through my study. There were also some unexpected field obstacles
7

that prompted me to put a primary field investigation focus on gaining insights of the
actions inside the civil society from the original plan (see Appendix 3.2.1). With a
“detour” approach, I was able to look at the recovery process through cross-sector actions
from the network data later on being collected.

Documents and Audiovisual Recordings
I kept a field journal for three purposes. One was to record my own experiences in
my field encounters. The second one was to provide a supplementary account of some of
the interview sessions. For example, in some incidences, the field encounters happened in

7
See Appendix 3.2.2 for detail.

87
such a rush and there would not be enough time for me to prepare and get ready for audio
recordings. I would then use the journal to write down the key ideas and conversation
themes. Thirdly, I would use the notes to jog down some of the ideas that came to my
mind regarding the research work in general or in reshaping certain parts of my interview
protocols.
I also collected public documents in the forms of official government memos and
guidelines as well as online archival information about civil society groups and
organizations. The former was collected to illustrate the planning and policy context of
the Wenchuan earthquake recovery, as part of the general policy background of my study.
The online archival resources were retrieved for the purpose of depicting the
establishment and activity-related information of civil society actors.
In addition to the traditional types of documentation, I also gathered visual
materials during my field visits. I started gathering these materials when I first visited
Dujiangyan city in July 2008 with a group of international planning experts who were
invited by the Sichuan provincial government. My second trip back to Dujiangyan in July
2010 resulted in another round of collection of photographs recording the housing
recovery progresses of the local people. During my dissertation fieldwork in 2011, I

88
primarily took my own photographs in order to elicit the lives and works of my
informants. Sometimes, the informants would provide me with their own photographs,
video recordings, and written biographies.

Methods of Data Analysis
Sequential Data Analysis
Figure 3.4 below shows the data collection and analysis procedures. The arrow
signs represent the sequences of each step taken.

Figure 3.4. Concurrent Data Collection and Sequential Data Analysis Procedures
Source: (Creswell and Clark 2007) adapted by Lu, 2013
Network
Data
Collection
Network
Data
Analysis
-Data collected at the same time
-Independent of each other
-Group/organizational level and
individual level
Analyzed for different purposes:
-NETWORK data analyzed for outcome-related
questions
-QUAL data analyzed for motivational and
origin-related research questions
Qualitative
Data
Collection
Qualitative
Data
Analysis

89
I analyzed the data in a sequential manner with the first step to conduct the
structural analysis of the relational data. This is to reveal the patterns and rules that could
explain the emergence and dynamics of institutional structures within the civil society
domain. The qualitative data analysis follows as a second step in the data examination
procedure. Qualitative data, in this research study, was designed to expand and build on
the relational data in order to bring forth the agency origin and motivational sides of the
experiences and stories. Therefore, first providing a structural description of the
interaction activities within the civil society domain serves as a relational basis within
which qualitative data can then be utilized to explore the issues of agency and emergence.
The two types of data are designed to answer different sub-sets of the general research
question. But at the same, the qualitative data plays a complementary role for the network
data in providing a comprehensive understanding of the discourse of Chinese civil society
after the Wenchuan earthquake.

90
Quantitative Data Analysis
Analysis of Descriptive Statistics using UCINET
The Handling of Relational Data
The method for analyzing social structures is called “Social Network Analysis”
and it is a set of methods that particularly examines the relational aspect of structures
(Scott 2001). The first step I took was to organize the relational data being collected.
Recall that each of the actors was given a unique identification number in the survey
roster list. I constructed a “node list” data structure in Excel to store the original data
form. The node list data format is consisted of 138 columns and each column is
represented by the identification code for one actor. Following the unique code of each
actor, each row is consisted of the code numbers of all other actors that this “ego” actor
nominated in the original roster. Two separate files are created to store the
communication and the collaboration network relations. The actor attributes data was
stored in a “case-by-variable” data matrix where each actor is represented by a row while
each column stands for one of the three attribute variables: 1) date of establishment; 2)
registration status; 3) location.
The two sets of relational data were then imported into UCINET, a social network
analysis software package. The storage of relational data can be either in “incidence” or

91
“adjacency” data matrices. The incidence matrix is structured as case-by-affiliation
matrix with its rows representing actors and columns representing the three attributes as
affiliations. This “two-mode” rectangular incidence matrix can further derive two square
incidence matrices. One one-mode square matrix is case-by-case where both of it rows
and columns will represent actors. The individual cells will show the number of common
affiliations that a particular pair of actors has. The other one-mode square matrix is in the
form of affiliation-by-affiliation incidence matrix. Both of its rows and columns show
affiliation types such as location, registration status, and date of establishment. The
individual cells in this matrix will be showing whether each pair of affiliations is linked
by common actors.
In providing the structural description of the network data, I treated the actors as
cases and conducted the analysis by using actor-by-actor (case-by-case) one-mode
incidence matrices to illustrate the social structures of communication and collaboration
networks. In this particular study, the strength of relations among pairs of actors was not
collected. Therefore, each individual cell in the adjacency matrices represents only the
existence of a tie. Essentially, the adjacency matrices are composed of “1” and “0” entries
with the former representing the presence of a tie and the latter representing the absence
of a tie.

92
Since in the original survey, each actor was asked to nominate others that it had
initiated communication and ties towards, the relational data inside the adjacency
matrices will represent directions. For example, an entry of “1” in cell (2, 5) would mean
that there is a presence of relation from actor #3 to actor #5, being interpreted as the
former took the action to initiate contact to reach to the latter.

Sociograms and Graph Theory
In this section, I illustrate some of the formal languages used in graph theory to
describe network structures and other features. First, the structuration processes inside the
civil society domain are all represented by a set of graph diagrams derived from the
adjacency matrices. In a graph, each point, or node represents one civil society actor. And
the lines between the points can represent two types of relations: communication and
collaboration. All the analysis being conducted in this study are based on directed graphs
as the actors can take the initiative to reach out to others and the nominations might or
might not reciprocate. Thus, the directions are represented by attaching an arrow to each
of the directed line. If two actors are connected by a line, it can be said that they are
adjacent to each other. For communication networks, this means that one civil society

93
actor is directly connected to the other. If the direction of the arrow goes both ways, then,
pieces of information can flow back and forth between the two actors. For any one of the
actors, the group of other actors that it is adjacent to are termed its neighborhood. And
the number of actors in the neighborhood is called the degree of connection, a numerical
measure of the size of its neighborhood (Scott 2001). For directed relations being
investigated in this study, the total number of other actors that each ego actor directed
communication or collaboration ties towards are termed as out-degree of the actor. The
number of actors who had ties nominated toward the ego actor is termed as in- degree of
the focal actor. The out-degree measure in this study can be generally understood as the
agency actions shown on the part of the focal actor. If the measure is relatively high for
one civil society actor for a given period of time, it means that this particular actor’s
“agency freedom” (Sen 1999) is being activated and it was willingly to engage in
communication or collaboration relationships with other civil society actor.
Actors not only can be connected by direct lines of relationships, but also can be
reached through indirect lines of connections. In order to delineate these indirect
relationships, a few concepts need to be clarified. A path is measured by a sequence of
lines in a graph in which “each point and each line are distinct”…and the length of a path
is measured by the number of lines that make it up” (Scott 2000, 68). The distance

94
between a pair of actors is measured by the length of the shortest path (geodesic distance)
that connects them.

Multi-level Network Analysis
The structure of the section on network descriptive statistics is arranged in such a
way to include both micro and macro-level analysis of the communication and
collaboration network structures. At the macro-level, I first explored the general basic
demographics that can be used to describe the structure of the two types of network
contents. As part of the general outlook on the actions taken by civil society actors
comparing before and after the 2008 earthquake, changes in actor out-degree and in
degree measures in the communication network was examined. For each period,
subjective ranking categories were created to illustrate the key changes of the
composition of top-ranking actors over time. Then, the basic descriptions of the network
structures are being explored from a macro perspective, named “connection”,
representing how actors are connected to each other through a set of different sub-
categories. These network properties include: density, distance, diameter, geodesic paths,
and flow.

95
Following the general macro-descriptions of the communication and collaboration
networks, I explored the micro-level of the actors’ experiences by looking at the
“embeddedness” of actors for each type of structures over the three periods of time:
before the earthquake, emergency response period, and the long term recovery period.
The concept of “embeddedness” is examined through the measures of reciprocity,
transitivity, and clustering, group-external and internal ties. Reciprocity measure
examines the smallest feature of embedding that can be extracted by the dyadic ties
between a pair of two actors. The degree of reciprocity of the network structures and the
type of network relations are being the focus of the attention. At the second level, the
measure of transitivity was being explored for triadic relationships that involve three
actors. The existence and the process through which a tie is created to close the triadic
connections among three actors reveal the emergence of the most fundamental forms of
social relationships (Hanneman and Riddle 2005). The characteristics of the triadic
relationships among actors depict aspects of hierarchy, equality, and exclusivity of
groups in a directed network. In the third level of the “embedding” analysis, I look at the
tendency for the two types of networks to have dense local neighborhoods or clustering.
The clustering measure expands from reciprocity and transitivity to see how actors are
embedded in its own local neighborhood. Higher degree of overall graph clustering in the

96
communication networks at any given point of time period, for example, would mean that
there are concentrations of information sources existing inside the network structure.
Identifying the actors with larger neighborhoods builds up the foundation of further
investigating the characteristics and the emergence of these actors. The fourth level of
embedding deals with the macro-level analysis of how ties exist inside the pre-
determined group as compared to outside the group. In this study, I am primarily
interested in the patterns showing how the group of actors with formal registration status
would interact with the group of actors without such status. The E-I Index was being
calculated and examined across groups and actors.
The macro-level analysis of the social structures is examined from two
perspectives. One is the “top-down” approach, which is chosen to look at the network
structures as a whole to detect ways within which the overall structure can be
decomposed into smaller units. This is another way of understanding “solidarity”
revealed through how actors were involved in group-selection. Actors are not only
actively building their own neighborhood, their decisions and behaviors are also
influenced by the larger sub-structures. Identifying these sub-structures and their patterns
of dynamics over time provide an outlook on the possible structural constraints or
opportunities faced by actors when they were constructing the networks at the same time.

97
The measures being analyzed in this section include: component, K-core, and
Community structure detection using Girvan-Newman Analysis
8
.
The second perspective in examining groups and sub-structures can be named as
the “bottom-up” approach (Hanneman and Riddle 2005). It is designed to understand
how the overall structure of the networks emerges from and linked together by smaller
units. I chose to study three kinds of sub-graphs: cliques, two-clique and two-clan. A
clique is a sub-structure that represents the most tightly inter-connected relationship
among its members. It is the “maximally complete sub-graph” where all the possible
pairs of points are directly connected by a line or are adjacent to one another (Scott 2000,
114). The emergence and changes in the clique structure provide indicators of a process
that can help explain the sub-grouping behaviors of the civil society actors. The two-
clique and the two-clan concepts relaxed the strictness in defining a clique and allowed
me to expand beyond the direct relationships. Overall, the micro and macro level of
analysis of the sub-structures of the communication and collaboration networks are
designed to explore the structural emergence and transformation of civil society
institutions after the earthquake. It allows me to explicitly point out the specific patterns

8
The Girvan-Newman Analysis is a method used to identify community structures in a given network environment and
it is made available in the UCINET social network analysis software.

98
of the structuration process inside the civil society domain, as well as the structural
change patterns across the domains of civil society, the state, and the market system.
The last two sections of the descriptive analysis are designed to explore how role
identities are being formed among civil society actors in the two types of network
relations. I also used this part of the analysis to re-examine two key concepts implied by
the central research question from a network perspective. One is the definition of the idea
of “power”. In the context of this study, this concept is not examined through actor
attributes but rather with a relational lens. A group/organizational actor can be “powerful”
not by the more readily observable outside traits, status or possessions, but by the way it
relates to others in a particular type of network setting. The measures being calculated
and analyzed in this respect are degree centrality and betweenness centrality. The other
related idea is how “roles” can be defined in the discourse of institutional change of civil
society after a catastrophic event in China. This part of the analysis is designed to look at
how “new” positions and roles can emerge as the result of agency actions. The term “new”
is being used because at times, one actor might communicate and collaborate in a
“similar” manner as compared to another actor, but neither of them will recognize such
relational patterns within which an emerging “role” can be defined. In these cases,
crystalizing these “hidden” identifications is a constructive step in conceptualizing

99
Chinese civil society in times of disaster response and recovery. I calculated the measures
for actor structural equivalence using the CONCOR method to execute this section of the
analysis. At last, a simple homophily model was proposed by using ANOVA Density
Models in UCINET. The main purpose of this analysis is to examine whether there is a
tendency for civil society actors to have a preference for within group ties based on their
registration status.

Longitudinal Modeling of Network Dynamics using RSIENA
Part I of the quantitative analysis is designed to lay down the conceptual and
structural frameworks within which the patterns of interactions among actors emerge and
being sustained over time. A set of specific mechanisms of actions and processes of
interactions among civil society actors are uncovered and depicted explicitly through
structural presentations. However, the descriptive part of the analysis is not sufficient in
specifying the underlying patterns of network behavioral change that can help explain the
dynamics of network evolution from before the earthquake to long term after the disaster
event. In order to further look into this aspect of the story, Part II of the quantitative data
analysis is designed to explain the impact of previously-found network patterns on
behavior and attitude, reflected by actor’s choice of ties.

100
This part of the analysis predominantly used the Longitudinal Modeling technique
in network analysis (Snijders et al. 2010) to particularly investigate the rules governing
network evolution over time. I used the Stochastic Actor-based (SAB) Models (Snijders
et al. 2010) to conduct my longitudinal analysis on network dynamics. The modeling
process is based on the paradigm of statistical inference. There are two advantages for
adopting the SAB models in this study context. First, the model is specifically designed
to represent network dynamics based on observed longitudinal network data. Two types
of such data were collected initially. One was the communication network data over the
time periods of before the earthquake, emergence response state, and the recovery stage.
The other is collaboration network data over the same periods of time. The second
advantage of the model arises from one of its underlying assumptions specifying that
actors can change their outgoing ties based on their and others’ attributes, positions in the
network, as well as their perceptions of the network as a whole (Snijders et al. 2010).
Essentially, the model takes into consideration of both agency and structure and provides
the flexibility for the inter-dependent factors to change over time. This is in direct
accordance with the theoretical backdrop of this research: the paradox of action and
institutional structure exemplified through the role of Chinese civil society in catastrophic
disaster recovery.

101
I then used the SIENA methods
9
implemented in the R statistical system
(RSIENA) to execute the analysis. Inside the first sub-section, I provide a specific
treatment strategy for the missing data. The procedures involved using a core-periphery
analysis in UCINET and creating a composition change file in RSIENA. I then provided
model specifications for examining the dynamics of uniplex networks and the cross-
dependencies between the communication and the collaboration networks.
Due to the particularity of this dataset pertaining to a rather drastic shift in
network ties especially comparing before the earthquake to immediately after the event, I
further look into the issues of “time heterogeneity” and “unconditional modeling
estimation”. I designed a set of modeling specifications that can incorporate the time
heterogeneity issue into the estimation processes. At last, the unconditional modeling
method is chosen based on this specific nature of the longitudinal network data.

9
SIENA is a statistical analysis program to examine network data. The name SIENA stands for Simulation
Investigation of Empirical Network Analysis.

102
Qualitative Data Analysis
Rationale for Sequential Data Analysis
Recall that the structural analysis of the relational data consists of two parts. One
part is the descriptive analysis of network structures. The results of that section
demonstrated that certain civil society actors occupied particularly “important” roles in
terms of their embeddedness, centrality, positions, and roles. And the mere presentation
of the actor attributes such as registration status, date of establishment, and geographical
locations cannot decipher the cognitive and motivational origins regarding how these
actions emerge. This issue became particularly apparent when trying to explain the
patterns of emergence and role occupancy of: 1) the smaller grassroots civil society
groups that were established after the earthquake; 2) the formal civil society
organizations that established first field offices in Sichuan after the earthquake event.
Although the second part of the longitudinal statistical analysis detects factors that
contribute to the network evolutions, the models offered a more forward-looking
perspective while taking each state of structural existence at different time stages as
given. In other words, explaining the source of emergence by taking into the social and
cultural contexts within which the agency actions arose is not a built-in focus of the
second part of the quantitative data analysis. This then, left a critical part of the original

103
research question unanswered, and it is the sources of agency inside the civil society
domain transitioning from before to after the disaster event, as well as the driving forces
for the durability of agency reflected in the processes of institutional changes through the
recovery period. Therefore, partially building on top of the results from the quantitative
analysis, a qualitative data analysis is designed to fill in the gaps in understanding the
structural discourses of Chinese civil society after the Wenchuan earthquake.

Coding and Thematic Categorization Using ATLAS.ti
First of all, I transcribed all of my interviews and stored them in separate files
arranged by the names of the actors. The reason that I chose to transcribe my own
qualitative interview data is because of my familiarity of the contexts where the
conversations were being held. This enhanced my ability to: 1) “fill in the unclear
messages” with background understanding; 2) “insert appropriate explanations and
clarifications” (Padgett 2008, 135). In order to stay close to the meaning of the account
provided by my informants, I conducted minimum editing and revision of their original
message, and used brackets only in times of clarification of certain terms.
In the survey questionnaire, I provided the respondents written promise to
maintain strict confidentiality by using unique identification codes for each actor.

104
Therefore, the actors’ names appeared in the interview conversations were all replaced by
their identification numbers when transcribing. Also throughout the process, I transcribed
the complete conversations with each informant one by one in order to maintain the
accuracy of the data and also as a way of showing respect for the information and stories
being offered by my informants. Since my fieldwork was conducted in Sichuan province,
in mainland China, the original transcription was in simplified Chinese.
The transcriptions in its original language are then imported into ATLAS.ti
qualitative data analysis software. The transcribed conversations with each informant are
maintained as separate files so as to make it easier for future retrieval. Throughout the
coding process of my interview transcriptions, I kept the files in their original Chinese
language. This is because as a native speaker of Chinese and having done the fieldworks
in China, the original search for meanings and interpretations will come more naturally
and come closer to the first-hand messages being passed on by the informants. This also
helped me stay alert and keenly identifying some of the hidden meanings that went
beyond the surface descriptions so as to maintain the purposes of finding “thick” and
“deep” ethnographic interpretations uncovering “the tacit meanings of cultural beliefs and
practices” (Padgett 2008, 140).

105
At the beginning of the coding process, I created code labels to relevant segments
of the transcripts based on the following primary topical ideas: 1) action and motivation;
2) the incipience of civil society; establishment process; 2) civil society, state, and market;
3) institutional formation, function, and structure; 4) self-awareness and solidarity
formation; 5) institutional sustainability motivations; 6) emergency response and
recovery activities; The decision to use these labels originated from the research question
that the qualitative part of the study intended to answer. For one, it is to investigate the
emergence of the key civil society actors found in the quantitative analysis. The accounts
of the informants further guided me to break down this “emergence process” into those
related to individual experiences, including individual action and motivations; and those
related to institutional formation when activities started to show patterns of group
formation. Secondly, I intend to investigate the long term institutional change that the
actors experienced over time, and the driving factors supported their sustainability.
Subsequent to the initial coding process being carried along after many more transcripts, I
gradually recognized a pattern related to how informants told their stories. The topical
ideas that I initially used to create my primary code labels can be further collapsing into a
set of categories
10
such as: 1) Individual-level actions and motivations; 2)

10
This further synthesis of categories did not prompt further changes to the original coding.

106
group/organizational actions that represent primary emergence of civil society; 3)
institutional change processes, both within sector and across sectors; 4) individual and
actor-level sustainability motivators. These categories also emerged in a chronological
order as each of the informants reflected the changes they experienced in both of their
personal lives and their engagement as part of the larger civil society actions after the
earthquake event. At this stage of the analysis, the coding process became more selective
as I pondered how the categories thus generated are interrelated to each other and in what
ways they can be organized as part of a conceptual framework.
As the original codes were being synthesized and the categories were being
compared in the last phase of the analysis, a set of themes and categories became
apparent (see Appendix 3.6). I organized these properties into three sets of themes that
can be ordered chronologically to illustrate a process-oriented approach in discovering
the sources of agency action in the civil society domain.
At same time when the three themes and their sub-themes were developed, I was
still open to generate new codes as new topics might emerge upon extracting meanings
from the stories being told by the informants. I stopped the process when eventually no
new codes were needed and the primary interpretations of informants’ accounts no longer

107
surpassed the already existing themes and their sub-themes. I also recorded memos in
notebooks in order to jog down the possible categories of interpretations. Incorporating a
multidisciplinary perspective into the interpretation process was helpful in bringing about
the different ways in understanding the accounts of informants. For example, one civil
society actor was engaged in activities related to community health issues after the
earthquake and the informant was reflecting on her original motivations into this aspect
of disaster recovery. Some knowledge in how social supports can be built through social
network interventions in promoting health in communities would be a plus in designing
and expanding the conceptual frameworks thus formulated.

Interpretation and Presentation
11

Further interpretation and presentation of the analysis was then organized into an
actor by actor format for illustration. For the case of each group/organizational actor,
both description of the case and the relevant interpretations were presented according to
the themes being developed in the last stage of the coding process. The highly relevant
direct quotes that can demonstrate the particular interpretation point being made or can

11
Please refer to Appendix 3.7 for discussion regarding the validity and reliability of data.

108
illustrate the contexts within the themes were being developed were then selected and
inserted.
I reserved the translation process into this last stage of analysis, after all the
relevant segments of quotations were embedded into the conceptual discussions and all
interpretations were completed. This is to reserve their original meaning in Chinese with
as little secondary interpretive obstructions as possible. One main obstruction is to
translate the passages too early on in the coding process or in the middle of the writing
process, which might lead to incomplete representations of the informants’ point of view
and the essence of the message that they were trying to pass on through their stories.
Therefore, I conducted my translation procedures
12
towards the end of the presentation
and the writing process. Another advantage in this kind of strategy for designing the
timing for conducting translation is that it can serve as a “safety net” to spot for any
pockets of incompleteness in the earlier meaning-making process.
Throughout my writing process in reporting the results of the qualitative analysis,
I used the interview accounts of informants in three types of quoting procedures
(Creswell 2007). I primarily used longer quotations in order to convey the complexity of
various processes. This is in accordance with the central phenomenon that the qualitative

12
All translations are my own.

109
section was designed to investigate. The emergence of agency action and the institutional
development in the civil society domain after the disaster event are themselves revealed
by a set of processes. In order to fully understand how the crisis event was experienced,
interpreted, and enacted through agency behaviors, this kind of longer quotes is used to
bring about the nature of the processes as close to the cultural context of the informants
as possible. The second type of quotation strategy is the usage of direct dialogues either
between the informant and me or between several key informants. Sometimes, the central
ideas being conveyed by one informant were being brought up through relatively short
statements, and to simply single those out as a shorter quote would not be useful in
understanding the context from which the ideas or thoughts arose. Thirdly, shorter
embedded quotes were also being used to be inserted into my narrative and
interpretations. This is done to emphasis on pinpoint a particular point with evidences
that can directly and clearly support the discussion of a theme.

110
Chapter 4
Tracing Actions and Processes

Overview of Actor Characteristics
One-mode Network Data Description
In this study, survey questionnaires were distributed to a total of 136 groups and
NGOs who have participated in the recovery of the Wenchuan Earthquake since 2008. A
total of 63 questionnaires were returned. The response rate was 46.32%.
In table 4.1 below, I illustrate civil society actors according to their attributes:
Table 4.1. Attribute Data Description
Non-registered social groups 21.3% (29)
Sichuan-based grassroots groups and NGOs 75.7% (103)
Established after 2008 Wenchuan
Earthquake
57.4% (78)
Note: Percentages and counts are based on 136 civil society actors.
The percentages in each category were calculated using the attribute data of 136 actors.
Although not all of them responded to the survey, it is possible that the non-respondents
could be nominated by other respondents
13
. In these incidences, network ties would be

13
In social network analysis, when one actor names another actor as communication or collaboration partners, one tie
between them exists. Such a tie is one-directional if the other actor does not respond to the survey. However, this uni-
directional tie can be recorded as part of the network structure with the inclusion of the nonresponsive actor.

111
counted as existing connections but without the possibility of being reciprocated. It is
therefore important to capture the comprehensiveness of the variety of attributes among
all actors.
First of all, among all the civil society actors being surveyed, non-registered
informal social groups constituted 21.3%. The emergent social groups formed after the
disaster was being accounted for in this category. By “emergent”, I mean the grassroots,
spontaneous, and the voluntary nature of these groups. In the Chinese context, the
registered civil society actors are further divided into two sub-categories. One is the
formal status of registration with the Ministry of Civil Affairs. The other one is the status
registered under the business category
14
.
Among the 136 civil society actors, 103 were based in Sichuan Province. This
category included not only the grassroots groups and NGOs formed locally, but also the
actors who established long term field offices for disaster recovery in Sichuan after the
2008 earthquake. Together with the “locally-born” actors, they counted as 75.7% of all
the civil society actors. The rest in this category are either domestic NGOs practicing
short-term for disaster response or international NGOs that had previous practicing

14
See Appendix 4.1.

112
experiences in China but did not have formally registered field offices within Sichuan
Province.
Another category I distinguished was based on the date of establishment of the
civil society actors. The primary purpose of this part of the design was to see whether and
to what degree did the earthquake event triggered agency actions inside the civil society
domain. A total of 78 social groups and NGOs came into being after the disaster and they
contributed to 57.4% of all the actors.

Two-mode Network Data Description
Two-mode network data was also collected in order to look at the specific
activities that the actors engaged in during the disaster recovery period. I distinguished 7
types of recovery activities in general and they were listed in table 4.2 and figure 4.1
below:

113
Table 4.2. Civil Society Participation in Earthquake Recovery Activities (70 actors
15
)
Types of Activities Counts Percentage
Housing 16 22.9%
Elder and disabled 28 40%
Women and children 32 45.7%
Environment 24 34.3%
Psychology 24 34.3%
Livelihood development 31 44.3%
Other 32 45.7%

Figure 4.1. Recovery Activities of Civil Society Actors
As figure 4.1 shows, recovery works related to the disadvantaged groups such as women
and children ranked the highest among all others. Livelihood development was another

15
See Appendix 4.2 for further explanation.
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
30.00%
35.00%
40.00%
45.00%
50.00%
Recovery Activities of Civil Society Actors
percentage

114
aspect of recovery that drew long term devotion of civil society actors. Some of the
functions being performed in the “others” category are worth mentioning. One area of
engagement that was named regularly was activities related to community re-construction
and re-development. Various terms were used by the respondents to describe their
activities under this category, and they include: “social recovery”, “community capacity-
building”, and “Integrated service provision to local communities”. In addition, survey
respondents also provided information regarding the types of their recovery activities
under the “other” category. They were: 1) Participating in community infrastructure
reconstruction such as water pipelines, community playground, and roads; 2) Financing
school reconstruction; 3) Basic infrastructure rebuilding; 4) Providing family education
and other educational supports; 5) Medicine delivery; 6) Information exchange
facilitation; 7) NGO capacity building and rural cooperatives capacity-building; 8)
Material resource collection and distribution; 9) General supportive type. From table 4.2,
it is clear that participation in these “others” category contributed to 45.7% of civil
society actors’ recovery activities, which is a tie with the activity related to “women and
children”.

115
Perceived Institutional Factors for Resilience-building
In the original survey, I also asked my respondents to name the factor(s) that they
perceived to be most important in enhancing the civil society’s capacity for disaster
preparedness. Table 4.3 and figure 4.2 illustrate the results:
Table 4.3. Perceived Importance of Institutional Factors for Resilience-building (70
Actors)
Types of Institutional Factors Counts Percentage
Government support 42 60.0%
Private enterprise support 11 15.7%
Social support
(collaboration with other groups and NGOs)
32 45.7%
Preparedness coordination
(communication platform-building )
26 37.1%
Other 13 18.6%

116

Figure 4.2. Perceived Resilient-building Factors
Civil society actors did perceive government support to be the most critical in their own
capacity-building, particularly in preparing for catastrophic disasters. Interview data
revealed that the activities of “support” from the government side can come in a variety
of forms such as recognition and acceptance of the existence of the civil society actor, or
provision of equipment and activity space. The second factor perceived to be critical was
making connections and collaborating with other civil society actors. Communication
platform building is also high on the list when it comes to preparedness and mitigation.
Since all of these factors were related to the qualities of how civil society actors would
hope to have or develop in the long term, I named them the factors for “social resilience”.
This concept is also developed further in terms of the actions put forth by the civil society
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
Perceived Resilient-building Factors
Percentage

117
actors throughout the rest of the study. Here, I focus on the perception aspect of the
picture. The perceived functions being named in the “others” category include: 1) the
professionalization of NGO services; 2) the role of professionals and organizations for
educating people to perform self-help immediately after a disaster; 3) risk awareness and
risk management skills; 4) practicing drills for disaster mitigation and preparedness,
educating the general public; 5) conceptual change; 6) societal support.

Overview of Network Structure Characteristics
Communication Network
The following three sets of network centralization
16
measures (see table 4.4)
illustrate a consistent increase in information exchange activities comparing the pre-
earthquake period, the emergency response period, and the recovery period.

16
Please refer to the network concept glossary tables (1) and (2) in Appendix 4.3.

118
Table 4.4.
17
Comprehensive Communication Network Characteristic Measures
Comparing Three Time Periods (138 actors)
Pre-earthquake

Emergency
Response

Recovery
Centralization
(cohesiveness)

Network Centralization
(Out-degree)=25.244%

Network Centralization
(In-degree) =6.862%

Betweenness
Centralization: 2.71%

Network centralization
(out-degree)=94.518%

Network
Centralization
(in-degree)= 21.727%

Betweenness
Centralization:
13.03%

Network Centralization
(Out-degree)= 92.90%

Network Centralization
(In-degree) =18.642%

Betweenness
Centralization:15.14%

Sub-groups
(micro-
structuration for
information
exchange and
communication)
4 cliques found

Weak components: 63

Strong components:121

Largest Strong
Component: 18 nodes
25 cliques found

Weak components: 1
(actor 1 being isolate)

Strong components:
91

Largest Strong
Component: 48 nodes
47 cliques found

Weak components: 1
(actor 1 being isolate)

Strong components: 91

Largest Strong
Component: 48 nodes

Network-level
analysis
(Action)

Number of ties=230

Density=0.0122

Reciprocity=0.0900

Transitivity: % of
ordered triples in which
i-->j and j-->k that are
transitive: 31.37%

Number of ties=1028

Density =0.0544

Reciprocity=0.0925

Transitivity: % of
ordered triples in
which i-->j and j-->k
that are transitive:
34.74%

Number of ties=1193

Density=0.0631

Reciprocity=0.1067

Transitivity: % of
ordered triples in which
i-->j and j-->k that are
transitive: 37.97%

17
Refer to Appendix 4.3 for definition of the network concepts

119
The out-degree network centralization increased from 25.244% to 94.518%. In-degree
network centralization increased from 6.862% to 21.727%. The betweenness network
centralization increased from 2.71% to 13.03%. This means that the whole
communication network might have experienced a structural change with substantial
amount of concentration in terms of out-degree during the emergency response period.
That is, the positional advantages could be rather unequally distributed in the overall
network. On the other hand, the networks showed significantly less concentration in
terms of in-degree (prominence) when compared to out-degree centralization. The
betweenness centralization also experienced a significant surge immediately after the
earthquake. This means that a higher amount of connections could be made in the
network with the aid of intermediary actors connecting others who otherwise never would
have known each other. When comparing the measures across the two time periods from
before to immediately after the earthquake, the power of civil society actor changed
substantially in terms of their ability to initiate information exchange ties.
During the long term recovery stage, the measures of network centralization
showed little variation with the emergency response period. This was represented by only
a slight decrease in out-degree and betweenness centralization, and a slight increase in in-
degree network centralization. In the long term, actors were getting to know each other

120
better and were not only able to recognize but also to stabilize their communication
relationships with others.
Knowing how a social group or NGO is embedded in a network sub-structure is
also important in understanding how certain traits or behaviors are developed. The
communication and information exchange ties among the actors in a group can have a
profound impact on the ways how these actors participate in the recovery process and
perceive their future trajectories in terms of organizational development. At the most
basic level, a clique is a sub-set of a network in which the actors are more closely and
intensely tied to one another than they are to other members of the network. The increase
in the number of cliques comparing before and after the earthquake represents a micro-
restructuration process being triggered by the disaster. For example, there were 4 cliques
before the earthquake and the number quickly increased to 25 during the emergency
response stage, and to 47 during the recovery stage. This means that as the
communication ties became increasingly weaved together, actors were also able to
cultivate close and tight relationships with others.
I further examined the overall communication actions taken inside the civil
society domain. The total number of ties increased from 230 before the earthquake to
1028 during the emergency response period, and to 1193 during the recovery stage.

121
Density, which provides an index of the degree of dyadic connection when all 138 actors
were taken into consideration, showed consistent increase over the three periods of time
from 0.0122 before the earthquake to 0.0544 in emergency response period, and further
to 0.0631 in recovery stage. The percentage of pairs of connections that were being
reciprocated also increased gradually, from the pre-earthquake period of 9% to
emergency response period of 9.25%, and to 10.67% in the recovery period. This means
that there was a gradual increase in cohesion and trust among actors over time.
Transitivity displays a type of balance when actor A directs a tie to actor B, and B directs
a tie to C, then A also directs a tie to C. The level of transitivity was at the 31.37% level
before the earthquake. It increased to 34.74% in the emergency response period, and to
34.97% during the recovery period. At this stage of analysis, it can be concluded that
there seemed to be an emerging pattern of institutionalization of the reconstructed social
structure. Table 4.5 below shows a selected number of those actors that ranked relatively
higher than the rest of the others in terms of their outreach ties and incoming ties over the
three time periods.

122
Table 4.5. Communication Network Three-Period Comparison of Local Centrality
Measures (138 actors)
Actors with highest in-degree
(receiving ties)
Actors with highest out-
degree (initiating ties)
Pre-earthquake #50,#1,#94,#51,#119,#118 #51,#61,#115,#6,#137,#100

Emergency response #3,#1,#49,#51,#2 #24,#34,#3,#32,#51,#6

Recovery #3,#1,#49,#51,#27,#38,#24 #3,#32,#93
Notes: Numbers represent actor ID identification
In this study, actor #51, a Chinese non-profit incubator whose Chengdu branch
was established after the earthquake, had consistently been a key player whom others
tended to reach and contact for information. For both outreach and incoming nomination
activities, changes can be observed when comparing before and after the earthquake. For
example, domestic non-Sichuan-based civil society actors such as #94 and #50 lost their
central positions in terms of the intensity of being reached out by others after the 2008
earthquake. Actors #49 and #3, both formed locally for the cause of earthquake response
and recovery quickly emerged and gradually sustained to be the key players in
information exchange and communication. Sichuan-based civil society actors, such as
#32 and #93, also emerged to become more active in reaching out to others in the
network. Figures 4.3(A, B), 4.4(A, B), and 4.5(A, B) illustrate a visual representation of
the changes of in-degree and out-degree centrality measures among actors over time. As

123
is shown in the circular layout graphs, each line connecting one actor to another
represents an information exchange tie was being established. It can be concluded that
right after the disaster, not only the communication network became further integrated by
forces drawing in previously isolated actors, but the sheer availability of the number of
information exchange channels in the network increased significantly, thus constructing
denser communication neighborhoods surrounding each actor.

124

Figure 4.3A. Communication Network (Pre-Earthquake)
18

18
The cluster of actors on the bottom of the right hand side of the figure were those who did not have any
communication ties with others. Thus, they are counted as isolates and not being included in the connected network.
Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree)

125

Figure 4.3B. Communication Network-Pre-Earthquake (Circular Layout)

Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

126

Figure 4.4A. Communication Network (Emergency Response)

Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

127

Figure 4.4B. Communication Network-Emergency Response (Circular Layout)

Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

128

Figure 4.5A. Communication Network - Recovery

Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

129

Figure 4.5B. Communication Network-Recovery (Circular Layout)

Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

130
Collaboration Network
I now provide an overview of the structural changes in the project collaboration
networks after the earthquake. Examining table 4.6 below, the betweenness centralization
increased from 0.19% to 7.59%, which are at a lower level when comparing across all
three time periods with those of the communication network. This means that although
the percentage of intermediaries that the actors depended upon increased, the
collaboration network as a whole tended to have less concentration on ties connected
through intermediaries than the communication network. The collaborative network was
also less clique-oriented as the number of cliques increased from 0 to 1 and maintained
the same through the recovery period. This can be interpreted as a sign showing an
embracing nature among actors in a collaboration network environment. They were less
likely to seek out certain characteristics of other actors as the basis for establishing
project collaborations for earthquake response and recovery activities. The out-degree
centralization measure increased from 6.212% to 22.814%. Over time, the collaboration
network is also becoming more centralized around a particular set of actors in terms of
their actions in reaching out toward others.

131
Table 4.6. Comprehensive Collaboration Network Characteristic Measures
Comparing Three Time Periods (138 actors)
Pre-earthquake Emergency
Response
Recovery
Centralization
(cohesiveness)

Network Centralization
(Out-degree) =6.212%

Network Centralization
(In-degree) =6.948%

Betweenness
Centralization=0.19%
Network
centralization (out-
degree)=16.362%

Network
Centralization
(in-degree)= 16.362%

Betweenness
Centralization=5.53%
Network
Centralization (Out-
degree)= 22.814%

Network
Centralization (In-
degree) = 15.462%

Betweenness
Centralization=7.19%

Sub-groups
(micro-
structuration for
information
exchange and
communication)
0 cliques found

Weak components:92

Strong components:
134
Largest Strong
Component: 3
1 clique found

Weak components: 38

Strong components:
117
Largest Strong
Component: 19
1 clique found

Weak components: 39

Strong components:
111
Largest Strong
Component: 30

Network level
analysis
(Action)

Number of ties=76

Density=0.004

Reciprocity=0.0556

Transitivity: % of
ordered triples in which
i-->j and j-->k that are
transitive: 22.62%

Number of ties=241

Density =0.0127

Reciprocity=0.0856

Transitivity: % of
ordered triples in
which i-->j and j-->k
that are transitive:
19.35%

Number of ties=272

Density=0.0114

Reciprocity=0.0924

Transitivity: % of
ordered triples in
which i-->j and j-->k
that are transitive:
18.46%

As table 4.7 shows below, actor #1 and actor #2 maintained to be among the most
“popular” project collaboration partners throughout the three time periods. It can be

132
recalled that actor #1 represents all the active local government agencies in the state
domain while actor #2 represents the private enterprises inside the market domain.
Table 4.7. Collaboration Network Three-Period Comparison of Local Centrality
Measures (138 actors)
Actors with highest in-
degree (receiving ties)
Actors with highest out-
degree (initiating ties)
Pre-earthquake #1,#2,#37,#100,#110 #137,#61,#100,#70,#93,#119,
#104

Emergency response

#1,#2,#3,#49,#119,#51

#3,#24,#119,#135,#137

Recovery #1,#2,#3,#51,#119

#3,#135,#119,#51,#24,#93,#134,
#32

I interpret this type of network behavior as perceived “preferences” (from the perspective
of tie initiating actors) for reaching out to the government and the private sector for
collaborations both before and after the earthquake. Whether this “preference” emerged
out of a motive considering the institutional necessity of the social groups and NGOs in
order to survive and sustain over time or out of altruistic factors to better serve the local
communities will be dealt with in another section. For now, I continue examining the
characteristics of those actors who were “prominent” or “popular” (high in-degree) and
“active” (high out-degree) in the collaboration network environment.

133
Before the earthquake, actors #37 and #100 were being reached by many others,
immediately following actors #1 and #2. Both of the former two actors were registered
Sichuan-based NGOs. Actor #110 was the only international NGO that was being
nominated by many others within the in the network. However, this actor’s place was
replaced by other local grass-roots groups after the earthquake. Actors #3, #49, #51, #119
became the key collaboration partners being named by other actors in the network. The
first two were grassroots Sichuan-based actors. During the emergency response period,
both #3 and #49 participated in being the hubs for information coordination among
groups and organizations that came into the area. While actor #3 maintained its
specialization in coordinating activities among actors through the recovery period, actor
#49 focused more on community social recovery in different locations in Sichuan
province. Actor #119 was part of an extension program of an NGO based in Hong Kong.
It had been active in terms of initiating collaboration ties since the pre-earthquake stage
and attracted attention from more civil society actors especially right after the earthquake.
Figures 4.6(A, B), 4.7(A, B), and 4.8(A, B) illustrate a graphical representation of the
changes of the in-degree and out-degree centrality measures for the collaboration
networks. In general, compared to the communication structure before the disaster,
collaboration activities were more dis-integrated with a larger number of isolated actors

134
and one completely separated sub-structure being self-sufficient on the periphery.
Immediately after the disaster event, the circular layout shown by Figure 4.7B illustrated
that despite of the existence of isolated actions, there was an expansion of ties with more
actors collaborating with each other. Such increase in network density continued through
the long term recovery stage as shown in figure 4.8B.

135

Figure 4.6A. Collaboration Network (Pre-Earthquake)
Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

136

Figure 4.6B. Collaboration Network-Pre-Earthquake (Circular Layout)

Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

137

Figure 4.7A. Collaboration Network (Emergency Response)

Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

138

Figure 4.7B. Collaboration Network-Emergency Response (Circular Layout)

Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

139

Figure 4.8A. Collaboration Network - Recovery

Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

140

Figure 4.8B. Collaboration Network-Recovery (Circular Layout)

Size: Out-degree, Color: In-degree (darker blue with higher in-degree, orange with lower in-degree

141
Agency Freedom Initiation and Capability Formation
Initiation of Agency Action (Out-degree)
In this section, I provide a more detailed account of the social structuration
processes within which civil society actions emerged throughout three time points: before
the Wenchuan earthquake (t1), shortly after the event (t2), and during the long-term
recovery period (t3). Two types of institutional environments are investigated specifically
and they are: communication and collaboration. An important feature of this level of
analysis is that it allows for an in-depth understanding of the duality nature of action and
structure. On the one hand, actors are part of the social structure and their actions are
dependent on whom they are connected with, the types of connections, and the depth of
connectedness. In other words, actors’ constraints and opportunities in terms of building
relationships arise from the way they are positioned inside the network structures. On the
other hand, the structuration processes of the two types of network environments are in
turn constructed by or grounded in the actions among actors. In the context of this
research, actors not only made decisions based on how they were connected in the
structure, the way each actor formulated its own connection decisions toward others over
time was also consequential to how the structure as a whole was integrated.

142
I utilized network descriptive statistics generated by the UCINET network
analysis program to examine the general features of the communication and collaboration
environments. I maintained all 138 actors in the original dataset even though only a
subset of them actually responded to the survey. This was made possible for the
following reason. From a network methods point of view, one important nature of the
network data is that when a tie is being nominated by one actor towards the other, the
reported existence of such a connection can be counted as in existence even if one party
did not respond to the survey. In other words, the 63 respondents did not just report
interactions among themselves, they also reported their connections across all other
actors in the original roster list containing 138 actors. This allows the network analysis to
capture a larger set of connections. In fact, some network studies revealed that for highly
competent informants, a whole network can indeed be measured by a small number of
respondents (Marsden 2005).
Applying this to the current research context, examining the complete network
actor list will provide a comprehensive picture of depicting the changes of outgoing and
incoming ties. For example, at one point of time, one could observe that actor A reached
out to certain number of remaining others in the network, and over the subsequent times,
the out-going ties for this actor stayed the same in terms of sum of ties. If this is observed

143
in a network composed only of those actors at the core rather than retrieving the complete
list, one can easily come to a conclusion that actor A has behaved consistently throughout
all of the three time periods. But a closer look at the original roster data can reveal that
actor A actually developed more ties over the long term and some of the ties were not
being counted for because they were cut out from the core list of actors. Therefore, in the
current section of the study, where the nature of the network formation is one key aspect
of exploration, I maintained the use of the original 138 actors.
I start out by looking at the “big picture”, which is how actors were connected in
general. The type of structural environment that I am looking at first is the
“communication” network. I defined this context in terms of general information sharing
and communication exchange as the result of proactive coping behavior on the actors’
side. Such a willingness to get engaged with others in the network implies that the
network ties are of a “directed” nature. This means that actors have the opportunity to be
both initiators of reaching out to others and receivers of being contacted by others and
being perceived as possible credible sources of information. The former mechanism is
measured by “out-degree” in network descriptive measures, calculating the sum of
connections from the actor of interest to others. The latter is measured by “in-degree”,
calculating the sum of ties that others have reached to the particular actor of interest. In

144
the following discussions, I illustrate both out-degree and in-degree measures for
communication networks. For those actors with comparatively higher level of degree
measures for each time point, I further expand the examination to trace the motivations
and origins of their actions.
The examination of the communication network is critical at this point because
information exchange after a catastrophic disaster often represents primary actions taken
in response to the crisis event and building a robust information infrastructure contributes
significantly to the “resilience of communities exposed to recurring risk” (Comfort and
Haase 2006, 328). Deciphering its patterns of changes can be informative in tracing the
emergence of the more institutionalized project collaboration relationships.

Pre-Earthquake Actions
Table 4.8 shows the summary of results illustrating three categories of the actors
with relatively higher ranking of their outreach activities. I adopted a cut-off point
restricting to those with outgoing ties greater than 10. The intensities for tie-initiating
action are shown inside each of the parenthesis in a corresponding order with each one of
the actors listed in the “Civil Society Actor” column.

145
Table 4.8. Ranking of Initiation of Communication Agency Action Intensity (Pre-
earthquake)
Civil Society Actor Tie-Initiating Action Intensity
Category 1 #51 (36)
Category 2 #61, #6, #137, #115 (19), (18), (18), (18)
Category 3 #100, #119, #20,
#70
(16), (13), (11), (11)

The Case of Actor #51 (NGO51-01)
In the first category, the findings showed that before the earthquake event, actor
#51 had the highest action intensity in terms of outreach activities as it sent ties to 36 out
of 137 possible others. After normalizing this information by expressing it as a proportion
of the number of those in the connected network, it is apparent that this actor reached out
to 26.3% of all the remaining actors in the communication network. In fact, this actor was
the only one among those listed in the three categories that reached out to over 25% of
the remaining others. All other actors in the second and third categories in table 4.8
remained at a level lower than 15% during this period before the earthquake. Actor #51’s
high degree of connection in reaching out to others can be traced back to the types of
activities it had been involved in as an “incubator” for philanthropy-oriented start-ups and
smaller social groups and organizations in China. As was first established in Shanghai in

146
2006 with the goal of supporting social innovation in the non-profit sector in China, the
organization developed into a “support aggregate” that assists the growth of various
grass-root groups in the country. The kind of supports can be in terms of skills and
training for capacity-building, provision of space and equipment for operation,
microfinances as subsidy, and thus help in their institutional registration process. The
organization itself was first registered as a NGO development support center specifically
focusing on developing more domestic non-profits whose visions are driven by the need
for innovative problem-solving techniques for society’s emerging problems and needs.
With this understanding of the actor #51 in mind, it makes sense that this actor tended to
initiate more connections compared to others even before the 2008 earthquake. In other
words, this actor can be thought of “influential” in the sense that it not only initiated
communication ties to others with the intention of information exchange, most
importantly, the nature of such ties originated from the actor’s intention in purposefully
developing and sustaining the functioning capacity of others in the network. Continuing
looking at the emergency response period, the actor’s communication connections
increased to 58 ties which counted 42.3% of the remaining actors. During the recovery
period, this figure further climbed up to 62 ties representing 45.2% of the network, thus
showing an increasing trend in the actor’s outreach activity. Indeed, actor #51 (NGO51)

147
established its field offices across different regions of the country and the one in Sichuan
was another expansion initiative for the actor to engage in the disaster recovery process
after the earthquake. Therefore, the sources of the expansion activities as an organization
became important to trace qualitatively. I paid specific attention to the interview accounts
of the head program officer (NGO51-01) working for the organization’s Chengdu field
office.

Sources of ORGANIZATIONAL action and motivation (Actor #51)
The actions being conducted by actor NGO51 was not completely new based on
its practicing history in China. Before the earthquake, it had field offices across Beijing,
Shanghai, and Shenzhen. The Chengdu field office in Sichuan Province first emerged
primarily due to the 2008 earthquake event. And according to the program officer
(NGO51-01), the Chengdu branch was already a registered formal nonprofit organization
at the time of the interview. With its specific practicing focus in being an “incubator” of
small or newly emerged civil society groups and organizations, the actor engaged in
projects that particularly aimed to develop civil society awareness and engagement
during the early phases of disaster recovery until year 2009. Those social groups and

148
organizations that were in their emerging or start-up phase of development were being
paid special attention. Since 2009, there was a gradual shift in the actor’s recovery-
related activities towards providing local community services, while in collaboration with
actors not only from the civil society domain but also with those inside the state and
market domains.
From the account of the program officer, sixteen earthquake-impacted local
communities were selected to establish projects for community service provision. Three
determining factors were in play for this actor to make the selections. One is that the
majority of the sites chosen were located in the areas that were significantly damaged by
the earthquake. The second factor was to see if there were sufficient local resources to
implement the projects, especially if there were enough physical spaces available. “At
times, we had to negotiate with the local governments to see if they would be willing to
approve us with a site for us to implement our community service programs” (NGO51-
01). The third selection criterion was based on the local circumstances of the re-
settlement areas, such as the differences in population coverage across locations
19
.

19
For further detail in Chinese, refer to Appendix 4.CaseNGO51.1.

149
Among these sixteen selected locations, the community service centers were
further divided into three groups based on the size of the community centers (three large,
three medium, and ten small ones). The operating mechanisms for each category were as
follows. First, for the ten smaller centers, the services were being implemented by its
collaborating civil society partners, called “third party social organizations” ( 第三方的社
会组织). The role of actor NGO51 was to provide some funding and technical supports.
Or it would step in to provide necessary assistance only when problems or difficulties
were encountered during the program implementation stage. As for the medium sized
centers, while the third-party actors were also being introduced, more assistance would be
provided, as compared to the functioning of the smaller centers, in the form of assigning
extra staff to help carrying out the field activities.
Aside from the action side of the picture, the concerns to better serving the needs
of the disaster-hit local communities became the primary motivation for actor #51 to
participate in these different projects. As described clearly by the organization’s program
officer:
This would depend on the number of people we are servicing and the available
space. So if the servicing space turns out to be very big, like one to two thousand
square meters, then, the services being provided would consist of more varieties.
The activity areas covered by only one or two civil society groups/organizations
would be limited, especially when the needs in the disaster areas varies a lot.

150
These one or two groups/organizations would not provide all of the needed
services in the area. In these cases, we need to coordinate many other
groups/organizations to come to a specific location to conduct activities. Our
ability to coordinate and facilitate would be critical at these circumstances. So
what we need is to play the role to build some kind of platform. (NGO51-01-01
20
)
Note first that the informant recognized the diverse social needs within local communities
during the recovery stage of disaster. This represents the cultural sensitivity being
implemented through the actor’s field practices. Second, there was also an awareness of
the importance in building connections with other civil society actors for sustaining a
collaborative effort in providing the needed services for communities. Third, the role
being perceived by the actor itself was essentially a relational one in terms of performing
its task as a platform for facilitation and coordination.
In order to illustrate how a particular project was being carried out in the field, the
program officer provided an example of a project implemented in one of the areas
suffered significant damages after the earthquake. Upon carrying out the project at that
location, three social service organizations ( 社会组织 ) were introduced into the
community. During the “early intervention” phase, typically some preliminary research
would be conducted and information be gathered regarding the needs of the community.
Then, based on the different kinds of needs, actor NGO51 will select those social

20
For Chinese script, Please see Appendix 4.4.01A.

151
organizations with matching specializations. Thirdly, the written details of how to carry
out the activities will be implemented through a collaborating effort with locally
established partnering organizations. “This is because they are more familiar and
knowledgeable with the local conditions and environment than we do and thus can
operate in an effective way that is culturally sensitive”.
In general, the actions of NGO51 were consistent with its primary purposes in
assisting the growth of other civil society actors both before and after the earthquake. But
the emergence of many smaller grassroots actors provided valuable opportunities for
NGO51 to expand its communication connections toward supporting more of its civil
society counterparts. This explained why the actor was able to establish an increasing
number of communication ties throughout the emergency and recovery periods after the
earthquake. Taking the initiative to become an “incubator” to develop other civil society
actors that were at the start up stage became the primary source of motivation for NGO51
soon after the earthquake. The persistence of organizational action beyond the emergency
response stage clearly arose out of the actor’s awareness of the co-dependent and co-
evolutionary nature of its relationship with others (see figs. 4.8.1A, 4.8.1B). This thus led
to the actor initiating field collaboration projects specifically drawing in others with
significant amount of local knowledge.

152

Figure 4.8.1A. Banner Placed Outside Actor #51 Chengdu Field Office Showing
Civil Society Groups and Organizations in its “Incubation” Program
(Source: photo taken by Jia Lu, 2011)

153

Figure 4.8.1B. Banner Placed Outside Actor #51 Chengdu Field Office Showing
Civil Society Groups and Organizations Completed its “Incubation” Program in
Shanghai, Beijing, and Chengdu
(Source: photo taken by Jia Lu, 2011)

Continuing examining the results shown in Table 4.8, I now focus on the actors
inside the second category. Actors #61, #6, #137, #115 exemplified relatively similar
behaviors in being the active initiator of communication ties as compared to actor #51.
Before the earthquake, actors #6, #137 and #115 reached out to 13.1% of all the
remaining others while actor #61 reached out to 13.9% of others in the network.

154
Actor #137 is a Beijing-based non-profit organization registered under the
business category and it was first established in April, 1994. The primary focus of
practice established by this organization has been in the area of environmental protection
and community problem-solving related to the sustainability challenge as a result of
China’s rapid urbanization process. Its primary goal was to construct a platform for
citizen participation regarding environmental protection issues and motivating voluntary
behavioral change. Immediately after the earthquake, the communication ties for this
actor almost doubled, climbing up to 37 (27%), indicating increasing emergency response
effort through the establishment of information sharing ties with others. According to the
organization’s online memorandum, in July 2008, just two months after the earthquake,
the actor initiated a seminar event inviting participants from the planning, construction,
and education departments of the central government and local government, foundations,
and other donors to discuss the issue of incorporating green technologies and concepts
into the construction of schools in the impacted areas. Such an effort had been carried out
through the long term recovery phase when the actor reported participation in areas of
housing reconstruction and environmental protection. However, there was a decrease in
the actor’s initiation in reaching out to the others from the emergency response period to
the long term recovery period. A close observation of the particular connection changes

155
revealed that all of the 26 (19%) communication ties maintained in recovery period were
part of the actor’s connection network in the emergency response period. What this
means is that for this actor #137, its efforts in initiating ties were concentrated during the
period immediately after the earthquake. And it was able to maintain the already
established information exchange channels through the longer term. No new ties were
being built into the recovery period. This could be the case that the actor’s focus over the
long term was in maintaining certain types of relationship that tend to enhance its ability
to conduct works in combining housing reconstruction and green technologies in
particular.
Actor #6 is a non-registered international NGO with areas of expertise in
providing educational opportunities for the disadvantaged and vulnerable individuals
across various developing countries. Right after the earthquake, the organization directly
participated in the emergency response activities through the establishment of an
educational system support project in the Chengdu region. From the organization’s
annual report in year 2008, it is apparent that the activities of the project included setting
up temporary classes in tents for the continuity of the affected children’s education. At
the same time, the organization has been active in reaching out to others and the number
of actors that it named as communication partners increased from 18 (13.1%) of pre-

156
earthquake period to 45 (32.8%) during the emergency recovery period. And during the
long-term recovery period, such ties reached to 58, resulting in its communication ties
with 42.3% of the remaining actors in the network.
Actor #61 originally entered China as an international NGO focusing on poverty
reduction and community development initiatives in rural areas of the country (see
figures 4.8.2A and 4.8.2B).

157

Figure 4.8.2A. The birds-eye view of the newly reconstructed post-earthquake
housing of YP village (located in the mountain regions of Sichuan Province), the
primary location where one of the rural recovery programs managed by Actor #61
was implemented.
(Source: photo taken by Jia Lu, 2011)

158

Figure 4.8.2B. Another birds-eye view of the YP village
(Source: photo taken by Jia Lu, 2011)
Before the earthquake, its connections extended to 19 other actors. This was 13.9%
of the remaining actors in the network, which ranked the top of all others in this category.
During the emergency response period, the organization actively participated in the
immediate relief by gathering both material and financial support for some of the most
significantly earthquake-damaged provinces. In November 2008, the organization was
successfully registered under the ministry of civil affairs in Sichuan province and was
able to continue its rural development and poverty relief initiatives in the earthquake-

159
impacted areas. A seven-year special disaster recovery project was also established with
the particular long term goals for community sustainable development and environmental
protection. During the emergency response period, its out-going ties increased from 19 to
37. This was an increase from reaching out to 13.9% of the remaining actors in the
network to 27%. However, like actor #137, such extensiveness decreased to 26 (19%)
during the recovery stage. Again, the stabilization of the 26 ties into this last period did
not come from this actor’s effort in building new ties in the network. Rather, it originated
from the initial surge in communication tie construction immediately after the earthquake.
Actor #115 is also based in Sichuan Province and it is an association that
specializes in community elderly care and nursing. It started out with extending its
communication ties to 18 others, accounting for 13.1% of the remaining actors in the
network. But after the earthquake, its breadth of relationships began to wane, sending a
decreasing number of ties to 6.6% of the remaining actors during the emergency response
period and 2.9% during the recovery period. Compared to the relationship patterns of
actors #137 and #61 mentioned earlier, this organization is the only one that experienced
consistent decline in the number of communication partners including the period
immediately after the earthquake. However, a closer examination of the particular ties
being built and terminated over time shows that #115 actually did actively established

160
two new connections with groups that were non-existent before the earthquake event.
One is the tie with actor #3 and the other is actor # 49, both of which came into being
only after the disaster and became central players in the network over time, as will be
shown in the earlier analysis. Actor #115 also built a new communication connection to
private enterprises during the recovery period. Therefore, in the case of looking at actors
with decreasing number of connections, an observation of the number of tie pattern
changes longitudinally will only provide a general picture of the extensiveness of
connections, but not considering the in-depth quality of the ties. By quality, I mean one
needs to take into account to whom this focal actor is connected and the characteristics of
these other actors.
In the third category, actors #100, #119, #20, and #70 have slightly lower level of
out-degree as compared to those in the second category. Both #100 and #119 reached to
consistently increasing number of other actors in the network since the earthquake. Actor
#100 reached out to 16 others before the earthquake, which accounts to 11.7% of the
remaining actors in the network. This number increased to 13.1% and 14.6% respectively
for the emergency response period and the recovery period. Actor #119 started out by
making connections to 13 others which accounts towards 9.5% of remaining others. The
intensity of such activities increased to 13.1% and 17.5% during the consecutive periods

161
after the earthquake. Actor #100 is a Sichuan-based NGO focusing on participatory-based
community development and poverty reduction. First established in 2003 and registered
in the business category, the organization participated in activities such as taking care of
women and children and livelihood development for local communities. It was also
involved in community infrastructure reconstruction and community funds during the
earthquake recovery period.
Actor #119, however, is a non-registered Hong Kong-based NGO. It was first
established in 2004 and has been developing a variety of community partnership projects
with mainland social groups and NGOs over the years. After the earthquake, the
organization further expanded its works in the areas of community arts and social
recovery. Throughout the three time periods, the communication partners of #119 has
increased from 13 ties before the disaster event, to 18 in the emergency response period,
and to 24 during the recovery stage.
Actor #70 is a Beijing-based rural construction center registered in 2004. The
organization is a sub-division of the Rural Construction Center established at a university
in Beijing. It focuses its work in policy implementation related to rural development. It
had connections with 13 communication partners before the earthquake, which was

162
comparable to the level of actor #119 in the same period. But rather than continuing to
expand its communication ties over time like the case of #119, actor #70 did not sustain
any of the ties established before the earthquake and during the emergency response
period. Right after the earthquake, the organization only had 4 ties remaining in its
network. While it could be interpreted as a decreasing activity effort in building up
connections at the aggregate level, a closer look at the nature of these 4 ties revealed that
they were all newly established communication channels with those actors that had not
have relationships with it before the earthquake. The disaster event triggered a type of
relationship-building agency action for this actor to initiate contacts with the government
branches, and other central actors in the Sichuan Province. However, none of these new
ties made it over time into the recovery stage. One possible explanation could be due to
the geographic location of the organization as being far from Sichuan Province. The other
reason could be due to its purpose of establishment in participating solely for the
emergency response period without the intention to stay active into the long term locally.
Actor #20 is another Beijing-based registered NGO first established in 2002. As
an organization engaged in community service and capacity building in urban
communities, the actor has been dedicated to provide tools to engage citizens to
participate in the governance processes for community problem-solving. The majority of

163
its activities are currently located in Beijing and none in Sichuan Province. Before the
earthquake, the actor had communication connections with 11 other actors in the network
and some of which included actor #51 and #137. However, none of these ties persisted
right after the earthquake. During the emergency response period, the only partner that it
reached out to was actor #49, a Sichuan-based NGO that was formed and established
initially for the purpose of earthquake response and recovery. This particular relationship
was maintained through the longer term recovery period. For this actor #20, the
earthquake event seemed to be a turning point for opening up an increasing number of
communication partners in the Sichuan Province over the long run. This can be seen from
its ties expanding from only one with actor #49 during the emergency response period to
a total of five during the recovery stage. And all of these other five partners were either
Sichuan-based grass-roots NGOs or formal earthquake-recovery oriented NGOs with
field offices established in Sichuan. When one only looks at the changes of the number of
ties, it is easy to come to a conclusion that actor #20 was not able to build consistent
connections over time. But a closer examination of the characteristics of this actor tells us
a different story. First of all, it is a non-Sichuan based organization. For such an entity to
participate in the response and recovery of the earthquake that happened in Sichuan, it
became necessary to get out of its “comfort zone” location-wise and start establishing

164
brand new ties with actors whose activities were grounded in the local culture and were
able to gather first-hand information on a timely basis. It might be a rather difficult task
for actor #20 initially especially when it did not have a field office that could have
facilitated the communication flow from its partners in Sichuan to those working at the
agency headquarter in Beijing. This location factor might partially explain the fact that it
took the organization all the way into the recovery stage in order to establish
communication connections with the local Sichuan-based NGOs. Secondly, the
earthquake event, however, did prompt this organization to actively build up, particularly
to those operating at a different location. The relationships that formed at this later stage
of recovery period could be the foundation of a platform for a multi-regional
communication network for the longer term stretching further beyond the 2 to 5 years of
the recovery stage examined in this study.
The first conclusion that can be reached by examining the nature of these actors
with higher level of outreach activities before the earthquake is the fact that they are all
formal institutions that had established themselves and practiced as non-profit
organizations in the civil society domain long before the 2008 earthquake event.
Secondly, actors #3 and #49, both established only after the disaster, were found to be
critical communication partners to connect. Their characteristics will be further discussed

165
in the later section. As a third conclusion for this section of examining the out-degree
activities for the period before the earthquake, I summarize the characteristics of the key
actors in table 4.9 below.
Table 4.9. Traits of Key Actors with High Tie-initiation Action (Pre-Earthquake)

Attributes
Actor
Registered Establishment
(Sichuan Branch)
Location
#51 Yes After the EQ Chengdu
#61 Yes After the EQ Chengdu
#6 No None Non-Sichuan
based
#137 Yes None Beijing
#115 Yes After EQ Chengdu
#100 Yes Before EQ Chengdu
#119 No Before EQ Chengdu
#20 Yes None Beijing
#70 Yes None Beijing
As we can see, there were two un-registered foreign entities—#6 and #119—but were
able to establish relatively high level of communication relationships before the
earthquake. Both of them were formally established international NGOs originally
operating outside mainland China and have started their work in the country before the
2008 earthquake. Since the 2008 earthquake, both organizations were able to sustain and

166
expand their communication networks, particularly with domestic grass-roots groups and
organizations. Similarly, domestic grass-roots organizations, such as actor #51 and actor
#100, that were able to maintain such a trend were all formally registered NGOs with
established field offices in the city of Chengdu, Sichuan Province. Among those that
showed a downward trend in tie-building activities over time, with the exception of actor
#115, the rest were all non-Sichuan based organizations that maintained their operation in
Beijing both before and after the earthquake event. This shows that a certain degree of
familiarity with the local culture, or in other words, the depth of local knowledge from
practicing in the field, where the focal organization seeks connections to implement its
work could be an important factor in sustaining information sharing ties throughout the
time span depicted in this study. However, tie-creation trend for actor #20 showed that it
was also possible for the non-Sichuan based organizations to start generating more ties
over the longer term.

Post-Earthquake Actions (Emergence Response)
One characteristic to remember about the 138 actors at the first stage discussed
above is that none of these actors existed before the earthquake. The actions of these

167
actors were triggered by the disaster event and were able to dedicate themselves solely to
recovery activities. This is the reason why we observe a large number of “isolates” in the
network graph figure 4.3A. In this section, I will examine the ways these newly
established members were being integrated into the network by further investigate the
out-degree measure.
I have already discussed the functioning nature of those actors that were not only
in existence in the first time period but also maintained a relatively high level of
communication ties for the following two periods. Here, I will pay particular attention to
the actors that emerged as key players only after the earthquake and examine the
persistence of their communication tie-building actions. In table 4.10, I used five ranking
categories to depict the top levels of intensity for actors’ communication outreach
activities. The cut-off level of intensity is set at above 20 outgoing ties.

168
Table 4.10. Ranking of Initiation of Agency Action Intensity (Emergency Response)
Civil Society Actor Identifier Tie-initiating Action Intensity
Category 1 #24 (136)
Category 2 #3, #32 (76), (59)
Category 3 #135, #137 (38), (30)
Category 4 #123, #4 (29), (27)
Category 5 #7, #109 (24), (21)
I distinguished actor #24 in the first category because of its increasing number of
connections that stood out from the measures of agency action among the rest of the
actors. During the emergency response period, it reached out to 136 actors in the network
which accounted for 99.3% of the remaining ones. What this means is that except one
actor in the network, #24 reached out to everyone for communication for this time period.
This is worth noting because no other actor in the network reached out to this level
activity that is comparable to actor #24. It signals an invitation to look further into the
characteristics of this particular actor in search for possible explanations of this type of
information exchange behaviors.
Actor #24 is a registered non-profit organization originally established in 2004.
The focus of its activities has been mainly towards collaborating with urban communities
for communicating information regarding sustainable ways of living and developing

169
projects that are not only grounded in the local culture but also assist the general public to
better understand the concept of environmental sustainability. Looking back at the period
before the earthquake, the actor actually had communication ties to 4 other actors in the
network. One was directed towards the government and the other towards the private
businesses. The two other connections were established with non-profit organizations that
were in the similar field of environmental protection as the actor itself. According to the
online archival record of its activities right after the earthquake, the actor not only
participated in a set of coordinated response activities among many of the domestic non-
profits at the time but also was responsible particularly for the coordination of
information exchange, resource allocation, and volunteer organization. Such a role during
the emergency response period could be the determining factor contributing to the
organization having established connections with 136 other actors in the network.
However, going on observing its level of connection towards the long term recovery
stage, the percentage of ties dropped more than 50% from reaching out to 99.3% of the
remaining actors to 40.1% of others in the network. In other words, 59.2% of the
communication relationships that it established during the emergency response period
had not been carried into the recovery stage.

170
Here, the case of actor #24 sets a context for investigating the meaning of
“capability” from a structural relation perspective. To say an actor is “capable” of doing
something, one often has to assume that there are factors that are inherent to the
functioning or the characteristics of the actor itself that pre-determines whether a goal can
be achieved or not. But when “functioning” is defined by the ability of an actor to
establish relationships with others, be it communication or collaboration relationships, the
capability lens becomes a way of seeing and thinking by taking into account the co-
evolution of “action” and “structure” side of the picture. In other words, if “capability”
can be defined in terms of the available choice set that an actor can actually act upon by
having the opportunity to choose from all the possible ways of functioning, then, the
focus is not to be on those who already have achieved functionings (Sen, 1999) but on a
set of possibilities that are available for the actor to attain when he or she actually enacted
such initiation. Take the situation of actor #24 as an example, the sheer number of its
communication connections decreased over time. But by no means is it interpreted as the
actor being in-capable of achieving the various ways of functionings. A closer look at the
specific types of connections of this actor revealed that not only all four ties that it
established before the earthquake were sustained into the recovery stage but also they
were accompanied by additional relationships from which the actor actively sought out

171
during the long term. Along the way, the changes of the network structure itself through
the emergence of new actors and formation of new ties right after the earthquake also
contributed to an increased opportunities for actor #24, thereby expanding the ways that it
could choose to build relationships. As a result, throughout this study, any changes in the
structure of the network environment are interpreted as a possible opportunity for a
particular actor to perceive the circumstance as one that it can act upon. This approach is
inherently different from perceiving the structure of a given social environment as one
that is imposed and fixed over time, or merely existing to put constraints upon the focal
actor. In fact, in the context of the conceptual framework of this study, actors themselves
are seen as change agents in forming their own social environment by making decisions
about whom and how to connect for relationship-building.
This way, the term “capability” in this study will be understood as a way of
looking at how actors activate their own “agency” based on the perceived opportunities
may be provided by the structure that they are embedded in rather than focusing solely on
what they have already achieved. One functioning characteristic of actor #24 during the
recovery stage is that it chose to focus in the area of environmental protection on top of
all other types of activities. This choice of conducting work within the boundaries of the
actor’s original field specialty can help explain the decreasing number of its immediate

172
network ties. By “immediate network ties”, I mean the number of others with whom this
actor has direct relationships with. In other words, “friends of friends” does not count in
this incidence.
So far, the case of the agency actions of #24 pointed to several important natures
of structural changes in the communication network environment. First, the action
response efforts among all the civil society actors shortly after the earthquake opened up
a specific opportunity for actor #24 to build up and expand its own network connections.
Secondly, the overall information exchange structure persisted into the recovery stage.
This can be inferred from the case of actor #24. When compared to connecting to 2.9% of
the remaining actors in the network before the earthquake, the actor was able to sustain
significantly higher level of relationships up to 40.1% all the way into the recovery stage.

The Case of Actor #24 (NGO24)
In the following text, I provide a qualitative in-depth examination of actor #24 as
a result of its significant outreach activity position in the communication network. It is
critical to further understand the driving forces of the agency changes from the actor’s
point of view. The processes of making the decision to take initial action, sustaining the

173
action, as well as its own development experiences in relation to other civil society actors
provide valuable evidences in tracing back the sources of the initial network formation,
thus the structural environment of an emerging civil society.
Initially, by the interview account of the director of NGO actor #24, the decision
was made to focus most of its actions by paying more attention to long term disaster
recovery rather than emergency and short-term response. “We were hoping to base and
develop our work connections mainly towards groups and organizations who aimed to
practice locally, and we have not positioned our work in such a way to concentrate solely
on the emergency response stage”, explained the director. With such a long-term
definition of its agency action after the disaster event, there was a strong tendency for the
actor to choose selectively in building communication ties with Sichuan-based grass-root
groups and NGOs, particularly during the long term recovery stage. In terms of the
collaboration tie formation, first of all, one of the primary motives for the actor to
develop a connection over time was the other actor’s similar practicing background in
environmental protection. Secondly, another source of motivation to establish long term
collaborative partners can be related to the location factor. According to the account of
the director (NGO24-01), the reason that some potential collaboration projects not being
realized was due to the lack of geographic proximity with the field practicing site of the

174
potential partner. As one of the main actions taken by this NGO during the recovery
times was in assisting the disaster-hit rural regions constructing an “environmental
friendly toilet system” by building eco-friendly toilets for each individual rural household,
the need to establish a long term relationship with the local communities was particularly
important.
One of the action characteristics of project collaboration network formation
among civil society actors, based on the account of the director, was in a pattern of
“regional clustering” with those practicing in the same regions tended to establish
collaborative relationships together. Along with this trait, he made further comment on an
ideal type of collaboration network environment for disaster mitigation and preparedness:
Looking at the current situation, many of the collaborations are characterized by
regional divisions. I personally think there is not enough work-related
collaboration among us. In fact, aside from the aspect of disaster recovery, the
NGO network in Shenzhen worked out well. They had a tightly-knit work
relationship among each other. The institutional environment provided by the
government was nurturing. In order to do well in disaster mitigation and
preparedness, it is very important to do professional works, passion and warm-
heartedness will need to be supported by a well-functioning emergency
preparedness mechanism. (NGO24-01-01
21
)
Apparently, the sources of the actor’s concentrated agency action in communication
relationship-building came immediately after the earthquake, from two kinds of

21
Original Chinese script in Appendix 4.4.01.

175
determinations. One was the thinking of activities beyond emergency response and
disaster recovery into the long term social development of China. The other one was its
practice focus paying particular attention to establish connections locally with others in
the field of environmental protection.
Examination of the remaining actors in table 4.10 shows that actor #3 and actor
#32 had comparatively higher level of post-earthquake communication outreach activities.
For actor #3, it was formed only a few days after the earthquake as a non-registered
social group originally defining itself to perform the function similar to that of a
communication platform coordinating the response actions among grass-roots non-profit
organizations. The actor’s communication initiatives in reaching out to others expanded
to 55.5% of the remaining network, which accounted for 76 communication connections
at the emergency response period. Contrary to actor #24 whose reaching out activities
encountered a sharp decrease over the long term, the information sharing connections for
actor #3 climbed up to 98.5% of the remaining network during this same period. This
accounts to 135 actors in the network.
Looking at the changes in its immediate structural environment for actor #3 over
time, one can generate some preliminary conclusions regarding the changes inside the

176
civil society domain. One is that the earthquake did trigger the emergence of this group
when ordinary local citizens voluntarily took actions to respond to the needs of the
response efforts immediately after the disaster. While the case of actor #24 demonstrated
the activation and persistence of the communication structure after the earthquake, the
experience of actor #3 also revealed the emergence and continuity of the actor’s
institutional structure internally generated. The other lesson is that the continuity of this
group’s activities in the field over the long term suggested that possible inherent drives
and motives can be identified to explain the sustainability of its functioning. In this
respect, I present the following qualitative examination to trace its motivational origin of
action.

The Case of Social Group Actor #3 (SG3)
Actor #3 first came into being as an established social group only 3 days after the
5.12 Wenchuan Earthquake. All of the three interviewees in SG3 directly participated in
the formation and the continued functioning of the group’s action since its establishment.
When trying to discover the motivational sources of their individual actions in terms of
forming the group in response to the disaster, it became clear that their spirit of devotion

177
to voluntary activities self-organization actions can actually be traced back all the way
before the earthquake event.
Recently graduated from college and having worked for a local newspaper, the
younger participant (SG3-01) was first introduced and drawn into the field of voluntary
associations as she joined the local “youth voluntary program” and started to become
active in different projects with her friends. This type of pre-earthquake action not only
expanded her personal network connections but also brought her a sense of awareness of
what the field of voluntarism involves and is all about.
I first came to know the field of NGOs ( 公益组织)and participated in their
activities because of joining the youth volunteer program. When they (SG3-03)
were launching the third phase of the program, one of my friends participated in it
as a volunteer, and she would invite me to be involved in some of their activities,
or joined them as a volunteer as well. Then, I started to get to know many people.
In fact, I rarely knew about the “circle” of NGOs before. But when I joined the
hospital volunteer team just shortly after the earthquake, one of the project
managers of the youth program called me and said that they needed some extra
help. That’s how I came here to participate. It’s around the 16
th
of May, when a
huge number of the wounded were transported here. I was taking care of a little
girl for three days back then. (SG3-01-01
22
)
With this prior familiarity in the field and a set of ready-built personal networks, which
included close connection with one of the senior participants in this emergent group, she
joined its emergency response volunteering activities immediately after the earthquake

22
Original Chinese script in Appendix 4.4.02.

178
event. Although her original role was not one that was primarily responsible in officially
organizing the group together, she became the first volunteer to work on information-
sharing-related activities when the group itself started to take form.
The primary information-sharing activities took the form of Internet blogs and
postings through the mediums of various online discussion forums
23
. With further
voluntary participation of the local Chengdu citizens in disaster response, the informal
connections between the group as an entity and other organizations expanded over time.
Familiarity to the local culture, including people and the works related to building
communication lines among those in need became important factors for her to make the
decision to continue being devoted to the field over time.
When bringing about the question of motivations in participating and committing
to the field activities of the group in the first place, the participant constantly referred to
her devotion towards broader background-oriented concepts such as “circle” and “field”.
The word in Chinese is often used as a slang, and can be interpreted as a large group of
people who share common interests and being strongly devoted to a particular field of
activity, so that those involved started to call themselves “insiders”. Obviously, the young

23
See Appendix 4.4.03 for details (SG3-01-02).

179
participant perceived her involvement in the group not just as a type of ordinary work she
was doing but as an act in personally engaging in the larger field of voluntarism. There is
a sense of emotion being involved in terms of her devotion towards a larger cause for the
society. And the constant identification of herself practicing in the “circle” demonstrated
her willingness to be defined as one of those “insiders”. This can be a primary indicator
of the individual’s capability to recognize her belonging and identification in terms of a
larger sphere above the work activities themselves.
Because when I just graduated from college by then, I still wasn’t so sure about
what to do with my life. Plus, from my experiences in working for the media, I
felt quite disappointed. (When I participated in NGO activities) I felt that was a
field that I had never encountered before and what went on in this circle…for one,
very intriguing and the people working in this area were all very interesting…how
to say it…it’s like each person has his or her own distinct individuality, not like as
if they were all of the same type. When I first came to know teacher YX from
actor #73…because I had only known his name through newspapers and news
reports on TV, and when suddenly seeing the real person, I immediately felt the
difference when doing things with them (his team). In fact, (I) haven’t thought
much about how long I will continue doing this back then, neither had I planned
for the long term, but at least at the current stage, I know that I am strongly
interested in this thing, and have a lot of energy whenever I am doing it, so that’s
why I stayed and kept working in this field. (SG3-01-03
24
)
As participant SG3-01 further identified her action with those practicing in the
“circle”, her own participation raised a level of self-awareness further empowering her to

24
For original Chinese script please refer to Appendix 4.4.04.

180
make the decision for long-term commitment. In this case, the motivation came from the
work allowing her being able to identify a type of shared interest and endeavors with
others while preserving her own individuality and interest development.
Compared to the young participant’s view of motivation, one of the senior organizers of
the group offered another perspective on the factors that promoted actions and
motivations towards voluntary activities after the earthquake.
…as for us, as long as we can still make a living, and for people at our age, I don’t
want to do businesses, neither do I want to make a fortune of wealth. I’d rather
prefer doing something that can promote the betterment of the society, that’s it.
No, nothing dramatic, don’t think of it as something that special… (SG3-02-01)
…our society was supposed to be a sound and healthy one, with everyone
supposed to possess something that is in accordance with…what can be called…a
moral bottom line, yes, it is a set of bottom line morals and ethics. If I have
money, but I don’t take out some for donation, then, I will have a moral deficit,
right? This is what’s like in a healthy society, and such is the most basic moral
bottom line, a code of conduct. Nothing so significant, no. It’s just because our
society has fallen too much below (that basic line), people’s behaviors became too
lowly, too lowly, and when you compare the two states of existence, the gap
maybe too big. Nothing more than that. (SG3-02-02
25
)
For this senior participant, one of the primary drivers for his action can be referred to as
one for the betterment of the society in general. There is a sense of self-giving-ness that
guided him for voluntary activities. As we can see, both the younger and this senior

25
For original Chinese script please refer to Appendix 4.4.05.

181
participant interpreted their action and motivations within a larger field of practices rather
than perceiving their work as merely part of the group functioning itself. To them, such
involvement brought further meaning and purpose beyond their immediate duties at hand.
For the senior participant in particular, he was able to recognize that this larger purpose
incorporated a desire to give back to the society and making personal contributions to the
social development of China. According to both of the participants’ reflections on their
actions and motivations, the action responses to the disaster no longer can be looked at as
merely a passive reaction simply trying to deal with the crisis situation. The event
became a channel through which the participants in SG3 becoming aware of the “things”
that they hold most valuable in life and being empowered to strive towards them
alongside with others who share similar aspirations. And in the case of these participants,
the valuable “things” are the motivations represented by serving and helping others not
just in times of crisis but also their inner desire for the good of the society in general.
In summary, the initiation of voluntary action and self-organization for actor #3
was motivated by a joint awareness of the meaning of agency action larger than merely
for the purpose of disaster response and recovery. The individual participation was
interpreted as a sign of self-giving-ness towards working for the betterment of the

182
Chinese society in general. There was a collective recognition of being empowered
through identifying their practices as part of the civil society action domain.
Following actor #3, actor #32 had the second highest level of out-degree in the
second category. Before the earthquake, the actor had only one connection which
accounted for 0.7% in the network. It quickly extended its ties to 43.1% of the remaining
network right after the earthquake. And not only did all the connections established
during the emergency response period continue through the recovery stage, but also the
actor reached out to 57.7% of the remaining network during this period. This accounts for
an increase of 20 more communication connections in the long-term recovery period.
Taking a closer look at the actor’s own attributes, at the time of 2011, it was a non-
registered social work association based in Chengdu, Sichuan Province. The association
was first established in 2010, after the earthquake event. However, #32 still named one
other actor in the network as a communication partner before the earthquake. This means
that there might be a blurred conceptual boundary in understanding relationship
formation from the actor’s perspective. In this context, the actor did not perceive that the
formation of a tie involves the group to be formally established as reflected in registration
status. In other words, the date when the actor started perceiving itself as a civil society
group/organization could be different from the date when it first gained registration status.

183
This can be illustrated by the fact that the expansion of its network connections started
before it became formally established. Since this study defined the emergency respond
period is within 7 days to 1 year of time after the earthquake in 2008, by the time when
the actor claimed to be established in 2010, it had already initiated contacts with 59 actors
in the network. What can be said for this actor in particular is that the earthquake did
trigger an action-oriented initiative or urge to connect with others primarily for the
purpose of short-term disaster relief, regardless of the institutional status at the time. By
further defining its field of work with a focus in providing support for other non-profit
groups and organizations over the recovery stage, the immediate disaster response
network actually enabled the actor to cultivate these relationships and expand its
connections in such a way that shaped its goals in the long term.
Two other newly established civil society actors worth mentioning are #135 and
#123. Actor #135 was non-existent before the earthquake and therefore started out with
zero connections. During the emergency response period, the actor built 38 ties which
accounted for 27.7% of the remaining actors in the network. The scale is further
expanded to 62 (45.3%) ties into the long term recovery stage. Three of the connections
that the actor established were not sustained into this very last period. The actor was able
to maintain all of the remaining communication relationships and initiated 24 more

184
connections from the short term response period to the recovery period. Different from
the other non-profit actors that were formed as one group or organizational entity as we
have discussed in detail earlier, actor # 135 was originally established as a collaborative
in partnership with two universities, one from mainland China and the other from Hong
Kong. By “collaborative”, the word exemplifies actions taken by professors and students
from both academic institutions and who practiced social work related disaster relief and
recovery in the earthquake impacted region. Although this particular entity was not
registered at the time when this study was conducted in 2011, it had been working in the
field as a social work station for almost three years since June 2008. And it was located in
Yingxiu—one of the worst damaged areas and also at the epicenter of the earthquake.
During the recovery stage, the station also became active in works related to women and
children, environmental protection, psychological counseling, and livelihood support at
the local community level. At the center of its social work practice was a value of
promoting macro social policy change through the micro changes in individual, families,
and communities.
Similar to actor #135, actor #123 was established as a social work oriented non-
profit group. It was originally started up by an ordinary Sichuan local who voluntarily
participated in the emergency response activities immediately after the earthquake and

185
then decided to devote his career in the long term recovery by establishing his own non-
profit organization. The actor did not have any communication ties before the earthquake
as it was non-existent at the time. Shortly after the earthquake, it reached out to 21.2% of
the remaining actors in the network that accounts for 29 communication connections.
During the long term recovery period, there was a slight drop in the number of
relationships to 26, counting towards 19% of the remaining actors in the network. As a
non-registered social group, the actor participated in the recovery activities such as
providing support for elderly and disabled, women and children, as well as psychological
counseling. One of the factors that could contribute to its minor decrease in
communication connections was that the actor’s defining areas of works became more
focused and aimed at developing a particular area of expertise over time as compared to
an all-comprehensive response-driven approach initiated at the emergency period. I use
the following qualitative examination to provide a detailed account tracing the source of
individual action that eventually led to the emergence of actor #123 at the
group/organization level.

186
The Case of Actor #123 (SG123)
The second day after the earthquake event, the organizer of actor #123 (SG123-
01), just as many other volunteers, came and gathered in one of the significantly
earthquake impacted cities to make their own contribution to those in need. He initially
opted to conduct disinfection works for the remains of the victims in order to prevent
further epidemics in the area. As was recalled by him, through the process, he witnessed
some of the most unforgettable tragic scenes in his entire life and the images left in his
mind often came back to “haunt” him in his dreams. Through those days working in the
disaster areas, the experience for the organizer was so traumatic that he eventually
recognized that he had to seek professional psychological help in order to recover. This
first-hand experience prompted him to be determined to stay in the disaster area and
establish his own team to assist the psychological and other health-related recovery of
children survived through the disaster. Just as how the organizer put it into his words with
passion, “I have developed a tremendous amount of affection towards the people in the
area where we are conducting our work since then and it has been the primary source of
motivation for me to keep it going until today…”. The main source of individual action,
therefore, came from the participant’s first-hand emergency response experience. This in

187
turn triggered his deep care for the recovery lives of the local people while forming an
emotional attachment to the people and the place impacted by the earthquake.
The level of outreach action of actor #4 immediately followed that of actor #123.
Its communication ties went from zero to 27 others in the network during the short-term
response period. This outreach level went down to only 8 connections over the long term
recovery period. As a non-registered grass-root group established in August 2008, there
was one distinguished property of this actor when compared to all the others in the
previous categories. And it was the diversity of member composition during the period
immediately after the disaster event. Originally initiated by an ordinary driver employed
in a private company in the city of Chengdu, the group was later joined by other migrant
workers, university students, and small-business owners during the emergency response
period and together, the actor participated in the housing-reconstruction activities as well
as providing support for elders and disabled population. In contrast to the participation of
those actors that were more established in the form of international non-profits and
domestic ones with more developed specialties, actor #4 was one of the most grass-root
groups formed after the earthquake. Such a bottom-up quality was similar to the nature of
action initiation of actor #123 discussed earlier. However, the difference was that the
communication actions of actor #4 became significantly less intense over the long term

188
recovery period. It is thus important to closely examine both the motivations for agency
action at the individual and the group level.

The Case of Actor #4 (SG4)
The first time I met the key participant (SG4-01) at the time of my interview was
in an early morning in Chengdu. We arranged to meet at a small coffee place near
Sichuan University. He was a tall and enthusiastic young man in his late twenties. After
we settled ourselves down at the cornered seats inside the coffee place, he started to tell
me the stories of how the group actor #4 came into being and how the experiences after
the earthquake had touched upon every part of his life.
From his account, the group was first established in August, 2008, which was
around three months after the earthquake event. The self-organizing action was initiated
by the founder, then an ordinary working class driver working for a foreign company in
Chengdu. Participant SG4-01 was voted as the second group leader after the founder had
stepped down. From the recall of the participant, the founder started recruiting volunteers
online and the group quickly started to grow as a team. During the emergency response
period, its volunteers participated in activities assisting “demolishing the earthquake-
damaged housing and the reconstruction of new ones”. The team had raised a significant

189
amount of operational donation funds through that time. However, as the official
emergency response period ended, like actor #3, the group itself had to re-consider the
variety of ways of transitioning.
At that critical point of transition, several factors could be traced to understand the
decision for the group to take a particular trajectory in order to remain as an informal
social group. First of all, the participant had a clear conceptual understanding of the
difference between “NGOs” and “social groups”. The former, especially the term
“organization”, was understood to involve “strict division of specialization and high level
of professional works being performed” (SG4-01). The functioning of SG4 was therefore,
only being perceived as a “group” rather than a formal “NGO”. However, what was
unique about the way that the participant saw the “group” status was in his openness in
bringing about a “natural” type of growth rather than forcing the team to develop pre-
maturely into a formal organization structure. For example, as one of the key leaders at
the time of the interview, he had never intentionally expanded the group. Instead, he
continued to lead the group with a guiding idea of “do everything with great diligence
and do well in everything we committed to do. Only then, others will find us and there is
no need for publicizing out works”. Therefore, the sources of agency action of this social
group arose from how its representative informant perceived the quality of its own

190
practices as well as its role in relation to others inside the civil society domain. From the
perspective of the participant, other formal NGO actors with more “professionalized
administrative structures” tended to be focused more on “publicizing and competing with
each other” and these are the things that the group (SG4) would refrain from doing
through the initial stages of development. This type of perceptual framework on the part
of the participant would indeed guide the group towards an activity-based functioning
focus. On the other hand, the intentional maintenance of an informal “group” structure
formed a dilemma when it came to making a decision to transitioning the group into
practicing long term in the field
26
.
The personal experiences of the participant himself were essentially intertwined
with the development of his group SG4. Before the earthquake, he was an ordinary
working class Chinese citizen. His daily lives used to be occupied with constant job
changes looking for the need of maintaining basic survival. From working as a car
repairman to a salesperson, it seemed to him that the meaning of living was to sustain the
pattern of “daily grind of work”, while dreams and purposes of life were nowhere in the
picture of his existence. After the catastrophic event and joining the group, the participant
recognized a particular change in his way of thinking about life in general. From the daily

26
For more information on the formal functioning activities of the actor, please see SG4-01-01 in the Appendix 4.4.06.

191
interactions with his group members, particularly from those opportunities to share
thoughts with university students, for the first time the participant recognized a gradual
change in him seeing life from narrowed view of his personal life to a reflection of the
Chinese society. Turning from being almost “ignorant of the social issues” to thinking
and pondering constantly about the “social justice concerns” that China faces and their
root causes, the participant posed the following questions: 1) “why do we always have to
wait until the last minute to recognize some measures needed to be taken or corrected?
(In this case, wait until the moment that civil society actors had to actively perform their
roles to complement the tasks taken by the state sector), and why not prepare the society
in such a way that social justice can already in place? 2) Social development is not just
economic growth, not like the extreme materialism exemplified in the Western countries.
It is one that involves the awareness at the level of minds and ways of thinking. At the
moment when the participant opened up and let out his inner thoughts to flow “freely” in
telling such a transforming story, as a researcher and also a Chinese citizen, I realized
that I have just witnessed a personal account of how a “civilian” whose cares for life only
involved a self-revolving matters, being transformed into a “citizen” whose life was
given a purpose towards giving and serving others in the larger society. Although this
particular participant and his group actor might not be recognized or event made their

192
names in the response and recovery history after the earthquake, his story demonstrated
the emergence of a civil society in its most primitive and bottom-up level.
As the participant led the group into the recovery period, he realized that the size
of the group was gradually shrinking. And one of the most distinguishing reasons that he
later recognized was many of its volunteers originally participated in the group left to
work for larger NGOs. While one can definitely sense a tone of disappointment, he was
quite proud by making a comment that the group started to become a “mediator in
transporting skilled volunteers” to other NGOs. Essentially, the participant did realize the
importance of transitioning into a formal organization to the survival and the influence of
the group in the long term. In this sense, “the power of a single person will be indeed
limited”.
Such a transitioning process explained why the actor’s outreach activities did not
persist over time as was being demonstrated in the results of social network analysis.
Although being less well-known among its peers and with less network influence, the
grassroots nature of the group and the life-transforming experiences of one of its key
participants provided a detailed illustration of how individual lives were being
transformed in such a way that the meaning and emergence of “civil society” in the

193
Chinese context is being demonstrated. The investigation in finding the sources of action
also revealed the creation of a sense of empowerment at the individual level through
agency actions based on collective and interactive voluntary efforts. Like the participants
of actor #3 and actor #24, the emergence of actor #4 as a civil society group was further
triggered by an awareness of issues of social justice and social development beyond
disaster response and recovery. The anchor of change was provided by actions that were
perceived to be self-giving and serving others towards a larger cause.
The last two actors belonged to the fourth category in table 4.10 were #7, and
#109. The number of out-degree ties of actor #7 immediately follows actor #4. The actor
was first established in July 2008, two months after the earthquake. But the effort was
originally initiated by a self-supported non-Sichuan resident who came to the Province as
a disaster response volunteer immediately after the earthquake. As the emergency
response activities receded, the initiator of the group decided to stay to pursue the
provision of educational services for children who were not able to access a regular
schedule to continue their education after the disaster. Similar to actor #4, actor #7 calls
itself as a “volunteer team” rather than “organization” or “center”. As we can see from
these two actors, the formation of this type of social group in the context of Chinese
disaster recovery is driven by a strong selfless and fearless endeavor at a personal level to

194
work for the betterment of lives of those impacted by the disaster. It is also characterized
by drawing on the volunteer energies from people with variety of backgrounds rather
than performing works as a continuation of a specialized formal organization, especially
at the earlier stage of response. Most importantly, the informality exemplified by their
structure and ways of utilizing the drive for volunteer service from members of society
with a wide variety of background at the formation stage of these groups seemed to act as
a driving force for these groups to carry their works into the longer term. Aside from the
example of actor #4, we can see that actor #7 also acted to sustain 21 out of its 24
connections into the disaster recovery period. A closer examination of its connection
partners reveals that over the long term, the actor lost its relationship with three other
actors that it established in the emergency response stage, two of which were based in
Beijing. The actor also developed communication connections with two new partners in
the network. One is the actor #6 and the other is actor #119, both of whom were
important players in terms of out-degree measures both before and after the earthquake.
The last actor in this third category that deserves attention is actor #109. It is a
non-profit association providing welfare for the elders. Based in Tianjin, a city close to
Beijing, it was first established in 1995 and was formally registered under the Ministry of
Civil Affairs. The actor started out by having no connections with any others in the

195
network before the earthquake. But after the disaster, it reached out to 21 others and
maintained these exact same set of relationships throughout the recovery stage by
carrying out its specialty works related to supporting elders, disabled, and providing
psychological counseling assistance. On the one hand, such a form of maintaining
stability by keeping the exact same number of partners over time can be interpreted as the
actor’s virtue of persistence in maintaining its line of communication. On the other hand,
this could also be interpreted as a lack of a spirit of entrepreneurship on the action side of
the picture when comparing it with those who sought a growing set of network partners
over time while maintaining what they had established in the period before.

Post-Earthquake Actions (Recovery)
Going back examining figure 4.5B, one other thing to be noticed is that for the
first time, actor #97 emerged to become one of the top active agents in establishing
communication ties aside from the level of action of actor #3, #51, and #24.

196
The Case of Actor #97 (SG97)
In terms of looking beyond disaster recovery in action outlook, actor #97 shared
the similar drive in making contributions to the long term social development of the
Chinese society. Its actions persisted to such an extensive degree that it became one of
the core actors in terms of initiating connections in the communication network
environment at the recovery stage. Examining the actor’s ego-network results
27
, it
emerged to play multiple roles in facilitating information exchange among actors with
different registration status, such as “brokering” and “gatekeeping” for post-earthquake
response and recovery. A further look at the “behind-the-scene” stories for how roles are
formed is informative to understand its emergence.
The experiences of the key organizer of actor #97 revealed several factors can be
attributed as significant sources of individual action and motivation for the initial
formation process of the actor. First of all, it was the emergency response activities
collectively experienced by the individual herself and those she defined as “her own
circle of friends”. This “circle” included her personal networks in terms of work,
friendship, and family relationships. Regarding the aspect of individual experience, the

27
Please refer to Appendix 4.4A to 4.4C for ego-network results of actor #97 over the three time periods.

197
organizer (NGO97-01) recalled that she had been conducting part-time works assisting
counseling and training services in the nonprofit area of practices since before the
disaster. However, this type of activity only occupied a “small part of her life” before the
earthquake, and she did not regard practicing in the “field of NGOs” as a profession that
would require full time engagement. After the earthquake, “things have changed
significantly” in terms of the way she experienced her immediate personal network
relationships:
This is because the earthquake stirred up a quite big impact, including my friends
who are professionals. They would all participate in the medical support or other
kinds of support efforts. So we were all drawn in because of the event, either
willingly or unwillingly. You will be involved anyway. And will put a lot of
efforts and time into this (NGO97-01-01
28
)
Note that the disaster event was perceived as an agent acting to “stir up” on how she took
action in becoming involved and eventually being engaged in the field of NGO practices.
Initially, the emergency response efforts were collectively experienced through the
intertwined relationships that the organizer had with her professional colleagues. Such
has led to the organizer’s own increasing amount of efforts and time being devoted to the
field of NGO practices, particularly during the immediate stage of disaster response.
What is worth further noticing is the “stirring up” impact of the disaster event on the

28
Original Chinese script refer to Appendix 4.4.07.

198
development of individual motivation and eventually sustained action. The key emphasis
here is to understand the motivational factors in this context. Based on the account of the
organizer, the factor can be summarized as an openly-recognized awareness of the
capability functionings (Sen, 1992) that became available in terms of revealing the
“resilient” ways that Chinese civil society copes with crisis situations. The following
account indirectly refers to how the organizer perceived such type of “resiliency” as part
of her experiences:
At the beginning, we were all engaged in this field of work, but I have never
thought of developing it into so and so, just felt that this was a good way of doing
things. Because I was also conducting this aspect of work before the earthquake,
then, I started to feel to that this was an interesting field of work to be engaged in,
and wanted to go further into this direction. So that’s why I drafted a project
application, which was approved later. I really wanted to do something at the
community level, and they (actor #51) also held the similar ideas, so got approved
by them. I thought this would be a good opportunity to do it by myself. (SG97-01-
02
29
)
Such awareness of acting out the available capability functioning choices to engage in
the civil society domain was not clearly recognized at the beginning stage during the time
immediately after the earthquake. Like the participants in SG3 (actor #3) and NGO49
(actor #49), the “resilient” ways of response was initially reflected in the individual’s
particular interest in practicing in the field of NGOs as a preferred direction of

29
For original Chinese script please refer to Appendix 4.4.08.

199
engagement in civil society domain. The factor in maintaining the functioning capability
reveals the second stage of “resilient” responses. With the assistance of another civil
society actor, the organizer of actor #97 was able to develop the group’s own autonomy
in functioning towards a professionally oriented civil society entity engaging in the long
term social recovery.
In this long term recovery stage, the motivations and actions at the individual
level experienced a direct shift from activities engaging in the area of emergency
response efforts towards social and community development that extended beyond
disaster itself, and towards long term mitigation and risk prevention. Similar to the cases
of #3 and #49, such transitioning dynamics of participants’ recognizing their roles over
time gradually became a more refined outlook towards the Chinese society as a whole
and how the civil society groups/organizations can contribute in developing a social
capacity for “resiliency” through ordinary situations. This is most clearly reflected in the
personal account of actor #97 organizer:
Later on I thought about taking opportunities to apply for doing projects relate
back to the urban communities. This is because I think human rescue includes
multiple facets, especially after a disaster. For example, at the earlier stage, we
have conducted a lot of work, good or bad, either providing assistance or relief, it
seemed like all the energy and efforts were concentrating on those groups of
people in the areas that were significantly impacted by the earthquake. But the
thing is, when the earthquake happened, I was also in Chengdu, also tried to run

200
away. In fact, all of us have experienced this process, including responding to the
security alarms and so on. (SG97-01-03)
So, it’s not just those who were living in the areas that were being damaged by
the earthquake were in need of attention, or can be called disaster victims. Or
there is no need to put on that ‘hat’ called ‘disaster victims’. If there is a need for
attention, we all need attention, including those living in urban communities.
This is because we find that there were lots of things that were missing even in
ordinary communities. Up until now, for example, we really don’t know how to
respond to the emergency situations, how to use emergency exits, how to provide
assistance. There are many things related to emergency response that we should
have known but really did not know. Also including those basic understandings
that should be part of the common sense, but the ordinary Chinese people really
don’t have. For example, many people don’t even know how to wash hands
correctly. We lack many of this kind of common understandings in our daily
lives. That’s why I am more inclined to go into the ordinary crowd. I also felt
like much more attention was being paid to those living in the disaster area, and
with the increasing assistances from the government on this group of people, the
good results must be accumulated. So I felt like I can redirect my attention to
focus on the common crowd and conduct my work related to ordinary
communities. (SG97-01-03
30
)
Her account here essentially reflected a particular lens through which how the recovery
period was perceived by way of a type of social development. First of all, note that how
the response efforts had prompted the participant to further ponder what it meant to
“rescue” and help the people whose lives were impacted by the earthquake. From her
first-hand experiences in physically and emotionally going through the happening of the
disaster, there was an eventual personal discovery of how to define the concept of

30
For original Chinese script please refer to Appendix 4.4.09.

201
“disaster victims”. Secondly, the lack of knowing the appropriate emergency self-help
activities at the very first moments when a disaster hit would, according to the participant,
put those living outside the areas of significant earthquake impact as the “victims”
category, specifically the vast majority of the urban population. The lens that she chose to
understand the “human rescue” aspect of disaster recovery, therefore, further expands to
include the everyday practices conducted by people in the society as a whole. Essentially,
the role for the organizational actor itself was not one that is being confined to response
and recovery from this one particular disaster anymore. The actor’s position was
perceived as one with a focus on “mitigation” and “preparedness”, which would
incorporate issues involving social capacities for coping with risks. Thirdly, this coping
and adaptation mechanism is one that needed to be developed inside the “ordinary people”
living in “ordinary communities”, rather than completely focusing the attention on those
in the areas “significantly hit” by the disaster event at the time. Such is the motivation
behind the sustained actions of the group that eventually led to its establishment of a
nonprofit entity functioning in the area of health for urban community population.
Fourthly, the last section of the account also revealed how the participant saw the
relationship between a civil society actor and the state. Note that she did not perceive
functioning of the two sectors as being one replaceable by another. Rather, it is one of

202
complementarity while confirming the essential role of the state especially inside the
areas that were significantly hit by the earthquake
31
.
The second conceptual re-evaluation reflected through the participant’s account is
the understanding of activities that can be counted towards “emergency response”. To the
organizer of actor #97, the concept is interpreted with a much broader perspective that
incorporates incidences beyond those conducted only after a catastrophic natural disaster.
Most of the times I would think that responding to disasters at our level of work
only involve the aspect of emergency response, but emergency events cannot be
confined to only one kind. Most of the times we will all encounter similar
circumstances or those that are even more severe and difficult to deal with. Then,
knowing how to respond to these other emergency situations would be very
important. So the reason that I wanted to do works related to ordinary
communities is because all have to face these emergency circumstances in our
lives, although of different kinds for different people. It’s not just after a
catastrophic disaster that we have to deal with such situations. Then, we need to
think about how to respond and face them, how to build a community social
support network, how to develop healthy communities. And these are the things
that I would prefer to choose to focus on. (SG97-01-04
32
)
Similar to her expanded interpretation to the concept of “disaster victims”, the organizer
provided another lens demonstrating the breadth of the scope of “emergency response”
activities. Note that she understood it within the context of how people respond to crisis
situations that arise from people’s everyday lives. Although different in their particularity,

31
For further related original accounts, see Appendix 4.4.Case97.1.
32
For original Chinese script please refer to Appendix 4.4.10.

203
the “normality” of having to face and deal with emergency incidences is perceived to be
part of people’s lives, and sometimes these circumstances will be even more difficult to
cope with than those come from a natural disaster. As a result, the initial motivation in
the search for sustainable action responses to position the functioning of the actor #97
came from the organizer’s awareness of taking a social responsibility for the
organizational actor’s long term role development. The establishment of a healthy
community support network assisting ordinary people to cope with these “emergency”
situations is recognized as part of the desired coping mechanism.
Such a motivation for the formation of actor #97 cannot be separated from the
organizer’s own personal transformations after experiencing the earthquake event.
In fact, change includes many facets. For example, how you change as an
ordinary person, because you experienced this thing, as everyone living in
Chengdu had experienced running from the earthquake, running down the stairs,
including receiving notices to avoid later earthquake impacts, living in tents, all
these experiences. This was a significant event in our lives. And your whole life
and ways of living will be touched and changed. Even though I did not participate
in the response efforts, as long as I have experienced the event itself, I will still be
changed. Many people will re-orient their ways of thinking about life in general,
including how to develop themselves in the future. This is why even if I don’t
participate in the volunteer activities and engage in the field of NGOs, my life still
would be changed as a result of the earthquake event. (NGO97-01-05)
The second type of change is that…like many of my friends…they are
professionals practicing in the field of NGOs, and would be drawn into the
response efforts after the disaster, whether it was proactive or reactive. Then, you

204
will have experienced the entire process, although it would be different for those
living in Chengdu and those living in the areas most significantly impacted by the
earthquake. But regardless, as long as we have experienced to be part of it, our
lives would be going through big changes one way or the other. It did not matter
whether you participated in a full-time manner or part-time manner, we were all
volunteers. As long as you participated in the process, it will be different. Some of
my friends are continuing doing it as part-time volunteers, and they kept doing it.
This is a type of change too, behavioral change, a change of living arrangements.
They wanted to arrange these kind of works into their lives now, which is very
encouraging. The development of volunteerism in the entire Sichuan Province can
be traced back to the happening of this earthquake event. So many people have
come to know and become volunteers, and treated it as a way of life. It doesn’t
have to be full-time devotion. But for me, I just decided to become a full-time
employee working for a formal NGO in order to be directly involved in the field,
and this is a type of change as well. (SG97-01-05
33
)
One factor of primary importance is the “life-transforming” characteristic of both types of
changes. At the personal level, the significance of the event changed people in ways of
how they see their own lives. At the level of action in terms of the voluntary coordination
and self-organization process, there is a type of transformation in the perception and the
behavior of those participated in such a process. Note that this kind of transformation is
characterized by action taken in the emergence of civil society almost becoming a way of
life blending personal lives with professional ones. What this means is that the
earthquake itself, was not simply an event that triggered certain temporary voluntary
response particularly from those who directly experienced it. The transformation is

33
For original Chinese script please refer to Appendix 4.4.11.

205
perceived as a type of social change that fundamentally altered how people look at
themselves as well as how they related towards others to make contributions to the
society. Essentially, such social change was first triggered by a life-altering event,
enhanced by the recognition of civil society actors’ social responsibilities in building up
the capacity for risk-coping for the society in general. And it was further distilled through
the actor’s participation in activities related to long term social and community
development.

The Case of Actor NGOLF (NGOLF-01)
The actor NGOLF was not listed as one of the civil society actors in the original
social network survey. Its unique existence status was discovered through the
acknowledgement and referral of the participant of actor #4. Recall that one of the main
concerns in transitioning the group towards a registered status, from the perspective of
the key participant of actor SG4, was a lack of independent decision-making mechanism
that could come along with such a formal institutional status. The participant of SG4 was
obviously being cognizant of the possible alternatives that the group could gone through
and well aware of the institutional routes being tried out by other civil society actors.
Therefore, during my interview session, he kindly mentioned NGOLF as an example.

206
The uniqueness of NGOLF lies in its institutional status in terms of registration format,
which differentiated itself from all the other civil society actors examined in the network
investigation. It was decided that such a characteristic renders a closer qualitative
examination of its nature of emergence and institutional status formation within the
Chinese disaster recovery context. The following thematic description of its experiences
will provide a further detailed documentation of the various types of institutional
transformation that civil society actors actually went through after the earthquake.
The motivational source for the establishment of this actor originated from a
determination to serve the grassroots groups and organizations stayed in the earthquake-
hit areas conducting long-term recovery works. As described by the participant NGOLF-
01:
At that time, many people were thinking this way (doing volunteer work), and
many of the grassroots volunteer teams emerged and later disintegrated. But since
the sheer number of emergence was so big, there were still many stayed and
maintained functioning long term. They’ve always wanted to do these kinds of
things, whether or not it is in disaster areas or not. There are some people I know,
they would continue doing similar works even when they went back to their own
provinces or hometowns, and became very active leaders regardless. The works
that we are focusing on is still related to disaster recovery. And this is because
what we wanted to do is to provide a communication platform for them, or

207
provide training programs, or other types of resources for exchange. (NGOLF-01-
01
34
)
This account depicted, from the perspective of the participant, that it was an intentional
choice being made for the actor to function in supporting the continuity of those newly
emerged social groups through the longer term. Note that after the disaster event, there
were two types of sustained self-organization processes been identified. One was the
action taken by groups that decided to carry out the works related to civil society into
places other than the disaster areas. This reflected the willingness of civil society actors
whose actions were brought forth after the earthquake event in bringing their devotion
back to their normal daily lives outside the context of disaster recovery. This suggested
the tendency of continued actions in the civil society domain long term after the disaster
event all the way into the phase of “resilience-building” beyond extreme situations such
as a catastrophic event. The other side of the picture was the action chosen by those
groups who decided to stay in the disaster area by being devoted to the long term
recovery phase and making contributions towards the social development of local
communities. It is in this latter action category that the actor NGOLF performed its tasks
in facilitating the communication channels particularly among grassroots groups
conducting recovery –related activities locally. From the outset, this motivation looks

34
For original Chinese script please refer to Appendix 4.4.12.

208
similar to the path chosen by actor #3. However, there were indeed differences among the
two types and the discussion will come back to this point in the following section related
to “source of civil society emergence”.
The participant also distinguished the date from which the actor became
registered and the date of the original initiative in action actually took place after the
earthquake. In other words, the question regarding “date of establishment” can be
interpreted by civil society actors from two perspectives. On the one hand, it can be
defined as the date from which the actor’s registration certificate first took effect. And for
this actor, it was towards the end of year 2010. On the other hand, it can be defined as the
first day that self-organization action was initiated. For the participant of NGOLF, this
latter type of establishment wasn’t as clear-cut as the former way of the definition. To the
extent of the exact date being identified, he was not able to recall and confirm the date of
the first agency act took place in the field.
LU: So when your team was first established?
NGOLF-01: “On the registration certificate, it says December 25, 2010. But I
cannot remember clearly of the exact date of our establishment date. For groups
like us, we can only tell the experiences for each person in that aspect. For
example, if you ask him or her when your group/organization did was first
established, they can only tell you when they first came to Sichuan. More or less

209
like this…the small and more grassroots groups slowly emerged out of a quite
hectic situation at the beginning.” (NGOLF-01-02
35
)
This account revealed the original nature of the self-organization process if it is being
understood from the date that the first individual action of a particular civil society actor
can be identified. In terms of understanding the initial formation of especially those
grassroots social groups that came into being after the disaster event, the second way of
defining this “date of establishment” can be used to trace the origin of action and
motivation during the primary stage of emergence of a civil society actor. Within the
context of disaster response and recovery in the case of China, the day of the participants’
arrival in the Sichuan Province after the event could also be a significant landmark in the
Chinese society realizing and acting upon the “agency freedom”.
As a nonprofit grassroots focusing on serving others civil society actors, the
“communication platform” role performed by the actor was carried out in the format of
providing training services. Initially, a notice will be sent to the local nonprofit groups
informing them the actor’s upcoming activities. Then, those who concentrated their
recovery works about in youth-development could send out representatives over to
participate in the training programs. At the time of the interview, the participant of

35
For original Chinese script please refer to Appendix 4.4.13.

210
NGOLF recalled having collaborative relationships with ten to twenty partners
36
. He
further recognized the important role of the online medium not only in connecting with
other civil society actors but also in opening up some of the first voluntary opportunities
for the participant himself right after the earthquake.
The internet has played an important role from the very beginning. This was very
obvious, and I can tell from my own experiences. After the earthquake, I felt the
need to do something, but not really sure how to start. Then, I went on QQ and to
take a look at what was going on. So gradually, I started to encounter and get to
know many others, some turned out to be professionals. For example, there were
some who came from Hong Kong or Taiwan. And some acted as our guide or
consultant, as they provide us with systematic trainings. Also like many
organizations from Taiwan, they also brought their experiences from the 9.21
earthquake and to show them to us. (NGOLF-01-03
37
)
Note that as the boundary of the voluntary coordination efforts expanded through virtual
means, the sources of actions and motivations also arose from the skills and experiences
being brought forth by foreign professionals in the field of civil society. At the level of
awareness and understanding for the actor’s own professional social work practices, these
learning channels found and established online provided valuable sources of guidance for
the development of the actor.

36
For detailed account, refer to Appendix 4.CaseNGOLF.1.
37
For original Chinese script please refer to Appendix 4.4.14.

211
Summary
This concludes the current discussion on the initiation and change processes of
actors’ information seeking activities comparing the periods before and after the
earthquake. I measured agency actions by using the network concept of “out-degree”
representing the number of communication connections that each actor reached out to. I
also looked at the normalized mean, which expresses out-degree as a proportion of the
remaining actors in the network. This is done in order to compare the measures across
networks with different sizes.
Among the top ranked actors with relatively high out-degree ties, the findings in
this section revealed that the earthquake event triggered two general types of out-reach
information seeking actions. For the actors that were already in existence before the
earthquake, the analysis showed mixed results as the non-Sichuan-based domestic NGOs
tended to recede in communication actions after the earthquake while others with
established local offices in the earthquake impacted areas tend to be persistent in
activating further relationships with others regardless of their country of origin and
registration status. For emerging actors that came into being after the earthquake, the
ones listed in Table 4.9 all reached a relatively high level of agency actions during the

212
emergency response period and were actively maintaining these relationships into the
long term recovery period. Some of these domestic grass-roots groups chose to be more
active in the longer term and this is demonstrated by the steady increase of
communication partners from the time when they were established to the time when its
three years after the disaster event.
On the one hand, “out-degree” measure represents actions taken on the part of the
initiator towards others for information and communication. On the other hand, it also
represents an outlook from which one understands how actors in a particular network
perceive the possible choice sets they may have and act upon such percepts. “Out-degree”
itself is thus an outcome measure that exemplifies such decision-making processes of
actors. I would like to formalize this concept as one way in measuring “capability”,
theoretically understood as a representation of the various combinations of
functionings—beings and doings—that the person (Sen 1992) or a group in focus can
achieve. Since the unit of analysis of this study is groups and organizations, the term
“capability” as this research revealed is a collective way of choosing. The availability of
the choices for each of the actors can be mapped out by the existence of the structure that
they were embedded in at one period of time. But I would argue that the availability of
choice sets themselves may not be sufficient to interpret “capability” within the realm of

213
“freedom”. The perceptions and behaviors of the actors also play an important role in
shaping their own ways of possible beings and doings. In other words, how they connect
with others, whom they choose to connect, and the circumstances that prompt their action
or in-action, are all crucial in defining the boundaries of their functioning capability set.
Therefore, as stated earlier, this section is a first step in my effort in exploring the
meaning of “capability” as originally reflected by Amartya Sen (1992). In the future
sections, I will continue investigate the different faces of capability from an action-
structure point of view.

Incoming Nominations and Status of Prominence (In-degree)
Overview
In this section, I examine the concept of “in-degree” in network analysis. This
term measures the incoming ties from others towards the focal actor. It is a way of
looking at action in reverse direction as compared to “out-degree”. In network analysis,
those actors with higher in-degree have generally been interpreted as the ones who are
“prestigious” or “prominent” among the others in the network. This is because those who
are being named by many others may possess certain degree of “power of knowledge”
that many would like to seek out to. Within the context of this section in investigating the

214
communication networks, one could interpret an actor with higher level of in-degree was
being perceived as a trustful source of information by others. In the following paragraphs,
I will look at the univariate statistics for actor in-degree measures in order to further
examine the various ways of understanding and interpreting “status of prominence” along
with considering the characteristics of the actors.

Treatment of Actors inside the State and the Market Domains
Before delving into presenting the results, I would like to introduce the way that
the actors in the state and the market system were being treated in this study. Recall that
the primary focuses of the research were the actors in the civil society domain in terms of
groups and organizations, their emergence, as well as their actions and roles in
constructing an institutional infrastructure that in turn shapes their own actions and
behaviors over time. In order to understand such a process of change, how civil society
actors relate to the ones in the state and the market system becomes an integral part of the
investigation. In the Chinese context, some of the most commonly discussed types of
relationships that an actor in the civil society has with the state and market domains can
be found in the form of legitimization with the government and financial support from the

215
private sector in the market system. Legitimization with the government often entails a
prolonged process of getting registration status from the Ministry of Civil Affairs by
finding a government agency that is willing to be the sponsor or by registering itself in
the business category. The private sector, often based on their orientation towards
philanthropy and the particular specialty of the civil society actor, can become a source of
financial support the latter. However, rarely do research look at the other side of the
picture, which is how the connections are perceived by social groups and non-profit
organizations themselves through the lens of different types of social environments, and
the consequential adjustments in their behaviors when establishing or making changes to
the relationships with each other over time. In order to examine these aspects of the civil
society in action, I designed the study in such a way that the state domain was
particularly referred to agencies, branches, and departments of the local government, and
they were all integrated into one entity coded as actor #1 in the survey questionnaire.
Similarly, the private businesses or any for-profit organizations were coded as an
integrated actor #2. This way, these two actors do not have any out-going ties in this
study. Any tie that directed towards them is understood as a perceived communication or
collaboration connection from a civil society actor’s point of view.

216
Pre-Earthquake Incoming Nomination Action
Power of Influence (Category #1)
Table 4.11 shows the categorical break-down of the actors who received high
level of incoming ties compared to the others in the communication network before the
2008 Wenchuan earthquake.
Table 4.11. Communication Structural Foundation (Pre-Earthquake Incoming
Nomination Action)
Actor Identifier Incoming Tie Intensity
Category 1 (#1, #50), #94 11, 9
Category 2 (#118, #119, #51), (#24, #27, #134) 8, 7
Category 3 (#100, #37, #38) 6
Category 4 (#137, #14, #19, #2, #25, #95, #110, #61,
#70)
5
The first category included actor #1, #50, and #94. Both actor #1 and #50 received a total
number of 11 tie nominations during the period before the earthquake. For actor #1, this
number meant that the government entity as an aggregate was contacted for information
and communication purposes by 11 different civil society actors during the period before
the earthquake. This number jumped up to 31 at the emergency response period after the
disaster event. In other words, only 8% of the remaining actors in the pre-earthquake
network perceived actor #1 as a communication partner. Its communication partners went

217
up to 22.6% during the short-term response stage. There were 30 actors, which accounts
to 21.9% of the remaining actors in the post-earthquake communication network
sustained their actions toward the state during the long term recovery period. Therefore,
there was a rather significant change (up to 14.6% increase) immediately after the
disaster and the level of incoming connections remained to be relatively persistent over
the long term. Such a trend can be interpreted in the following way. One is that the state
actor was considered as a trustful source of information among civil society actors when
it comes to disaster emergency response and became a “popular” point of contact as
compared to the connections it had before the earthquake. It also remained as a prominent
source of information throughout both the short term and the long term recovery periods
after the disaster event. The “status of prominence” of this kind of actors can be
interpreted as a power of influence for others to take actions to make connections toward
them.
As for actor #50 and #94, neither of them turned out to be a respondent to the
survey. The decision for keeping them in this section of the study is that they all have
connections that were named towards them and were, strictly speaking, still weaved in as
part of the fabric of the social structure that other actors were embedded in, or in other
words, depended upon. These connections were particularly important in communication

218
networks because having a high level of in-degree is a representation that others in the
network perceived the focal actor as a source of the needed information and the actor can
have certain degree of “power” over those who approached it in the first place. In the
context of this study, this kind of “power” can be manifested in the different
characteristics that these non-responsive actors might have, such as areas of
specialization and registration status. Therefore, we take a closer look at actor #50.
Different from the actors we have seen so far, actor #50 came into being as an
online information service platform for Chinese domestic non-profits who were seeking
institutional development and growth. First initiated in 2005, the entity existed as an
exchange platform similar to that of a public sphere that promotes public discussions
among emerging non-profit social groups and organizations in bringing about an
initiative of public participation of Chinese civil society education. The main medium for
such a public sphere in this Chinese context is through the interaction between an online
virtual society and a real world networked society. With its purpose of encouraging and
providing opportunities for those who would like to participate in the work of social
services, this non-profit entity did not simply play a role as an online information
provider but more importantly as an active participant in building Chinese civil society at
the most grassroots level. Therefore, I treated it as an “actor” in this research. The actor’s

219
website used the term “grassroots NGOs” to refer to the non-profits groups and
organizations that were formed by actions of Chinese citizens. Browsing through the
information posted on its webpage, one can navigate news, ideas, and tools related to
assisting the advancement of domestic non-profit actors. This partially explained the
popularity of this actor before the earthquake. Same with the state aggregate, it received 9
incoming ties. After the disaster, more civil society actors sought information from it and
the number of connections rose to 16 during the emergency response and 17 during the
recovery stage. Although this surge of incoming connection-seeking behavior from others
towards the actor is less in number when compared to those directed towards the state
actor for periods after the disaster, the results still showed that the earthquake brought
forth a jump in information-seeking partners for #50 and also a steady increase over time.
Another actor that received a high level of nomination from others in the network
is #94. It is a registered non-profit organization based in Beijing. It was first established
in 2007 as a Chinese domestic non-profit and its area of expertise is in environmental
protection. Although the actor was one of the prominent actors in terms of the high level
of nominations from others before the earthquake, this status was gradually given over to
a number of other actors across the network over time. Comparing its own nominations

220
over time, the actor had incoming ties with 10 others in the network for both emergency
response and recovery periods.

Communicator and Facilitator (Category #2)
The second category of actors that were being approached by many others
included: #118, #119, #51, #24, #27, and #134. With the exception of #118, the rest of
the actors in this group were also active in initiating connections which reflected in their
high out-degree discussed in the last section. Let’s first look at some of the characteristics
of actors #118 and #119. Both actors were originally based in Hong Kong and the
difference was that the former established its field office in Sichuan to devote works in
earthquake relief and recovery while the latter has been involved with the local
communities in the province before the earthquake since year 2004. Although as a late-
comer, actor #118 had a formal registration status with the Ministry of Civil Affairs and
its area of works during the recovery stage of the disaster had been concentrated in
livelihood-building activities, particularly in basic infrastructure reconstruction. One
factor that should be noticed is that even though the date of the field office establishment
of #118 was after the earthquake, during this period before the disaster, it still was

221
nominated by many others as a source of information and perceived as a “popular”
communication partner. This was due to a specific characteristic of how non-profit
groups and organizations function in the context of China. Such actors can start their field
work and practice long before they become a formally registered entity with either the
Ministry of Civil Affairs or as a business entity. For actor #118, it actually entered
mainland China and engaged in works related poverty reduction and disaster
management since 1987. Its projects expanded across multiple cities in mainland over the
20 years. Eventually, the actor got established its formal registered field office in
mainland China in 2004. Therefore, it is possible that the actor had built its reputation
over time and others were more willing to extend their communication ties to it for the
period before the earthquake leading to the disaster event.
Actor #119, in comparison, was not registered. But its field experience in
engaging communities dated back almost 4 years before the earthquake event. Regardless
of its institutional status in terms of registration, the areas of its participation in the
disaster recovery stage encompassed a group of much broader focuses than that of actor
#118. Aside from conducting activities in building up livelihood in local communities,
the actor was also involved in housing reconstruction, taking care of the elderly and
disabled, women and children, environmental protection, psychological counseling, and

222
most importantly in activities pertaining social recovery at the local level. The actor was
also among those who initiated high level of connections before the earthquake as well.
From this, it can be concluded that status of being “prominent” illustrated by the case of
actor #119 was enacted in two aspects: “communicator” and “facilitator” in the
communication network for the period before the earthquake. As a communicator, it was
active in reaching out to others to forge new ties not only to seek out information from
others but also to spread its own influences, such as letting other actors know its areas of
expertise and its values. As a facilitator, it shares information requested by others and
also becomes a medium connector among those who had ties directed towards it and
those to whom it initiated such ties. From a long term perspective, if actors with such
roles can be identified and the sustainability of these roles can thus be examined over
time after the earthquake. It is one of the beginning steps in defining the
institutionalization process within the civil society domain.
Looking across the three time periods incoming ties for actors #118 and #119, the
in-degree measures for both of them increased steadily after the earthquake. The former
had 15 ties named toward it and the latter had 13 ties. Over the long term recovery stage,
ties for actor #118 increased to 17 and to 14 for #119. Compared to the type of surge in
incoming ties for actor #1, the type of increase for these two actors is relatively mild.

223
The other two actors in this category were #24 and #51. Recall that actor #51 also
had the highest number of out-degree activities in this period before the earthquake.
Therefore, it exhibited similar role characteristics with that of actor #119, being both a
communicator and facilitator in the network. During the emergency response period, the
incoming ties of actor #51 increased from 8 to 23, accounting for 16.8% of the remaining
actors in the network. This number climbed up to 28, representing 20.4% of the actors
during the recovery period. Therefore, this actor was also perceived as an “influential”
communication partner, not only before the disaster event, but most importantly shortly
afterwards and all the way into the longer term. The level of its perceived significance
continued to have a steady increasing trend after the initial surge during the emergency
response stage. Also considering its high level of out-degree both before and after the
disaster, what we can conclude about actor #51 is that its embeddedness in the network
originated from two sides. One is what I would call the “action” side of the story. It came
from the actor’s inherent willingness to put itself “out there”, initiating connections and
in network terms—the act of attempting to be more “influential”. But when examining
the specific types of actions that the actor involved in, we get a different picture for the
motives behind such “influence”. Established itself pursuing towards an “incubator”
function to provide support for the smaller grass-roots non-profits in China, we can say

224
that actor #51’s activity agency comes from its courage in developing the strength of
others in the network. In other words, its “influence” does not have to be interpreted
within the boundaries of counting the number of connections this particular actor reached
out to as measured by out-degree. The development of the out-degree activity of those
whom this actor has connection to over time can be utilized to examine the focal actor’s
degree of agency. This way, actor’s “attempt to be influential” not only can be reflected
in an inward-looking way through its own immediate outreach activity, but also can be
modified by turning the lens towards others revealing their actions. The other side of the
“embeddedness” story involved the perceived level of prominence of others in the
network toward the focal actor, which can be revealed through its in-degree measure.
Because #51 was considered as an important and trustful source of information both
before and after earthquake, it was reached out to by many others over time. What this
revealed was that the level of “acceptance” from others also plays a role in how an actor
can be involved in the network structure. Note that I decided to use the word “acceptance”
rather than “popularity” to represent the level of incoming ties. The reason is similar to
that of using “agency” rather than the phrase “attempt to be influential” in the out-degree
activity. “Acceptance” is an act of others in the network that can reveal the underlying
drives of those actors to take the actions to initiate those incoming ties toward the focal

225
actor. It is more others-oriented when compared to using “attempt to be influential”,
which is essentially self-oriented. Within the context of non-profit activities in disaster
recovery in particular, I argue that such an outward-oriented interpretation of the
embeddedness of actors could be an alternative representation of this type of actors’
motivations and drives than seeing their actions as a result of merely wanting to be
“influential” on top of others or “popular” for their own purposes. This can be revealed
by the types of activities that they participated in, and the emergence as well as
sustainability of ties the actors forged over time.
Actor #24 also shared these two sides of the story by being both an active agent
and being accepted by others in the network over time. The exception is that the actor
was not among the most active in initiating connections before the earthquake like that of
actor #51. It had few communication partners for this period of time. But its outgoing ties
reached to almost every other actor in the network during the emergency response period,
while close to half of these ties were maintained into the recovery stage. In terms of how
much others perceive it as a significant source of information and outreach partner, the
actor not only received the same number of nominations as #51 before the disaster, the
measure more than doubled into the short term response stage with a steady upward trend
into the long term recovery.

226
Both actor #134 and actor #27 received the same number of nominations as #24 at
this stage, which counted towards 5.1% of the remaining actors in the network. Since
actor #27 was a non-respondent to the survey, I will briefly discuss some of its attributes
and moving on to other actors who were embedded by both out-going and in-coming
connections. Originally coming to mainland China as an international NGO functioning
in the area of environmental protection, actor #27 established its field office in Chengdu
in 2003, overseeing projects related to sustainable development and environmental
education in rural areas of the region. The entity was formally registered as a non-profit
organization in 2008. This actor was further recognized as an important communication
partner right after the disaster as it was named by 14.6% of the remaining actors. The
number climbed up to 16.8% during the long term recovery stage. What this means is that
this actor was perceived to be a renowned source of information both before and after the
earthquake event. However, it is difficult to comment on its agency role as a “facilitator”
in the network because we do not have information regarding it’s out-reach activities.
For actor #134, its level of incoming ties increased from 5.1% before the
earthquake to 9.5% immediately after the disaster, and further advanced to 13.1% at the
last stage. We can see that there was a steady trend of the actor gaining acceptance by
others in the network over time and being contacted as communication partners. One

227
special role that can be observed about this actor when comparing to the others in this
category is by looking at its agency activities reflected in its out-degree. The actor was
not particularly active in building communication relationships before the earthquake.
This can be shown by it reaching out to only 1.5% of others in the network at that period
of time. In other words, actor #134 received a lot of information but did not send much
out. To use a phrase in network analysis, these types of actors are often called
“information sinks”, which means that they collect facts more than they create them for
distribution. To put the interpretation into the context of the actor’s characteristics,
certain attributes may help explain the emergence of such a role and its change
trajectories over time. First of all, the actor functions as a research center formed by a
collaboration initiative of two Chinese universities. The Center conducts civil society-
related research projects by particularly focusing on the development of Chinese
domestic non-profits entities. In other words, for the period before the earthquake, other
actors in the network perceived it as a trustful source of information but as an academic
research entity, it did not focus its attention solely on outreach and sharing information
with other non-profit entities as it primary purpose. Secondly, the actor also conducts
action-oriented research, from which its goals were defined as providing strategies for
non-profit entities to develop and promoting general volunteerism in the society.

228
Therefore, aside from its research purposes, the actor did strive to be an agent of change.
This role was revealed and brought forth after the earthquake disaster when it initiated
contacts with 19 actors, accounting for 13.9% in the network. Compare to its 1.5%
outreach level for the first period, one could say that the disaster prompted an actual
realization of the action side of its goal. During the emergency response stage, its role
also turned to be one that was more of a “communicator” rather than an “information
sink”. This can be seen by its outreach activities spread through 13.9% of the network
while its incoming nominations reached to 9.5% of the network. However, this trajectory
did not fully sustain itself into the recovery stage. The actor’s outgoing activities lowered
in intensity to 8% of those in the network while the incoming ties reached at 13.1%. But
this did not mean that the actor was switching back to a role close to an “information sink”
as in the first time period. Based on its increasing outreach activities from building
connections to only 1.5% before the earthquake to 8% of the remaining actors in the
network during the recovery stage, what can be concluded was that the actor transformed
itself into not only a “communicator” with a growing level of agency efforts but also
being empowered by others’ recognitions to be a “facilitator” with more actors willing to
share information with it.

229
“Home-grown” Communicator (Category #3)
The third category is composed of actors #100, #37, #38. They all received 6
nominations. For the period before the earthquake, each of them was named by 4.4% of
the remaining actors. Actor #100, a Sichuan-based community service and development
center, was also an active initiator of communication relationships at this period of time
when it reached out to 11.7% of the remaining actors. Comparing this measure with the
level of incoming nominations at 4.4%, it is safe to say that the actor was at a stage of
actively building ties and proactively sharing information towards others in the network
before the earthquake. From the figures for the period immediately after the disaster, it is
apparent that the tie numbers for both of its out-degree and in-degree went up with the
latter figure doubled. When examining the mean for this period, the actor reached out to
13.1% of the remaining network as compared to being nominated by 8.8% others. What
this meant is that while continuing trying to be an “influential” actor, it was also being
increasingly recognized and sought out as information sharing partners shortly after the
earthquake. However, such opportunities for the actor to be on the receiving end of
communication ties and being a “facilitator” to pass on the information to others were not
sustained into the long term recovery stage. While keeping itself up with creating ties and
expanding its network to 14.6% of the remaining others during the long term recovery

230
period, the actor’s incoming nominations went down to 5.1%, which was almost
comparable but slightly higher than the figure before the earthquake. In general, as the
actor came into existence before the earthquake event, the disaster seemed to have
triggered a temporary surge in the in-flow of information as well as incoming
communication ties during the emergency response stage. However, such a trend did not
sustain itself for the long term. The actor, overall, can be categorized as an agent in
communication tie construction throughout the three periods of observation.
The commonality among actors #37 and #38 is that both of them are Chinese
grass root non-profit research centers. Their areas of expertise focused on poverty
reduction and sustainable growth in the rural regions of Sichuan Province. Both were
established before the earthquake event. For actor #37, its network outreach connection
activities were completely triggered by the happening of the disaster. This was revealed
by its outgoing ties increasing from 0% to 15.3% before and after the event respectively.
Comparing this to the level of its incoming nominations over time, the actor started out as
an “information sink” in the sense that during the period before the earthquake, it did not
send out any initiation signaling its efforts in information-sharing and connection-
building towards others in the network. This happened at the same time when the actor
was often being contacted by others as its incoming ties were counted among the highest

231
reaching to 4.4%. For the period shortly after the disaster, the actor became highly active
in that it connected to 15.3% of the remaining actors in the network while its incoming
nomination level climbed up to 8%. Over time, as the actor was increasingly being
recognized and contacted by 11.7% of others in the network during the recovery stage, its
initiation activity went downward to 8.8%. This shows signs that the actor role was being
transformed from an information receiver and a reactive agent to an active agent in
building its communication relationships through its response of the disaster event.
However in the long run, the actor showed some tendency to reverse back into its role as
a receiver during the long term recovery stage.

“Home-grown” Sources of Information (Category #4)
Moving on to the last category, actors #137, #14, #19, #2, #25, and #95 had
relatively high level of incoming information and contacts from others in the network.
Recall that actor #2 is an aggregate for all the private businesses that were involved with
others both before and after the disaster. Together with actor #1, the two of them were
conceptual approximates for the state domain and the market domain. Any incoming ties
towards them were to be interpreted as how civil society actors perceive their roles in

232
relation to the state and the private enterprises. In other words, the changes in the number
of incoming ties towards the two actors can be thought as measures by which to look at
how “influential” they were through the “eyes” of the other actors in the civil society.
Such a measurement for perception originates from the behavioral changes of how civil
society actors make their decisions to reach out to actors in either or both of the state and
the market. In this study, I argue that being able to tell explicitly the particular changes in
network behavior from one set of actors to another is an important step in designing
future policy interventions that could encourage cross-sector communication and
collaboration, especially for issues like disaster recovery and mitigation.
The number of incoming ties for the business aggregate was 5 at the stage before
the earthquake and that accounted towards 3.6% of the remaining actors in the network.
Comparing this level of nomination with that of the state aggregate at the same period,
the latter was contacted by 8% of the actors in the network. But at the emergency
response stage right after the earthquake, private businesses are perceived to be a
“popular” communication partner by an increasing number of civil society actors. The
nomination level reached to 16.1% during emergency response, which became almost 5
times more than that before the earthquake. This amount of tie changes was higher than
the changes occurred for the state actor comparing before and after the disaster. The

233
incoming percentage of connections for the state actor went from 8% to 22.6% before
and immediately after the disaster event. It was about a 3 times increase as compared to a
5 times increase for the business actor. What this means is that the emergency response
stage of the disaster generated certain amount of synergy among the civil society actors
towards their perceptions of private businesses and further triggered their actions to
initiate contacts with them. With three less incoming ties for the recovery period, the
level of synergy dropped by a few percentage points from 16.1% to 13.9% through the
long-term recovery stage. An overall comparison of the incoming tie trend for the state
actor and the business actor revealed that the government entities were still considered to
be the primary source of communication contacts both before and after the earthquake.
Although the civil society actors recognized the important role of private enterprises
immediately after the disaster, the level of actions towards the business actor did not
reach the intensity as those towards the state actor when looking across all three stages of
changes.
Moving on to the rest of the actors in this fourth category, actor #137 was being
reached out from 3.6% of the remaining actors in the network all the way to 10.9%, and
11.7% for the three periods respectively. Again, the majority of the change was “jump-
started” during the emergency response period and remained stable afterwards.

234
Comparing its level of in-degree with its out-degree discussed in the last section, we can
see that the actor actively initiated communication ties with 13.1%, 21.9%, and 12.4% of
the remaining others for the three periods respectively. The general trend is that the actor
tends to play a role of agency in terms of reaching out to others more than a role of
facilitator in information sharing. This can be inferred from its higher levels of tie
initiation than tie reception across the three time periods. Both in-degree and out-degree
experienced a “jump-start” right after the earthquake with a more dramatic change on the
side of its in-coming ties. The actor did gain attention from others in the network and was
approached by increasingly many others over time while its agency initiatives tended to
wane down towards the longer term.
Actors #14, #19, #25, and #95 are all Chinese domestic grass-roots non-profit
organizational entities in existence before the earthquake. With the exception of #14 and
#25, the other two actors responded the survey so that their out-degree measures can be
observed. Actor #19 is a Beijing-based and registered non-profit organization. Its
functioning focus has been in the area of poverty reduction specifically targeting rural
women in China through empowerment education with an emphasis on their gender
awareness. The entity was first registered in 2001 under the business category. During the
disaster recovery stage, it engaged in actions assisting women and children,

235
psychological counseling, and livelihood development. Examining across the change
patterns of its in-coming ties, 3.6% of others in the network named it as a communication
partner before the earthquake. This was followed by 8% and 8.8% for the emergency
response and recovery periods respectively. It was recognized by more than two times in
terms of the percentage of other actors reaching out toward it right after the earthquake.
And the incremental change remained at a relatively stable level into the long term. For
the out-degree activities, the actor started out by it reaching out to 5.1% of others before
the disaster, to 2.2% during the emergency response period, 19.7% during the long term
recovery stage. Instead of actively building connections with more actors in the network
right after the earthquake, actor #19 reached out to a level of contacts that was less than it
started out with. But the surge of the number of out-going ties exemplifying its agency
effect actually came during the recovery stage. This was a distinguishing pattern among
the actors that so far have been examined. On the one hand, actor #19 was increasingly
recognized by its peers throughout the periods after the earthquake. On the other hand,
the actor actually narrowed down the number of its communication partners to three
others in the network, with two of them being actor #25 and #95. The other one is actor
#3, a Sichuan-based grass root non-profit that was established after the earthquake. Then,
actor #19 became highly active only during the longer-term stage to initiate contacts. One

236
possible explanation for this kind of behavioral change pattern can be attributed to the
actor’s location and its field of expertise. The lack of an established operating field office
in the area where the earthquake happened could hamper its ability to communicate
effectively with others that were operating in the field. But this factor would not explain
how the actor became active during the recovery stage. Based on the actor’s practicing
field, while it did have the motivation to actively participate in the response of the
disaster, finding the communication partners that it could share information particularly
regarding the development of rural women could be the reason why that there was a
delayed effect on agency activities. In other words, reaching out to the “right” ones could
take time.
One other possible contributing factor that helped the actor to gain agency
response is to whom it was connected at the second stage. Actor #3 became an important
coordinator of information immediately after the earthquake, with its out-degree reaching
to 55.5% of the network and 27% nomination rate by others. The connection with this
type of facilitator could have assisted the focal actor further identifying its targeted
connection partners during the recovery stage.

237
Actor #95 also specializes in women’s rights and capacity-building in areas of
health, disaster relief, and rural community development. It was first established before
the earthquake and registered under the Ministry of Civil Affairs. During the long term
disaster recovery period, the actor participated in activities such as housing reconstruction,
taking care of elders and disabled, assisting women and children, and the livelihood-
building. This actor only initiated 3 contacts including to the state actor, the business
actor, and actor #3 during the emergency response stage. Its in-coming ties did increase
from being contacted by 3.6% to 6.6% and further to 7.3% of the remaining actors in the
network. While the actor became increasingly “popular” among others in the network, it
did not seem to open up to others and showed limited agency activities towards others.
The earthquake did trigger its initiative in communicating with some critical actors, but
these relationships did not sustain in the long run.
Since actors #25 did not respond to the survey, we can only observe the changes
in how others had reached out toward it. As we have stated earlier, incoming nominations
can be interpreted as the level of acceptance and influential-ness among others in the
network. If both actors #25 and #95 were also perceived as important source of
information after the earthquake, then, we can state that #19 has indeed reached out to
some of the “helpful” and “right” actors in terms of sources of information that even such

238
a limited amount of relationship contacts at an earlier stage could still generate
increasingly active responses later on.
Actor #25 was “reached out” by 3.6% of the remaining others for information
before the earthquake. The level jumped to 11.7% and 12.4% for the two periods
following the earthquake respectively. The actor received an increasing attention from
their peers after the earthquake and was being contacted more over time. It is a formally
registered nonprofit organization and officially based in Chengdu, Sichuan Province. As
it came into existence in 2003, the actor’s areas of expertise have been environmental
protection, particularly in water quality control including environmental education and
information exchange with local communities. After the earthquake, the actor started
coordinating intensively with actor #3 and the two actors often held joint educational
seminars in order to assist the growth of domestic grass root non-profit groups and
organizations. Therefore, it did make sense for actor #19 to get to know actor #25
through #3 and communicated directly with actor #3 to seek the needed information. This
type of “friends of friends are friends” relationship exemplifies one of the most basic
natures of social structure that actors can be embedded in the network. This study will
further discuss such characteristics throughout the later sections. For now, I will continue

239
examining the characteristics of actors in different categories in terms of their level of
communication connections.
Looking across the characteristics of the status of influence and the agency
actions of the actors in the fourth category, one thing they had in common was that they
were increasingly approached by others as important sources of information since before
the earthquake event. Although some of their agency actions fluctuated over time, the
level of acceptance toward them from other civil society actors’ point of view continued
to have consistent upward trend since the disaster. The source of this kind of influence
can indeed be attributed to their being Chinese “home-grown” civil society organizations
that persistently had been involved in local practices and the lives of people in the local
communities.

Post-earthquake Status of Influence (Emergency Response)
In this section, I examine the behavioral changes of civil society actors during the
emergency response stage. This is another way of illustrating action changes and
identifying emerging informational sources in response to the earthquake event. The
results are shown in table 4.12 below:

240
Table 4.12. Communication Structural Change (Emergency Response Incoming
Nomination Action)
Actor Identifier Incoming Tie Intensity
Category 1 #3, #1 37, 31
Category 2 #49, #51, #2, #27 26, 23, 22, 20
Category 3 #50, #24, (#109,#25,#125),
(#137,#14,#118)
19, 18, 16, 15
Category 4 #38, (#61, #119, #134), (#100, #70) 14, 13, 12
Category 5 (#110,#122,#123,#15,#19,#34,#37,#94) 11
Note: actors included in each parenthesis had the same level of incoming tie intensity.
Looking briefly across the five identified categories, numerous new civil society actors
that did not have high level of incoming ties became important sources of information
right after the disaster. Another observation was that the number of ties in general went
up significantly especially for those with higher in-degree measures.
New civil society actors, in the form of informal social groups and formal NGOs
organizations, came into being during this particular period of time. The variability of the
in-degree measurement can be looked at as an indicator of behavioral responses as a
result of the overall shift in the structure of the network after the disaster. The levels of
nomination were divided into different categories and they were created subjectively in
relation to how others were being perceived at the same period of time. For cross-stage
comparisons, I focus on discussing those actors that were newly created after the

241
earthquake as well as those that were in existence before the earthquake but did not make
it into the top categories for the pre-earthquake period.

Empowered “New-comer” (Categories #1 and #2)
The first actor that deserves attention is actor #3. As discussed earlier, it was first
formally established as a grass root nonprofit group just three days after the earthquake
on May 15, 2008. The entity remained un-registered until 2012. Despite of its informal
status, it actively participated in the emergency response stage by defining itself as an
information coordinator for non-profit groups and organizations. These other civil society
actors came to Sichuan with a desire to engage in the disaster response process but
weren’t exactly sure of what kinds of assistances were mostly needed. Actor #3 assisted
in directing them to the right places and shared information upon requests with these
actors. Indeed, such activities can also be inferred by the words appeared in its name,
such as “disaster relief assistance” and “service center”. Looking at the actor’s in-degree
measurements, it received nomination from 27% of the remaining actors in the network.
This put it in the first place among all the other actors and also on top of the state actor
during the emergency response period. For a Chinese domestic grass root nonprofit, this

242
degree of nomination or “acceptance” rate by others in the network should not be
overlooked. From what we have explored so far, the actors that had maintained high
levels of incoming ties were all formally registered domestic or international nonprofits.
These organizations often had been operating in their fields for certain amount of time
even before the disaster event. One is tempted to expect these organizations that were
being “well-respected” before the earthquake to continuing occupying some of the top
“influential” positions after the disaster. While some of them did sustain their roles at
later time stages, there were two distinguished aspects about actor #3 that rendered it a
candidate for special attention in this research. One is the correspondence between the
timing from which it was first formed to the point that it was recognized as the “go-to”
actor among the other actors in the network. One would imagine that it normally takes
some time before a “new-comer” can be accepted by others in the network, especially
from those who occupied certain positions for a long period of time, let alone considering
whether this new actor can challenge the ways information exchange connections were
structured. The tie pattern changes for actor #3 demonstrated that the institutionalization
process of social structures can be “interrupted”, or more precisely be “renewed” at the
same time when the agency actions of certain new actors are being initiated.

243
The second aspect of this specific case was that the unregistered informal status of
actor #3 had not been the determining factor for others to make their decision on whether
to establish a connection towards it, especially at the stage of emergency response. The
actor was reached out by 27% of the remaining actors in the network and this
outnumbered the 22.6% received by the state actor. Actors that received higher levels of
incoming ties such as #50 and #94 from the previous stage were also lagging behind actor
#3 for this time period. Considering these two factors behind the emergence of this “new-
comer” in the civil society domain, examining the characteristics and the stories behind
its establishment at the group level became a necessary step.

The Case of Social Group Actor #3 (SG3)
First of all, the establishment process through which the group was
institutionalized informally as a whole can be traced back to the actions and experiences
of another key founding participant shortly after the earthquake. This participant came
from an academic background at one of the local research institutes and had been active
in the field of voluntarism before the disaster event. Years of practicing voluntary
activities provided her with close friendships and work connections with members of

244
other nonprofit groups and organizations. From her account, the proto-type of the group
(SG3) originated from an in-depth discussion among her and representatives from two
other partnering organizations amidst of the emergency response period.
Because at that particular situation, there were so many things that needed us to
do, but what were the things that we can actually do? A few of us were already
acquaintances, DX, TJ, and I were collaboration partners for many years, and we
were very familiar with each other. At the time when the center was first
established, many of us already knew each other well, either from previous
project collaborations, or through regular communications, all familiar with each
other. This was very different from Z’s team, many of the participants in his team
have never met each other before. They were completely strangers to each other
and got together under a special condition. This is the most obvious difference
between us two organizations at the very beginning. Their situation was different,
many people did not know each other, while only a few knew each other. I
remember when we first began our work, we decided not to engage in close
collaborations with complete strangers that we had never known about. What this
means is that all of our collaboration partners should be our acquaintances. (SG3-
03-01
38
)
As we can see, former organizational partnership relationships and thus friendship ties
established long before the earthquake event played an important role in the formation of
this group. Familiarity and trust were the two most valued factors that tend to determine
its future collaboration partnership-building processes. Here, the participant openly
admitted that one of the group’s first self-organizing rules was to find collaboration
partners only with those certain degree of familiarity has been established beforehand.

38
For original Chinese script please refer to Appendix 4.4.15.

245
Secondly, the decision to focus the group activities on information sharing was
also based on the organizer’s (SG3-03) awareness of the then available human and
financial resources to the group at the time. On the one hand, having worked side by side
with another grass-root organization that did participate in coordinating materials
resources that came into the area for earthquake relief, the organizer of group SG3
realized that the shortage of human resources that could be available to carry out such
purposes. On the other hand, the difficulties for obtaining its own financial resources also
became a constraining factor for the group to conduct activities that require large scale
involvement of human and material resources.
Another decision we made at that time was not to engage in fund-raising activities,
and a lot of us seemed to agree upon this. It was because we heard from Z’s team
that they had to coordinate trucks and truck of materials for transportation, and we
did not have the human and physical resources to conduct such grand scale
activities. And we not have the financial resources either. To be honest, our
organization was truly in short of financial resources. By then, during this meeting,
two of the other organizations agreed to provide us with some necessary finances
to support our work…so neither did we possess that kind of human resources, nor
did we have other kinds of resources, and that’s why we did not engage in
material aspect of response activities. (SG3-03-02
39
)
Thirdly, although the senior organizer (SG3-03) was well aware of the two types
of constraints mentioned earlier, by no means they had become un-surmountable

39
For original Chinese script please refer to Appendix 4.4.16.

246
obstacles for the group to be persistent in actively participating in the emergency
response efforts at the time. The group organizers quickly realized that the aspect of
information services was an important activity to consider. The decision-making process
was being revealed by the senior participant as follows:
We thought of the things that we could do at that time, one is information services,
to us this was very important, because…even those companies such as Wanke
would come to us to seek information, so we want to do this work related to
information resources. If material resources did come here, we also thought of a
way to coordinate “point to point” for locations. For example, if I am at Xiang’e,
there they urgently needed blankets, when outside donation arrived, if your
organization is here, then, we could direct you to go and directly transporting the
needed materials to Xiang’e. This way we won’t be acting as a mid-transit-point
for material resource transportations. We have made it very clear at the beginning.
We were an information platform, and if one had a need and the other had the
matching resources, then they can directly contact each other for further actions.
What we were responsible for was the release and the distribution of information.
So this is why you could see that we were more engaged in activities related to
information dissemination.
Of course we also conducted some other kinds of activities. Back then, we would
organize an information release session every morning. Many NGOs would come
to us and get information on where to go and those locations with open
transportation accesses. At that time, there was a group of volunteers from an
auto-traveler’s club, they all had communication stations in their cars, and that
became very helpful to retrieve on-sight information. As they traveled to different
disaster impacted areas, they could easily contact and inform us whether the
rescue team had arrived, which locations were in need of food and water. Then, as
many (NGO groups and organizations) from Beijing, Shanghai, and Guangzhou,
or other places around the country came to us, we would provide them with the
information we had to help them make the decisions on where to go. But they
would eventually have to make the direct contacts by themselves. So this was how

247
we operated back then and all the way until the end of May, when the situations
were basically stabilized. (SG3-03-03
40
)
There are several factors that can be characterized as the guiding sources of action when
the group was first self-organized and emerged as a formal entity. First of all, it was the
ability to clearly define the desired area of focus within a short period of time. Within
only several days after the disaster, the group participants were able to identify the
establishment of an information platform servicing other grass-roots groups and
organizations as its primary functioning focus. From an organizational perspective, this
told us that the agency performance of this particular civil society actor was purposeful at
the very first moment of emergence. This further reflected the “resilient” nature of
individuals who formed the group after the catastrophic disaster. The state of “resiliency”
in this case, can be characterized by clear goal-orientation and purposeful actions in a
period of uncertainty and extremely stressful conditions.
Secondly, the group emergence activities were also exemplified by the group’s
ability to carry out its goals within certain degree of flexibility. For example, according to
the senior participant (SG3-03), there were situations when other organizations came in
with material resources at hand but did not know where to direct them. Rather than

40
For original Chinese script please refer to Appendix 4.4.17.

248
turning them away, the group participants would establish a channel specifically directing
information to match local needs and the material resources being received from other
nonprofits. This way, not only was the information platform function fulfilled, but the
communication ties being made through the coordination processes also served to further
expand the group’s information-sharing network.
Thirdly, the guiding source of self-organizing actions also came from the group
organizers’ desire to serve other earthquake response and recovery nonprofit
groups/organizations. An aspiration to serve the needs of other civil society
organizational actors was one trait that distinguished this group from all others. The
senior participant made it clear by re-stating the group’s functioning purpose as one that
focuses on “orderly participation, effective servicing”.
One thing that I also remember is that on the first day, we called upon the focus of
our work as ‘orderly participation, effective service”. This was our own slogan
and mission. When we were deciding on what aspect of disaster response works
to focus our attention on, it became clear that our center would not directly go into
the disaster area, nor would we face directly toward disaster victims. We provide
services towards NGOs, particularly those focus their supports on disaster
response and recovery. We positioned our works clearly at the very beginning.
We are not an organization that directly participates in disaster relief efforts. What
we do is to provide services to those NGOs that worked directly in the disaster

249
impacted areas. This has been the case from the beginning until now. (SG3-03-
04
41
)
Some of the other types of activities that nonprofit groups/organizations have been
involved were: 1) assisting housing recovery, 2) helping elders and disabled; 3) women
and children; 4) environmental protection; 5) psychological counseling; 6) livelihood
reconstruction. As we can see, during the process through which the group (SG3)
initiated its coordination efforts helping all these other civil society actors coming
together with these diverse background specialties shortly after the earthquake event, the
opportunities for building up its own information network also expanded.
When it comes to understanding and tracing the action and motivational sources
for sustaining the group emergence structure over the long term, the driving factors, from
the perspective of senior participant SG3-03, was one that still intended to fulfill its
original establishing purpose of serving others. When asked about the long term plans in
the area of disaster recovery, it became clear that the motivation for the group’s
sustainability over time derived from the continued existence of other nonprofit
groups/organizations in the field practices.
What we are thinking now is that as long as there are NGOs working in the
disaster area, our organization should be committed to continue functioning. This

41
For original Chinese script please refer to Appendix 4.4.18.

250
is because there are many grassroots groups emerging, and they will come to us
for assistance such as information. We would also inform them of various kinds
of seminars, training sessions, or providing them with temporary office spaces
here, or other helps of similar kinds.
If none of the NGOs exists, then there will be no need for our existence anymore.
We have never thought about changing our work focus to another area. But this
will also depend on the situations. At the current stage, there is still high demand
for our services. It’s only been three years after the disaster. Take the NGO
activities in Taiwan and Japan for example, they continued their work on disaster
recovery even it’s been more than ten years. So this field of work here will
continue as well, as the impact of the earthquake here was even more catastrophic,
with more areas affected. So we think this would be the type of work that requires
long term commitment. And during these three years, our work has been
relatively stable, mainly through our service provision on our web and other daily
services. (SG3-03-05
42
)
Note that the participant chose the word “survival” to emphasize on the importance of
continued existence of other groups/organization and how it relates to the group’s desire
to continue performing its functioning through service for other civil society actors.
There was a sense “co-dependency” and “co-evolution” among this particular group and
all the others that it tried to provide services towards. On the one hand, one would not
survive without the survival of others. On the other hand, the capability to flourish among
others will also determine the ability for actor SG3 to flourish. In other words, “growing
together” is perceived as the key to making the initial emergence sustain over time. The
participant’s definition of “information sharing and exchange” activities was also

42
For original Chinese script please refer to Appendix 4.4.19.

251
expressed in a way to exemplify such as process of “growing together” and “co-evolve”.
Not only is the regular information sharing activities important, but also various kinds of
meetings, training seminars, and even the provision of temporary operation spaces, were
a part of what she accounted as “information delivery”. This is clearly a much broader
interpretation of what we normally think of activities of information sharing and
exchange. In this case, it incorporated a message of “solidarity” with relationship-
building in both personal and work functioning spaces. Therefore, the process of
“solidarity” or “cohesiveness” formation that we observed in the section of network
structural changes was not just a dynamic in the physical tie compositions, but most
importantly, it incorporated a willingness to build deeper connections that sometimes
involve relationships at the emotional level among group/organizational actors.
This point can be further demonstrated and elaborated by the young participant’s
(SG3-01) reflection on the decision-making process, from which the group was trying to
make the transition from an emerging volunteer entity at the emergency response stage to
a formally established institution with the intention to function in the longer term.
Our regular meetings every morning went on until the end of May, by then, the
most urgent emergency response period already passed. During that time, we all
did things together without distinguishing who was from which organization. But
after that emergency stage, many of us went back to our own original work
positions. But for some entities, they would immediately consider what aspects of

252
works they could focus on as a group/organization afterwards, for example, either
in livelihood or in information services.
In fact, the center is like a spontaneously self-organized entity, but at that
particular point of time, we all need to consider something…we also conducted
meetings to discuss whether there will be a need to continue our works. Many of
our partner organizations would suggest that we continue. On the one hand,
having such a communication platform would greatly facilitate the connections
among us. On the other hand, the existence of such a platform would also
facilitate the coordination of not only information exchange but also material
resources. In fact, for things like these, I think, it would be better to provide a
platform for connection and communication activities among local NGOs that are
based in Sichuan. Now we already have a foundational momentum going, it is a
good thing to provide opportunities for all of us to make connections with each
other, not just for collaborations but also for deeper emotional connections. (SG3-
01-04
43
)
First, note that the timing factor transitioning from the emergency response stage
into the long term recovery stage did prompt the need for the group to decide on the
continuation of its functioning. This suggested that in addition to the disaster event acting
as the original change agent inside the civil society domain, the evolution in the self-
organization process of the group was also directed by a set of self-defined timeline
guided by the behaviors of other civil society actors at the time. Towards the end of May,
which was about two weeks after the happening of the earthquake event on May 12, came
a period of time when the group had to make a decision on whether to continue
functioning as an information servicing nonprofit entity. That was also a point in time

43
For original Chinese script please refer to Appendix 4.4.20.

253
when most volunteers representing different groups and organizations who worked
together during the most immediate emergency response period went back to their own
routine work schedules.
From the perspective of the young participant, the motivational sources of the
continued functioning of the group SG3 as an information servicing center after the
emergency period were two-folds. One was the maintenance of a platform for mass
communication among group/organizational nonprofit actors. The existence of such a
facilitating entity was perceived as a foundation for not only channeling information but
also material resources. The other motivational source was the need for connection
among local “NGOs” in Sichuan Province. Note that the term “connection” was
interpreted in such a way that was more than just formal information sharing. Like the
interpretation of the senior participant, the term incorporated a type of desire for a
bonding environment among nonprofit groups and NGOs. It was also expected that being
“connected” in such an environment will not only “breed” further collaboration
opportunities but also building up closer relationships where emotions were involved. In
this case, these participants envisioned their group SG3 could act as a provider of this
kind of environment so as to facilitate the type of connectedness that incorporated
emotional bonding rather than simply the formal activities of information exchange.

254
It is also important to recognize how participants perceived and acted out
“solidarity” and “cohesiveness” behaviors as the group itself emerged through a process
of interactions with other participating groups/organization in just a few days after the
disaster event.
Since the very beginning, we were all very enthusiastic and dedicated to do all
these things, but there were still some issues along the way. Take those who have
been doing conducting works in this field for a long time like our organizer G,
they will often have an in-depth and forward-looking perspective to look at things.
At that time, we made it clear to defining our works as complementing what the
government was doing, and that means what we were doing must be of little
amount as compared to the government tasks. But at the same time, those were
also the things that the government might not have time to pay attention to
particularly at that emergency response period. And that is…the
saying…government lead, and our job was to participate, facilitate, and
complement. Also, they did pay more attention to volunteer works at that time,
and they understood that you came here to do good. So they helped print some
volunteer training brochures to distribute to those coming to the resettlement areas
and along with providing some training sessions. (SG3-01-06
44
)
At the very first moment of emergency response stage, the group’s emergence, as
experienced through the young participant (SG3-01) was characterized by the following
three traits
45
. One was the coordinated actions among sub-groups of volunteer activities
with different work focuses. The other was the “coming together” initiatives among
groups/organizations from other parts of the country working alongside with each other.

44
For original Chinese script please refer to Appendix 4.4.21.
45
For details of the interview, refer to SG3-01-05 in the Appendix 4.4.22.

255
The timeliness of their action to be at the site of the disaster demonstrated a sense of
eagerness and “togetherness” for civil society actors to serve the earthquake impacted
areas at the first moments of needed support. The third factor that characterized the
source of emergence can be attributed to a type of enthusiasm that incorporated the
willingness to complement the responsibilities held by the government. It is important to
recognize that the role of the group was perceived as one that was not confrontational
towards the action of the state, but one that was enacted by joint participation during the
emergency response period. This can be understood as a primary sign of how the civil
society and the state domain came in touch with each other at the group/organizational
actor level since the event of the earthquake. The source behind the “joining” process can
be summarized as a passion in service through the practicing areas that complement the
state activities. To this particular group, such area was being actively identified as the
need for information sharing among nonprofit groups and organizations through
emergency response and the long term recovery times.
The inherent nature of this group was also intertwined with its emergence process
as a stand-alone entity. At the very beginning, the proto-type of such a group was
founded upon the existence and interactions among many “NGO partners”. As each one
of these actors brought with them their own resources, organizational partners, and

256
volunteers when interacting with SG3, not only the group’s own network connections
started to build up but also more of the other civil society actors came to be aware of its
existence, thus reaching out towards it. At the emergency response stage, some of the
information service related activities practiced by the group included posting the most up-
to-date response related news on the website and holding daily report meetings gathering
information from the on-site teams. Transitioning from the short-term response towards
long term recovery, the agency actions of the group continued its primary motivation in
providing a relationship bonding environment for other civil society actors who came to
the area conducting disaster recovery-related works. Below are two accounts from the
young participant’s reflection:
This is because the foundation that formed our organization is based on many
other groups/organizations, on many other NGO partners. And they will come in
with their own resources and partners, or their volunteers. So, gradually, we all
got to know each other, and would come here. We would post information on our
website, because after all, we had the first-hand and just-in-time information at
that time. People would come and gather here. We always had a morning meeting
at nine back at that time and every task team would report on the important pieces
of information they had, where they had been on that day, the relevant safety
issues as such.
After the emergency response period, one important aspect of our work was to
assist (the development of) other groups/organizations through seminar programs,
trainings, and activities. All they need to provide is the attendees, and we will
facilitate and provide all other kinds of services for them. For example, if one
seminar is specifically related to projects financial management, or training
related to disaster management or environment related topics, we would act like a

257
resource ‘warehouse’ for our partners. We would invite the most appropriate
working partners to our training programs taking into consideration of the
requests of all the participating partners. We then would help them arrange all the
logistic parts of the conferences. These were the types of works that we were
occupied with into the recovery period. (SG3-01-07
46
)
As the long term activities transitioned towards more of a training and seminar-based
type, it can be seen that the group’s purpose of the service in the recovery phase focused
on facilitating the connections among other actors as well as on the provision of a
nurturing environment for newly formed civil society actors to grow and develop as
stand-alone formal entities. Again, this demonstrated the willingness of the group as a
whole to have an inter-dependent relationship linking its own long term survival to the
sustainability of the other civil society actors. Such a tendency as a service-oriented agent
was also brought about through an observant nature of the participant working for the
group, especially in terms of the awareness of local grass-roots nonprofit formations after
the earthquake event. For example, the young participant of SG3 realized that during the
period from year 2009 to 2010, many more grass-roots entities were being formed locally,
especially in the earthquake impacted areas in Sichuan Province. One type of emergence
was in the form similar to informal volunteer groups and has yet developed an idea of
perceiving themselves as “NGO associations/organizations”. Activities conducted by this

46
For original Chinese script please refer to Appendix 4.4.23.

258
kind of grass-roots groups had a more informal volunteering-orientation. Another type of
emergence was characterized by formal NGO organizations supporting local
communities in the hope that their action can empower the people inside the communities
so that a group of locally-grown “backbones” can eventually arise for long term self-
support. According to the young participant, many of the grass-roots formed after the
earthquake functioned with this latter type of purposes. Aside from the cognizant nature
of those representing the group as being fully aware of the state of emergence of other
civil society actors, the group’s practicing focus on providing services in capability
development and training for its partners during the long term recovery stage further
demonstrated two of the most important sources that drove the emergence of this
particular actor. One is the intertwined nature of the functioning of SG3 with the
emergence of grass-roots nonprofits through the provision of a bonding environment
shortly after the disaster event. The other was the continued self-defined action in
services that specifically targeted the long term growth of other newly emerged nonprofit
groups/organizations, thus developing a co-evolving process of inter-dependency
between the survival of the group itself and the capability for others to sustain themselves
over time
47
. Essentially, the group SG3 was a survivor on its own and at the same time a

47
Please see SG3-01-09 for detailed interview accounts in Appendix 4.4.24.

259
provider exercising its agency through a network that connected actors beyond the act of
information exchange and project collaboration. The connections reflected a type of
actors’ relationships at the cognitive level that not only could facilitate mutual growth in
the long term, but also a self-awareness of each of actor’s own responsibilities and
experiences. The following account illustrated the core of such an emerging process of
civil society:
When it was into the year of 2009 and 2010, our works mainly lean towards two
directions. Essentially, we were still doing service-related works, be it
information communication or coordination. Then, we gradually found out that
many of the very grassroots groups were being formed after the earthquake, and
they all emerged locally in Sichuan. One form (functioning) is that these groups
still don’t have the concept of “NGO” in mind, and function more like informal
volunteer teams. They still would think of themselves simply participating in
volunteer activities. Another form is...those did function as formal NGOs would
hope the ‘indigenous power’ would arise once they are gone. Maybe they
intentionally trained some of the local key participants to establish some of these
groups/organizations. These type of situations has occurred in large numbers
especially after the earthquake. (SG3-01-08
48
)
One other actor in the second category that I would like to pay particular attention
here is #49. Note that like actor #3, it came into existence right after the earthquake.
During the emergency response period, 19% of the remaining actors reached out to actor

48
For original Chinese script please refer to Appendix 4.4.25.

260
#49 to establish communication relationships. The percentage climbed up to 21.2%
during the recovery stage.

The Case of Actor #49 (NGO49-01)
Like actor #3, actor #49 also emerged as a self-organizing civil society entity after
the earthquake. Here, I focus on the characteristics of its individual action and
motivations in the self-organization process, or in other words, how the respondent lived
through his experience of the organization’s initial formation. Although the formal
establishment date of this NGO as a formal institutional entity can be traced back to 2008
after the earthquake event in May that year, it had actually been three years before the
happening of the disaster that the original group came into functioning related to “civil
society” practices. Here, the organizer’s usages of two term phrases were worth noticing.
One was that he specifically referred the entity as a “team” or “group” before its formal
establishment as an “NGO” after the earthquake. This meant that the respondent was
consciously aware of and distinguished the issue of formality. Secondly, the usage of the
term “civil society” when referring to the organization’s work practices signified the
respondent’s awareness of the concept and his willingness to identify both the “group”

261
and the “organizational” practices as belonging towards a larger domain called “civil
society”. The specific ways of how the “civil society” concept was understood in this
context was further revealed throughout the development of the respondent’s actions and
motivations throughout the emergency response and recovery stages.
Several characteristics can be depicted in terms of how the initial actions taken
that would eventually lead to the emergence of the organizational actor #49 during the
emergency response stage. First of all, the primary initiatives for the self-organization
process were collectively taken by the actions of about one hundred “Chinese domestic
NGOs” during the first moments after the disaster. The organizer himself (NGO49-01)
was “elected” as the lead coordinator at the newly established “joint office” across the
groups and organizations came into the Sichuan Province participating in the emergency
response. The initial decision-making process regarding the functions of the “joint-office”
as a whole was made collectively among the participating civil society actors. Just within
the timeframe of a few days, it was decided that the functioning of the organization
would involve actions of “specific rescue operations” during the emergency response
period. These activities included the distribution of the disaster relief supplies and the
transportation of volunteers. The following account illustrates the specific actions taken
among the individual participants of the “joint-office”:

262
At that time, the joint office had its own division of labor, and on my part, I would
mainly be responsible for fund-raising and managing daily affairs. For example,
the coordination of our transportation team and the investigation team, and raising
the material resources in our warehouse would be my responsibility. There were
also sub-teams, and every team would have its own duties, some were in charge
of gathering the updated information in the disaster impacted areas, some were
responsible for other types of information. And according to their information, we
would send the needed people, resources, and the transportation vehicles to the
proper location. So these were basically what happened during the emergency
response stage until the 30
th
of May. By then, the works of the joint office was
completed. Afterwards, my thoughts were…because so many of the civil society
groups/organizations participated in the emergency response activities, and the
disaster recovery stage after the emergence response should be a very long term
process, also according the experiences from Japan and Taiwan. So this is why I
decided to stay and continue doing long term disaster recovery-related services.
And afterwards, we thought about moving our headquarter to Chengdu and
change our name as well. (NGO49-01-01
49
)
Thus, during the emergency response stage, self-organizing actions basically can be
characterized by a differentiation of duties being performed by sub-groups of participants.
The particular role of the organizer (NGO49-01) was one that focused on general
coordination, administrative duties, and fund-raising. The end of May in 2008 marked as
a change point for the “joint-office” to make its transition from performing emergency
response related actions towards short-term as well as long term recovery actions. This
led to the discovery of the third trait of the self-organizing process. The organizer’s
awareness of the importance of disaster recovery phase and his perception of the

49
For original Chinese script please refer to Appendix 4.4.26.

263
functioning capability of all other participating civil society groups/organizations played
a critical role in the institutional transition of the actor #49. Drawing from the
experiences of Japan and Taiwan, he first understood that the disaster recovery phase
could involve actions that are intended to be long term. Then, the action taken
exemplifying the organizational actor’s further commitment to the field of practices was
reflected in the respondent’s decision to relocate the original functioning headquarters to
Chengdu, Sichuan, and also in the process of changing its title afterwards in order to
commemorate the rise of the “joint-office” after the catastrophic disaster event. As a
result, the organization under the new name was formally established on June 1, 2008
50
.
Therefore, when it comes to the initial drive in his initial participation in the
voluntary activities and establishing the organization, the younger informant of SG03-01,
NGO49-01 showed similar type of awareness in being able to be involved in something
more “meaningful” when compared to the daily tasks that he had to perform back at work
in the government. At the same time, they also recognized a force that empowered them
through the joint actions of participating in the disaster response and recovery among
many of his peers.

50
Refer to Appendix 4-Case49-1 for more detailed accounts.

264
I don’t feel like there is much future when working for the government…our
organization has a couple of us who are like me, three of them worked for the
government for many years. All of them resigned after getting their deputy
secretary title. They just didn’t see much future when working there…didn’t feel
like doing anything meaningful…we feel what we are doing now is a much more
meaningful and interesting one, so we chose this one. We also have a lawyer, he
has seven to eight years of working experiences. So that’s why the structure of
our team is also an interesting one. (NGO49-01-02
51
)
Aside from the emergence of new actors such as #3 and #49 and their rising up to
occupy some of the influential positions in the first two categories, they were followed by
actors #1, #51, #2, and #27. Note that #1, #51, and #27 were all among those with higher
in-degree measures for period before the earthquake. At the emergency response stage
after the disaster, they remained trustworthy communication partners from other actors’
perspective. Actor #1 is the state aggregate and it was consistently perceived as a “go-to”
source of information for these two periods. The percentage of civil society actors in the
network that reached out to it almost tripled after the earthquake.
The perception of nonprofit actors reflected by their choices of action is one thing,
the other side of the story is the policy implications for the various government agencies
in the state and their response upon receiving the information requests and
communication outreaches from civil society actors shortly after a disaster. In social

51
For original Chinese script please refer to Appendix 4.4.27.

265
network analysis, an initiation of a tie from one actor to another can be interpreted as the
former regarding the latter of such a high value so that the latter was chosen to be
contacted among all the other possible choices in the network. Establishing a new tie,
from any actor’s point of view, will cost the initiating actor’s time and resources while
the actor itself has to bear the risk of being rejected or the “interest” not being
reciprocated from the other party. Therefore, the fact that the state actor was contacted by
consistently higher percentage of actors in the network over time meant that the
government branches indeed played an important role in the changes of “lives” of civil
society actors. Within the context of this research, I would argue that the mere existence
of a tie between two actors, is a form of “power” executed by parties on both sides of the
picture. On the initiation end, the power comes from the actor’s ability to choose from all
other possible opportunities that it possesses at the time. In other words, the sheer
existence of an abundance of connection directions for the actor to act upon based on its
own intentions and goals can be interpreted as one form of agency power for actors that
focus on building ties. On the receiving end, the power not only comes from the
acceptance and recognition of others but also from it is actually being reached out to
through others’ actions. The usual interpretation of this kind of power is that the actors on
the receiving end must already possess certain reputation, resources, or positional status

266
that the other party would perceive as useful. This indeed could be the case for the state
actor. But in this study, I have already illustrated that this did not necessarily have to be
the case. Actors such as #3 and #49 certainly did not possess these achieved statuses at
the time when they established themselves right after the earthquake. But what they did
have was a drive that motivated them to serve for those in need at the time of crisis and
they acted upon it in the form of establishing new relationships towards others. And these
agency actions were being reciprocated by others at the same time. Therefore, “power”,
on this end represented an act of being empowered.
Therefore, at the same time that the government is trying to respond to the actions
of nonprofit actions or informal social groups, it needs to take into consideration of the
power on both sides. This means considering not just its own position by counting the
number of incoming ties, but also pursuing the motivations and meanings for those actors
who sent out those ties. This is critical because how its own action changes will also
alternate the structure of the social environment that the civil society actors are embedded
in and such changes will have an impact on how their behavior changes over time.

267
The Rise of Domestic Civil Society Organizations (Category #3)
Looking at the third category in the emergency response period, it is apparent that
actors #109 and #125 became the two other newly emerged “prominent members” among
all others in the same category. By “newly emerged”, I mean that the two actors did not
receive as many nominations as the ones discussed in the period before the earthquake.
Actor #109 was being nominated by 2.2% of the remaining others for the first period and
during the emergency response stage, the percentage climbed up to 11.7%, which was
more than five times from the previous stage. This level of nomination rate came down
slightly to 10.2% during the recovery stage. Observing its out-degree activities for the
three time periods, the actor went from a non-initiator of any ties before the earthquake to
one that established contacts with 15.3% of the actors after the disaster. And its level of
outreach remained exactly the same for the long term recovery stage. Overall, actor #109
was an “information sink” before the disaster happened. Initially, there were other actors
reaching out towards it but the actor did not send out any information. But its agency
action was triggered by the earthquake event during the emergency response period. The
actor’s level of agency as a tie establisher overcame its receiving nominations for both
the short term and the long term after the earthquake. This showed the primary tendency
for the actor to be an agent of change in the structure of the network.

268
The difference between actor #125 and the others that I have thus covered so far
is that it is a Chinese domestic foundation with a particular focus in disaster relief and
assisting the development of nonprofit actors in China. The foundation was first
established in 2007 and in year 2010, it became the first formally registered independent
foundation that can raise its own funds from the general public in mainland China. Based
on this specific characteristic of the actor, the communication ties that were being named
towards it not only suggested an interest of other actors to exchange information with it
but also their interests in seeking the opportunity to be funded by actor #125. With this
additional piece of interpretation, let’s look at how behaviors toward this actor change
over time. Before the earthquake, only one actor in the entire network sought to connect
with the foundation. But during the emergency response period, the extensiveness of its
incoming nominations went up to 11.7% and the level stayed the same over the long term
recovery stage. In this case, the earthquake did trigger a wave of recognition from others
in the network and this was most likely due to the fact that there was not only a surge the
number of newly emerging civil society groups but also an insurgence of the need to seek
funding as these groups develop over time. In other words, the level of in-degree for this
type of funding-oriented actors can be seen as an approximate indicator of the
development of grass roots civil society actors in China. The indication for this study is

269
that on the one hand, civil society actors in general were becoming more active after the
earthquake by looking directly at their out-degree activities. On the other hand, an
indirect way is to observe the change in the number of in-degree for actors that had
functioned as foundations. An increase in ties that were established towards a Chinese
domestic foundation formally registered to have the lawful status to raise funds from the
general public, can be a sign indicating a growing number of grass-roots
groups/organizations not only emerged at the time but also were aware of seeking to be
sustainable over the long term.

Expansion of Empowerment (Category #4 and #5)
In the fourth category, one actor worth paying attention to is #100. Recall that it
was an active relationship initiator throughout the three time periods. But it had relatively
few incoming nominations before the earthquake. After the disaster, along with its
increasing outreach activities, the actor’s in-degree level went from 4.4% to 8.8%. For
the long term recovery, its nomination dropped to 5.1%. While the behavior of others in
the network was fluctuating towards communicating with this actor, it maintained its

270
agency action level by kept establishing connections with others both before and after the
disaster.
In the last category, actors #122, #123, #15, and #34 started out to be increasingly
recognized by the rest of the network actors during the emergency response period.
Among them, actors #123 and #34 were newly established grassroots nonprofits after the
earthquake. Actor #123 was organized by a few volunteers who participated in the
emergency response activities and decided to stay active in the social recovery process
for people in the disaster impacted areas. Its incoming communication ties were initiated
by 8% of the remaining others in the network during emergency stage and the percentage
went up to 10.9% over the long term recovery. In other words, the actor was being
recognized and contacted by a growing number of other civil society actors in the
network over time. Examining its out-degree activities, it reached out to 21.2% of others
since its establishment after the earthquake, and this was one of the highest among others
at the same period of time. Over the recovery stage, the level of outreach dropped by a
few percentages to 19%. But in general, the actor sustained its efforts in being an active
agent for relationship building over time.

271
Like actor #123, actor #34 was also a newly formed social group established in
response to the earthquake. The participants were composed of Chinese citizens who
committed themselves to activities related to long term earthquake recovery in the area
and one of the main guiding purposes of their actions was to serve both the urban and
rural communities thus contributing to the sustainable development in the region. It was
formally registered in the business category and participated in activities such as
environmental protection and livelihood support during the period of long term recovery.
During the emergency response stage, the actor received nomination from 8% of the
remaining actors and this percentage remained at the exactly same level for the long term
recovery stage. Compared to #123, actor #34 was more active in building ties right after
the earthquake. Its outgoing connections expanded to 65% of the actors in the network at
the emergency stage, showing its eagerness to communicate with others and sharing
information. However, this level of agency activity did not maintain over time as the
number slipped down to 7.3% during the recovery stage. Therefore, we can say that the
earthquake indeed triggered a significant sense of agency and prompted it to drastically
expand its outreach activity during the short-term period after the disaster. However, the
“agency” interpreted in terms of the out-degree measure was not sustained over time.

272
Although some actors’ overall agency experienced great fluctuations over time,
such as the incidence pertaining to actor #34, it is still early to say whether the actor’s
role was indeed being transformed over the long term by the disaster event. This is
because the measures of in-degree and out-degree are just two of the many kinds of
network measures that are available to depict the varying roles and institutionalization
characteristics that actors may possess. Later on in the analysis, I will illustrate other
types of these measurements to look at the different sides of the processes behind the
actors’ actions. In these instances, the concept of “agency” can have various meanings.
For the case of actor #34, as its actions to connect came down significantly during the
recovery stage, it is also important to look at to whom the actor was connected to and
their positions when taking into account of the whole network structure. Sometimes,
getting connected to the “right” actor may be more important than knowing many of the
“isolated” actors who themselves were not well connected in the network.

Post-earthquake Status of Influence (Recovery)
In this last section of discussing the in-degree connections, I chose to discuss the
actors that were neither among the top in receiving nominations during the pre-

273
earthquake stage nor during the emergency response stage. It is important to look at these
“late-comers” during the recovery period two reasons. One is because they could be
newly emerged grassroots groups to participate in the disaster response and were able to
maintain existence in the long term. The other reason is that it could take a long time
before a grassroots group develop and become active in the field and finally gets
recognized from others in the network. Examining the characteristics of these “late-
comers” can help the analysis draw some inferences on how the origins of structural
change came into being, especially in regard to agency actions. Table 4.13 illustrates the
five categories of actors who held higher level of in-degree over the long term recovery
period. The categories were created in a ranking order from those with high in-degree to
lower in-degree
52
.

52
The actors in the parenthesis are the ones with same level of in-degree.

274
Table 4.13. Communication Structural Endurance (Recovery Incoming Nomination
Action)
Actor Identifier Incoming Tie Intensity
Category 1 #3, #1 33, 30
Category 2 #49, #51, #27, (#24, #38) 29, 28, 23, 22
Category 3 #2, (#134, #50), (#118, #25, #14),(#12, #125,
#137,#37)
19, 18, 17, 16
Category 4 (#123, #33, #61), (#109, #119) 15, 14
Category 5 (#135, #76), (#110, #19, #5, #71) 13, 12

Capability Formation
At this stage of long term recovery, note that actors #3, #49, #123, and #109, who
received high levels of nominations immediately after the disaster, maintained to have
relatively higher levels of nominations from other civil society actors in the network. And
among them, #3, #49, and #123 were all grassroots groups established only after the
earthquake. They remained to play an influential role in the network by continually
ranking ahead of the state actor and other long-established nonprofit organizations. This
piece of evidence signified the extensiveness of the influential capability of nonprofit
groups that were formed by Chinese citizens themselves. The fact that these groups were
able to attract a consistently high number of connection initiatives from others in the
network suggested that “power” can emerge from dimensions other than the ones people

275
are most familiar and attached to, such as in terms of money, position, status, and tenure-
ship. “Power” can also be interpreted within the dimension of motivations and drives
which can be manifested through network behaviors observed over time. To put it in
another way, the sustainability of the high level of in-degree for these newly emerged
actors is a form of power that further builds up the capability formation process first
being initiated immediately after the earthquake. On the one hand, the capability
functionings in terms of having the availability channels in communicating with others
was driven by agency action revealed through out-degree measures. On the other hand,
the ability for actors to sustain and the possibly expand the capability functionings was
empowered by the acceptance and trust from others, which can be measured through in-
degree. Agency action and the formation of status of prominence work together to set the
stage for further expansion of actors’ capability set. This line of argument emphasizes on
the indirectness of the execution of power and a sense of sustainability from the
perspective of civil society actors.

276
“Late-comer” Emergence
In the third category of the table 4.13, we can see that the level of incoming
nominations of actor #12 got a tie with #125, #137, and #37. The last three actors also
received relatively higher levels of incoming connections during the emergency response
period. The difference between #12 and the rest of the actors in the whole category lies in
the fact that it was the only actor that came into being after the earthquake and was
organized by the initiative of a group of ordinary Chinese citizens in response to the
earthquake. It was established in May 2008, just three days after the Wenchuan
earthquake. From a group of temporary volunteers coming together to help the disaster
response, the members later engaged themselves in the long term recovery and
development of a region that was severely destructed by the earthquake. The actor itself
also became the first registered grassroots nonprofit organization in the entire region
where it was active. During the recovery stage, the actor participated in activities such as
helping women and children, psychological counseling, and the livelihood support for the
earthquake-impacted communities. Since the initiation of its operation, the actor attracted
ties from 5.1% of the network during the emergency response stage. And its nominations
doubled during the long term recovery with the percentage climbing up to 11.7% of the
remaining actors. Looking through its outgoing connection activities, the actor had only

277
one tie-initiating act during the emergency response stage. But over time, it expanded its
communication partners to 23.4% of others in the network. As others started to recognize
the actor and reached out towards it, its own agency activities also started to grow in
number. Unlike the trend for some of the other actors whose incoming ties and outgoing
ties experienced significant “jumps” immediately after the earthquake and then tended to
go down once into the longer term, the changes in measures of in-degree and out-degree
for actor #12 signified that it is possible for emerging grassroots groups to experience a
type of “lagged” growth especially in their tendency to launch agency activities. What
can be inferred is that on the one hand, a disaster event can trigger immediate grassroots
voluntary actions among citizens and the behaviors can be reflected through the focal
actors’ outreach activities and incoming nominations. On the other hand, the long term
recovery stage should not be ignored when considering the starting point of a type of
latent agency that signifies the emergence of sustainability for social structures, thus the
initiation of an institutionalization process.
Moving on to categories 4 and 5 in the table, we find that actor #123, the
nonprofit group established after the earthquake, occupied a leading position among all
others. Except actor #33, all others in category 4 had relatively high number of
nominations towards them in the previous emergency response period. In category 5,

278
actors #135, #76, #5, #71 availed in receiving more ties as compared to the period of
emergency response. Among these emerging new actors during recovery stage, actors
#135 and #76 were the only two actors who responded to the survey and thus had both
incoming and outgoing ties. Actor #135, a social worker station established by inter-
university collaboration initiatives between mainland China and Hong Kong, had its
outgoing ties going up from 27.7% to 45.3% during the two periods after the earthquake.
At the same time, its nominations increased from 4.4% to 9.5% for emergency and
recovery stages respectively. It indeed exhibited traits of an active agent in building new
connections rather than an “information sink”.
Like actors #3, #49, #123, actor #76 is another “home-grown” Chinese social
group established in response to the earthquake. It particularly aims to assist the
livelihood and capability-building of local ethnic minority women whose lives were
significantly affected by the disaster. Upon until 2011, the group had not been formally
registered. But its institutional status did not seem to affect the high level of its incoming
and outgoing ties through the periods after the earthquake. The actor was reached out to
by 5.1% of others in the network at the emergency response stage, and the percentage
expanded to 9.5% over the recovery stage. For its actions in establishing relationships to
others, the actor reached out to 9.5% of the remaining others during emergency response,

279
and the number jumped to 23.4% during the long term recovery period. Therefore, how
the others in the network behave towards this focal actor and how its own outreach
behavior changes are relatively similar to that of actor #12. The peak of their agency
initiative was “delayed” towards the long term recovery stage rather than immediately
after the disaster. And their agency actions were further empowered and supported by the
increasing acceptance from others over time.
The primary findings in investigating actor in-degree and out-degree network
measures indicated that the newly emerging grassroots groups, such as #3, #12, #123, and
#76, can show an increasing strength of their agency in communication outreach all the
way into the disaster recovery stage and the level of their network expansion can
experience upward “jumps” during the long term period as well. This contradicts two
types of popular claims regarding the role of civil society after a disaster. One type of
claim is that the disaster event will create a “window of opportunity” for actors who are
trying to make a change to the system, and once this short time frame of opportunity
passes by over time, not so much change can be made to be sustainable to create any
institutional change of the structure. From the results of the current analysis, I argue that
within the context of looking at the newly emerged Chinese civil society actors, the start
timing for creating opportunities for structural change in their social environments can

280
occur long after the significant event. Another type of claim states that after the initial
surge of the volunteerism immediately after the earthquake in China, the sustainability of
the grassroots efforts becomes questionable. The results analyzed in this section showed
that for certain Chinese grassroots actors, not only were they able to build their strength
by expanding their outreach network all the way into the long term recovery stage, their
capability of functioning were also actively being supported by others through
information-seeking behaviors toward the focal actors, especially into the later stages of
recovery.

Summary
With this, I conclude the discussion on the out-degree and in-degree measures
from the univariate statistical outputs for communication networks. I generally focused
on exploring the characteristics of the actors whose level of incoming and outgoing ties
ranked themselves relatively higher than others in the network.
From the analysis thus illustrated, two themes can be derived regarding change in
Chinese civil society after the Wenchuan earthquake. First, at the actor level, evidences
of the preceding action and behavioral conditions that lead to institutional formation

281
characterized by group/organization internal generations were found. They were
illustrated by the persistence and sustenance of agency actions of the newly emerged civil
society actors after the earthquake. These grassroots actors not only maintained to be
highly active in reaching out to others long term after the disaster event but also were
being consistently perceived to be important sources of information and sought out by
others at the same time. The persistence or duration of the outgoing and incoming ties of
these actors demonstrated their primary forms of institutionalization. Secondly, by going
through the categorization of the level of actor out-degree and in-degree activities
comparing before and after the earthquake event, I also illustrated descriptive changes in
actions and interactions occurred inside the civil society domain. Some preliminary
investigations into the behavioral changes of actors over time were being delivered and
the general trends in agency initiatives were being summarized based on comparisons of
degree measures across time periods. In the next section, I will go further into the
analysis to look at the more concrete measures of the structural dynamics within the civil
society domain.
One lesson to be learned in future network data collection procedures is the need
for clear specification of how an actor may be defined as a formal organization or an
informal social group. In the Chinese disaster recovery case, the “date of establishment”,

282
in certain incidences, was reported as the day the actors gained formal registration status.
But discrepancies occurred when actors in this category still named communication
partners even before the day of their official existence. This suggests that the original
questionnaire can be improved by defining the establishment day as when participants in
a group/organization first came to Sichuan and started working collectively without
having the registered status. This way, the data will be able to capture more nuanced
connections among actors and make more accurate descriptions on the origin of actions.

283
Chapter 5
Enduring Civil Society: Sustainability of Actions

(Part I)
Communication and Persistence of Agency Action
Cohesion
Structuration of Solidarity
In this Chapter, I focus on an in-depth analysis of the social structuration process
of communication and collaboration networks by comparing the periods before and after
the Wenchuan Earthquake. I began by discussing the immediate “neighborhood”
structures of the more active and influential actors. Then, I expand the horizon to
examine the overall inclusiveness of the two types of networks over time. Such
characteristics of the general cohesion were revealed in the density measures by
calculating the percentage of all the possible ties that are actually present. The patterns of
connection revealed that the social structures of the communication and collaboration

284
relationships were both being significantly altered by the agency orientation of actors
seeking connections with others shortly after the earthquake.
From a general examination conducted in the previous Chapter, as actors became
increasingly connected to one another through direct and indirect ties, both types of
networks became denser. Over the long term, the agency structure also illustrated signs of
being institutionalized with increased numbers of project collaboration ties. Comparing
across the two network types, the communication network did appear to have more
solidarity than the collaboration network both before and after the disaster. This means
that actors were being embedded in a more “tightly knit together” kind of inter-
relationships when sharing information and communicating with each other.
Contrastingly, the ties for collaboration were “sparsely knit together”. Such a difference
was expected because the required commitment for collaborating in projects operations
involves a deeper level of motivation, dedication, and capacity for actors’ professional
growth.
In the first part of this Chapter, I delve further into the network structures to
particularly investigate the emergence and development of communication networks. The
different ways of network embeddedness can represent both opportunities and constraints

285
for the focal actor. And one also needs to be aware of the different types of relationship
environments, such as the distinguished features of communication and collaboration
networks. In this section, I am primarily concerned with the communication or
information exchange relationships because of the likely changes in action intensity
immediately and in the long run after the disaster event. Compared to the structural
changes in collaboration, the communication networks experienced a full integration of
actor connectedness immediately after the earthquake and maintained the overall
connection through the longer term. (See figure 4.3A and 4.3B). Therefore, examining the
structural features of how such integrated-ness came into being becomes an important
step in understanding the development of Chinese civil society in times of a catastrophic
disaster.
A basic measure that I used to analyze actors’ connectedness considering the
network as a whole is called distance. It is a concept that embodies a variety of ways to
examine how one actor can reach out to another and represent the different channels that
information can flow among agents or mediums one can navigate through to find
communication partners.

286
Creation of Efficient Communication Channels
In network analysis, two nodes (or two actors) may be directly connected by a
tie/line or indirectly connected by ties/lines. The sequence of such lines connecting two
actors is called a “walk”. And a walk by which each node and each line are distinct is
called a “path”. So, the “length” of a path is defined by the number of lines or “steps” that
it takes to get from one actor to another. For network with directions like the ones in this
research context, there might be multiple information exchange paths from actor A to
actor B, but it is possible that B cannot reach A. Moreover, the length for these multiple
paths from one actor to another will vary. For example, as shown in figure 5.1.1, there
can be five distinct ways for actor D to reach actor A.

Figure 5.1.1. Illustration of Communication Channels
Actor A
9
Actor C
Actor D
Actor B

287
In graphic terms, these five paths from actor D to actor A are: DA at length 1, DBA at
length 2, DCA at length 2, DBCA at length 3, and DCBA at length 3. The geodesic
distance, however, is the length of the shortest path between A and D. And that is when
actor D reaches actor A directly. Therefore, the calculation of geodesic distance only
considers the length of the shortest path between two actors. Thus, I chose it to represent
the efficient communication channels that the focal actors established.
The efficiency consideration of the shortest path length among actors is
particularly important for understanding communication networks in the case of disaster
response and recovery. During the emergency response period, when there is a time
sensitivity factor that has to be taken into account, reaching out directly to the targeted
information source will most likely to be the most time and resource-saving strategy for
actors who initiate communication relationships. As the number of information sources
with shorter geodesic distance increases in the neighborhood of the focal actor, the more
opportunities the actor will have to either sending out information to others or reaching
out to receive the requested information from others. Starting with this understanding of
the basic concepts, I examine the network analysis results of geodesic distance counts for
the three time periods before and after the earthquake.

288
Pre-Earthquake
For the period before the earthquake shown in Table 5.1.1, the largest geodesic
distance found is at length 6.
Table 5.1.1. Efficient Communication Paths Distribution (Pre-earthquake)
Geodesic
Distance
Length 1 Length 2 Length 3 Length 4 Length 5 Length 6
Frequency 230 473 417 137 26 6
Proportion 0.178 0.367 0.324 0.106 0.020 0.005
This means that there were actors in the network whose shortest path connection
between them was at length 6. It will take 6 steps and 5 intermediary actors before
information can flow from one actor to another. Note that there were 6 incidences when
such extended distance appeared in the communication network. The majority of the
actors could reach out to the others within 2 to 3 path length, with one or two medium
actors in between. Looking at the
“proportion” figures, 36.7% of the geodesic distances were at length 2 and 32.4% were at
length 3. Those actors who can be connected by one path length appeared in 230
incidences, which counted as 17.8% of the total geodesic distance cases. Overall, the
communication network before the earthquake can be said to be rather extended and

289
sparse. Information cannot travel quickly enough as most of the shortest connections
between two actors were mediated by one or two other actors in the network. The more
the number of such intermediaries, the more time it will take for information to pass
through towards its targeted destination. The result also revealed that the network is not
fully connected as there were pairs of actors not “reachable” to each other through any
categories of path length.
In these disconnected cases, the geodesic distances were defined with a number
that is calculated by one length greater than the largest distance in the network, which is 7
in the pre-earthquake communication relationships. This type of disconnectedness
appeared across the entire network. One reason for such cases is that there were actors
not in existence at this stage of time and the analysis treated them as “isolates” being
disconnected from those in the connected network.

Emergency Response
Compared to the pre-earthquake period, the emergency response network became
more compacted as the largest geodesic distance found was at path length of 4, which is 2
length fewer than the earlier stage (see table 5.1.2).

290
Table 5.1.2. Efficient Communication Paths Distribution (Emergency Response)
Geodesic
Distance
Length 1 Length 2 Length 3 Length 4
Frequency 1028 3902 1544 107
Proportion 0.156 0.593 0.235 0.016
This is one piece of evidence because showing that actors at this time could reach out to
others faster, if needed, with less mediation of other actors in between. As a result, this
network did evolve in such a way that the overall structure facilitated the information
flow, which was actually needed for the short term response period. One point worth
noticing as the communication network evolved from the first stage to second stage was
the intensiveness of how new actors were being integrated into the whole network. Note
that as new groups and organizations emerged during the period of disaster response, the
network experienced “contraction” pulling these actors into further connectedness. From
the previous analysis in tracing the origin of such a process, such contraction force can be
enacted by the agency actions of civil society actors, particularly the newly emerged
ones.
The longer path length of 5 and 6 disappeared altogether during this period. This
is an important signal of the behavioral response on the part of civil society actors. The

291
tendency for structural cohesiveness and solidarity enacted by the agency initiative of
actors immediately after the disaster event overcame the dispersive tensions exhibited in
the period before the earthquake. The frequencies of each category of geodesic distances
also corresponded to the increasing density of the emergency network. The number of
geodesic distance at path length of 2 “jumped” from 473 to 3902, still being the
dominating connection distance during this stage. To put it proportionally, 59.3% of the
distances “traveled” for information sharing were of path length of 2. This means that the
majority of the information can be reached from the initiating actor to the targeted actor
with one facilitating actor in between. This process was accompanied by a decrease in
the percentage of path length at 3 from 32.4% to 23.5%. Finally, the quantity of direct
connections at length 1 increased nearly five-folds from 230 to 1028. But the proportion
went down from 17.8% to 15.6%. This was an indication that actors were still not quite
aware of the existence of others in the network and this was possible because of the mass
emergence of new actors over such a short period of time after the earthquake. However,
such dynamic was being compensated by the increasing incidences when actors were
being able to connect indirectly through one other “facilitator”.

292
Recovery
By the long term recovery period, the whole communication structure became
even more cohesive when the maximum path length went down to 3 (see table 5.1.3).
Table 5.1.3. Communication Efficient Paths Distribution (Long-term Recovery)
Geodesic
Distance
Length 1 Length 2 Length 3
Frequency 1193 4574 812
Proportion 0.181 0.695 0.123
The communication network became increasingly tightly knitted together and
information flowed faster between two actors than any other previous periods. The
quantity of the most efficient communication paths at both length 1 and length 2
increased further. The connections that needed one mediator between two actors went up
from 3902 to 4574, and almost 70% of all the lengths among the three categories of
distances fell into path length of 2. Over time, the agency actions did bring actors closer
to each other to exchange information, when the proportion of those who could be
connected directly increased from 15.6% during emergency response period to 18.1%.
Note that this figure was higher than that of the period before the earthquake. This
demonstrated that in the long run, the opportunities for actors directly reaching out to

293
others steadily climbed up. Overall, two factors provided primary evidences showing that
the communication structure after the earthquake not only became inclusive but also
denser. One is the decreased proportion of those connections that needed two
intermediaries. The other is the further increase in the percentage of direct connections.
The actors became more embedded in the direct relationships with one another and had
more efficient communication channels to pass information through to another.
From the whole network point of view, the table 5.1.4 below confirms the process
of the increasing solidarity of the communication network.
Table 5.1.4. Combined Geodesic Distance Results (Communication Network)
Average Distance Distance-based
Cohesion
Diameter
t1 2.437 0.034 6
t2 2.111 0.186 4
t3 1.942 0.198 3
First of all, as a general trend, we can see that the average distance among actors
decreased from 2.437 before the earthquake to 1.942 during the recovery stage. In general,
the communication network evolved in such a way that actors were more intertwined
with each other and information flow became more unobstructed by mediators. More
direct exchange of information and contacts were possible. As there were more

294
connections with shorter geodesic distances, the opportunities to reach out to others and
to be known by others increased. This further promoted a sense of solidarity among the
actors in the network and the measure of “distance-based cohesion” appeared in table
5.1.4 captured such feature of “compactness
53
”. During the period before the earthquake,
such cohesion of the communication network was 0.034. The value of this measure can
be between 0 (entirely disconnected network) to 1 (everyone is adjacent to each other).
Therefore, the larger values can be interpreted as the greater the cohesiveness of the
network. This is exhibited through the incremental changes from the period shortly after
the disaster to long term recovery, with measures went up to 0.186 and eventually to
0.198
54
.

Emerging Network Boundary
The increasing compactness of the communication networks can be further
elaborated through the concept of diameter, which is defined as the largest geodesic
distance in the connected network. The changes of this measure were illustrated in the
last column of table 5.1.4. Before the earthquake, the longest distance one actor can reach

53
In UCINET, it is calculated by the normalized sum of the reciprocal of all the distances.
54
For complete output results of Geodesic Distances of communication networks, refer to Appendix 5.1A, 5.1B, 5.1C.

295
towards the other was with length 6 and this means that information has to pass through 5
different “facilitating” actors in order to get to the targeted actor on the other end. If
having “efficiency” is interpreted by how much time and resource that can be saved by
going through a particular path from the sources of information to the receiver, then, a
network with this size of diameter will not be as facilitating for the speed of information
flow as those with smaller values of diameters. But for some actors, this path length was
the most efficient communication channel for them to connect with others. And the
existence of such extended efficient paths for the pre-earthquake period confirmed the
estrangement of the relationships among actors.
As the network became more “compacted” after the earthquake, the diameter
measure decreased from length of 6 to length of 4. During this period, actors were
particularly active in reaching out directly to others and as a result, the longest geodesic
distance one piece of information can go through in the network is no longer 6 but 4. The
number of mediums that one actor had to go through in these cases also decreased from 5
to 3. As groups and organizations actively chose to stay devoted in the field, the long
term recovery period continued building up the strength of actors’ connection as they
became more acquainted and familiar with each other. Communication channels were no
longer dependent on as many extended “routes” of intermediaries as that of the

296
emergency response period. More shortened efficient information exchange paths became
available to actors as their agency freedom was being activated by a drive to contribute to
the long term social development and building up a Chinese civil society at the group and
organization level.
This resulted in further decreased diameter from path length of 4 to path length of
3. There were 12.3% of the geodesic distances fell into this category at this stage and
pieces of information only needed to go through two medium actors. In other words,
within the communication network, no actor was more than 3 steps away from any other
actor. If efficiency of information transfer from one side of the network to the other is of
primary concern, then, an upper bound of three steps exhibited in the recovery stage will
be of advantage to that from the previous two periods.

Capability Set Formation
55

The geodesic distances and the diameter measures matter when considering the
efficiency of information flow across the network as a whole. However, when bringing
the focus down to each actor and looking at opportunities and constraints on an actor-by-

55
Please see Appendix 5.1D, 5.1E, 5.1F for sample UCINET Output of “Number of Geodesic paths” matrices.

297
actor basis, it is not the length of the distance but the total availability of the different
geodesic paths from one actor to another that renders importance. In other words, the
existence of the variety of ways of connecting is of significance in this case. The ability
of the focal actor to get pieces of information through the network essentially depended
on the availability of these different channels of communicating with others. I call these
paths the capability set for one actor to efficiently conduct information exchange in a
structural environment. Amartya Sen (1992) originally brought forth the idea of
“capability set” as a conceptual tool to reflect on agency freedom for a person to “choose
from possible livings” (40). Along this line of thinking, the capability set, in this research
context, can be defined as all of the various combinations of communication (or
collaboration) channels available to an actor (at the group/organizational level) for it to
function in a structural context understood in terms of different social network
environments.
Take figure 5.1.2 for example, if actor D did not know the existence of actor A in
the first place (when the arrow in color yellow was non-existent), there would be two
geodesic paths that it can take to reach or to get to know A.

298

Figure 5.1.2. Illustration of Efficient Communication Channels
One is the route to reach out to actor B and the other is to reach actor C as treating them
as mediators. Both directions can get information signals from actor A to actor D in two
steps. If the characteristics or traits of B and C are not taken into account, the two paths
can be considered equally efficient in connecting actors A and D. But civil society actors
in this research context were indeed different in their attributes and thus behavior of
embeddedness in the network. For example, actor B can be a grassroots organization in
existence before the earthquake while actor C being another newly emerged social group
like actor D during the emergency response period. It is therefore possible that actor B
could have developed a rather longer term relationship with actor A than that of actor C.
Actor A
9
Actor C
Actor D
Actor B

299
This will put actor B in a favorable position in improving the odds of passing the
information signals on to actor A. Therefore, from actor D’s point of view, it would be
better to have multiple such efficient paths in its immediate “neighborhood” so that its
“voice” can be easily and correctly heard by the intended targeted recipient.
The number of geodesic paths is implemented to perform such task of detecting
“efficient” paths in particular actor’s neighborhood. Looking across the output matrix for
the period before the earthquake shown in Appendix 5.1D, it can be concluded that
information channels were easily interrupted by observing the large number of cells
consist of zeros between pairs of actors. The occurrence of zeros in the matrix tables
meant that the corresponding pair of actors would not be able to reach each other and
information flow will easily break down. Also, most of the pairs of actors had only one or
two available shortest paths to connect with each other, with the exception of a few actors
who had higher number of alternatives. From the perspective of choice availability, or
capability set formation, this pre-earthquake communication structure imposed more
constraints on actors because in order to pass information efficiently on to the other, most
of actors were either disconnected from others or only had one shortest path to reach out.
The limitation on the actors’ part comes from the fact that their intentions in getting the
information flow quickly were stringently dependent on the action or characteristic of the

300
other actor. There were no alternative choices available for information to pass on
through the rest of the network if the current route did not turn out to be successful in
getting the information through. In other words, the alternative efficient ways for
connection among pairs of actors were extremely restricted.
The increased intensity of agency actions during the short term after the
earthquake changed the structural patterns toward opening up more alternative
opportunities for pairs of actors to be able to reach out one another. Observing across the
matrix output for this period of time (see Appendix 5.1E), not only there were less
disconnected pairs with zero shortest path length, the number of available alternative
efficient routes also increased for pairs of actors across the network. Take actor #134 for
example, the number of geodesic paths to reach out to certain other actors increased from
0 to 23 at the maximum. What this meant was that compared to the connections in the
previous period, information flow at this stage was much less likely to encounter
disruptions and more actor had the choice to choose from alternative efficient channels to
send out or receive information if one route did not work out.
The long term recovery period continued such a trend while the network as a
whole is becoming more compact (see Appendix 5.1F). Considering across before and

301
after the earthquake, the number of efficient communication paths for actor #123 toward
other civil society actors exemplified sustained incremental changes. As soon as it was
first established during the emergency response stage, the maximum number of efficient
paths for it to reach out to others was at level 5. The maximum efficient paths of this
actor increased to 32 for the recovery time, which made it having the most alternative
communication route toward another civil society actor at this period of time. Therefore,
it is now necessary to explore the source of its institutional emergence for this actor to
look for the motivations and origin of change in terms of its increased capability to pass
information through variety of channels.

The Case of Actor #123 (SG123)
Throughout his experience during the emergency response period, the organizer
of actor #123 (SG123-01) was provided an opportunity to work side-by-side with other
volunteers and get to know those who eventually became team members of his group. His
devotion to improving the psychological health for children suffered traumatic
experiences after the disaster further connected him to many professionals and academics
who worked in similar areas of specialization at local universities. Together, they formed

302
the initial structure of actor #123 as the proto-type of the later formally established social
group. As for the organizer himself, it had been a long way before he finally found
something, or more precisely, a career path, that he was enthusiastic and passionate about
doing. Before the earthquake event, his work was related to electric engineering and for
most of the time he would just conduct the activities in such a mechanical way that no
human contact needed to be involved. To him, that type of work was dry and repetitive.
After the disaster, having had the opportunity to meet with other volunteers to serve the
need of others, the organizer finally recognized that it was in the nonprofit field that
enthusiasm and devotion of work energy were triggered and maintained. “Right now, not
only can I help others but also started getting to know myself as a person at the same time,
which is significantly meaningful” (SG123-01). As he handed me his business card, he
specifically pointed at a few words representing the guiding principle of his work:
“Helping others means helping ourselves, awakening ourselves means awakening others”
(助人自助，自觉觉他). The term “awakening” in this context would represent a type of
awareness within oneself in having the capability to exercise a type of freedom to assist
others.
As for the group activities transitioning towards long term recovery phase, the
organizer emphasized the importance of adolescent psychological health condition in the

303
disaster impacted areas, particularly the role of schools and families in providing this
group of youth a healthy growth environment. However, one of the critical issues that the
local communities still faced at the time was re-establishing their livelihood after the
earthquake event. From the account of the organizer, at one time during the short term
response period, it became very popular for some nonprofit groups and organizations to
support the local people making and selling cross stitch products. But over time, as
people’s attention to the earthquake faded, the business itself also came to a halt as the
products themselves made by people from the disaster areas no longer held buyers’
attraction. This example illustrated the importance for nonprofit groups and organizations
conducting long term recovery activities rather than just rushing into the local
communities providing any kinds of assistances without closely examining and
understanding the types of activities that can assist sustainable livelihood support at the
local level.
One thing worth noticing from the way that the organizer described the works of
nonprofit groups and organizations was in terms how he distinguished the conceptual
relationships among “ordinary people”, “NGOs”, and the “government”. From his
perspective, the interaction is one that NGOs being standing out as a separate sector of

304
entities free from the influence from and towards the mass and the government. For
example:
NGOs cannot stand on the side of the government and speak for the government.
If conflicts arise between the government and the general public, and part of cause
is probably due to certain government policies. If government wants us NGOs to
stand out and persuade the public on behalf of the government, we will definitely
not do so. What I think the ideal situation would be one where we will have our
own way of thinking, and that will be independent of the state and the mass. We
won’t have to lean on any sides and at same time, can encourage different parties
to express their own point of views without representing any one of them.
(SG123-01-01
56
; summary of account of SG123-01)
From here, we can see the organizer gradually realized the emergence of the civil society
domain through distinguishing the associative activities coming out of the general mass
to form NGOs and their separated-ness with the domain of the state performed by various
branches of the government. Such a separation process is perceived to be characterized
by NGOs’ stand-alone positions particularly in its way of thinking. The description of the
“ideal situation” in the functioning of NGOs in the civil society domain represented the
organizer’s initial awareness and a further pursuit of a need for a “matured” and
“standing up” civil society that not only rose up from the ordinary mass but most
importantly is capable of developing its own “boundaries” of thinking and conduct

56
For original Chinese script please refer to Appendix 5.1.5.01. For details regarding actor #123 institutional status,
refer to SG123.01.02.

305
distinct from the “government”. Note that it was not a confrontational position being
advocated from the account of the organizer, rather, it was an outlook and desire for a
dynamic that can be called “the emergence of civil society” in the Chinese society
57
.
With the awakening of serving others toward a larger social cause, as well as an
increasing awareness of the distinguished role of civil society groups and NGOs existing
apart from the state and the general public, the actors’ choices of alternative channels in
general were no longer as limited in the longer term as its situations right after the
earthquake. If one option of reaching out to the other actor did not work as expected,
there were ways to by-pass these obstacles to move forward in reaching its goals. One
important point to make is that such a process cannot be disconnected with the fact that it
was the actors themselves by their own agency in building their immediate neighborhood
networks that contributed to the increasing opportunities for others at the same time.
Action does initiate a process of creating social structure. But the dynamics of the
structuration process in turn can provide further opportunities for actors interacting with
each other. Therefore, each one of actor’s actions in reaching out towards others matters,
especially in times of a crisis. The more actions taken, the denser the network becomes,

57
For related field notes and accounts, refer to Appendix 5.CaseSG123.2.

306
and the more efficient alternatives for actors to choose from in order to facilitate
information flow throughout the network.
The concept of “capability” in this case therefore, can be perceived as an agency
act to pursue connections and relationships despite of a crisis situation. It does not start
off with certain self-oriented interest for the focal actor’s economic or status gains, but
begins with a motivation to benefit the others. Once such motivation is brought forth
through a type of outreach action, the “capability set” (Sen, 1999), which represents the
availability of opportunities for actors to exercise their choices for relationship formation,
can then be measured by the level of solidarity that we have discussed throughout the last
section.

Strength Formation
I have used the measurement of geodesic distance to look at the possible ways for
information to flow efficiently throughout the network. However, the inherent
assumption was that alternative efficient paths for a particular actor to reach out to its
target were indeed implementable. By “implementable”, I mean the strength of the ties
between focal actors as well as the mediators on the efficient paths was “strong” enough

307
to get the pieces of information through towards its target destination. For emerging civil
society actors in particular, the types of information being communicated went beyond
emergency response activities. During the period transitioning to long-term recovery,
more of these actors were willing to get in touch with others who could be helpful for
them to develop their field expertise and grow from an informal social group towards a
formal nonprofit organization.
It is possible that even if there are multiple efficient pathways counted as geodesic
distances, a “weak connection” from the mediator to the target actor would not be able to
get the desired message through. One source of such “weaknesses” can be a lack of
familiarity between the mediator and the target actor, especially during the period when
many grassroots actors were emerging shortly after the disaster. In incidences like these,
the initiating actor might be better off by having established many other alternative
communication channels that did not necessarily count towards the efficient paths. There
might have been connections that could take up more steps than those on the geodesic
paths. But the more connections of all kinds—including both efficient paths and other
less efficient ones, the more likely that the initiating actor, particularly from the informal
social groups’ point of view, would be able to reach out to more pieces of information
and other types of assistances from others in the long term.

308
The network measure I use to take this factor into consideration is called
maximum flow. It takes one pair of actors at a time and considers to what extent the two
actors are being maximally connected not just by shortest path connections (efficient
communication channels) but also all the other available “routes” that are implementable
for one to reach out to the other (alternative communication channels). The general
premise of the “flow” approach as compared to the “geodesic distance” approach is the
former’s emphasis on the strength of weaknesses. This means that although it might take
several intermediaries for one to communicate with the other, the availability of these
indirect (“weak”) paths will be counted as an advantage as compared to the situation
when it is only through an efficient path that one can reach to the other.
The formal calculation was made by counting the number of different actors in
the neighborhood of the source actor that would lead to the target actor. The following
results are based on the examination of output matrix of maximum flow that depicted the
overall patterns of change.

309
General Strength Formation (Maximum Flow)
Before the earthquake, we can see that the alternative routes that flow across the
different pairs of actors were generally limited to one or two intermediaries
58
. This means
that pieces of information were not as easily getting through as those with multiple points
of intermediaries to direct alternative connection routes. The higher the number of actors
performed the roles as intermediaries between a pair of actors, the higher the likelihood
that the information would be passed from the source to the target. At a closer
observation, actor #51 had a relative advantage in getting its own message sent out to its
target because the number of flows from itself to the remaining actors in the network was
higher when compared to others. In some cases, it had more than five medium actors
lying in its local neighborhood that could facilitate the flow of information from the focal
actor to another. In the incidences when one route did not work smoothly in facilitating
the information flow, the focal actor had the choices so that it could redirect towards
other alternative connections to get its voices heard through these other possible
channels. Compared to those with only one medium across the pairs of actors, they would
encounter difficulties when this one available route was not able to serve its purpose in
realizing the information exchange. This type of connection pattern is more likely subject

58
Please see Appendix 5.1.5A1 for sample output.

310
to the changes in the outside influential factors such as actor attributes. This is a weak
and “vulnerable” connection because the likelihood of the focal actor getting to know its
target is dependent solely on this one intermediary, and if the communication tie between
intermediary actor to the target actor wasn’t realized, there would be no other pathways
to turn to for another try. In the pre-earthquake time period, not only more actors were
engaged in this type of vulnerable connection patterns, many pairs of actors were not
even reachable to each other. This was the result of the amount of isolated actors who
were disconnected from the network. Overall, during the time before the earthquake,
communications among actors were rather difficult to be achieved and many seemed to
be stuck in positions where the actual occurrences of communication ties being built were
entirely dependent on the existence of only one choice. This would be a weak social
structure that is normally vulnerable to disruptive changes as there is a lack of alternative
opportunities and the overall circumstance can result in easily broken connections. Actors
in this stage also tended to work in their own “circles” and there were no signs of
relationship-building initiatives that would provide opportunities for them to get to know
each other.
However, the actors’ behaviors encountered a significant wave of change
immediately after the disaster event in 2008. Examining across the maximum flow output

311
matrix
59
, one of the most noticeable changes was that there was a surge of those acting as
intermediaries in between pairs of actors. Not only the number of flows from one actor to
another increased drastically, actors across the entire network also established more
medium facilitators in their immediate neighborhood. Take actor #51 for example, the
highest number of flows for it to reach out to the remaining actors was 8 while the
majority of the rest of the flows stayed at the level of one or two during the period before
the earthquake. During the period of short term response, the highest number of
intermediaries jumped to 25 while the alternative routes in flows from itself to the other
actors in the network also experienced significant increase in numbers. Actor #3 is
another example. It was not in existence before the disaster. And right after the
earthquake, the actor built up its own connections in such a way that the alternative routes
available for it to communicate with its targets increased multiple-folds. This kind of
group behavior did not just happen among a few selected pairs of actors but became a
phenomenon across the entire flow pathways of the communication network. Similar
observations can also be illustrated with many other newly emerged groups during the
time of emergency response. The numbers and the trends reflected a sense of relationship
and connection-driven efforts by a motivation of voluntarism immediately after disaster

59
Please see Appendix 5.1.5A2 for sample output.

312
event. Compared to the weak and vulnerable connection situations before the earthquake,
actors in the emergency response period grew in their strength in their capability of
sheltering against the possibility of general communication break-down. With multiple
alternative connections to choose from and to fall back to when one choice did not work
out in getting the message communicated through, actors then would have more
opportunities to select another route of flow pathway for communication to occur. In
other words, the more intermediaries in existence in the neighborhood of one actor, the
more “safe-guarded” the strength of the connections, and the less likely that the tie-
building efforts be disrupted. The social structure at this period of time can be regarded as
more strengthening in terms of building up the opportunities available for pairs of actors
to establish communication relationships. Such a pattern was maintained into the long
term disaster recovery period
60
. Observing the maximum flows between pairs of actors
revealed that the general increase in availability of multiple communication channels
appeared to be endured in the long term. The endurance factor contributed to the strength
of “resilience” as new relationships were being stabilized and possibly institutionalized
through collaborative projects among actors.

60
Please see Appendix 5.1.5A3 for sample output.

313
Structural Formation from Action to Persistence
So far, my investigation in examining the communication network structures has
been tie-based. It is centered in the intensiveness and extendedness of relationships
among the actors themselves. I have used it to examine how actors were responsible in
building up the connections in each of their own immediate social neighborhood, how
that affects the connectedness of the structure of the whole network in terms of the
distances between pairs of actors. Action itself also changed the social structure beyond
the immediate neighborhood of the particular actor who initiated the act. The “solidarity”
and “cohesiveness” of the structural environment in its entireness can also be subject to
the dynamics of actors’ network behavior over time. As more relationships were being
built and more connections were weaved together across the network, the shapes of the
structure and how information pathways were embedded in it through both direct and
indirect relationships were all subject to change. In order to bring an in-depth picture of
the development of such structural environments, this section will look at their
differences across the various time stages. The focus is on the “structuration” side of the
picture, as the following section depicts the dynamics of such a process in the formation
of micro-structures from a “bottom-up” perspective. I particularly used the term
“embeddedness” throughout the later discussions. And in this research context, this

314
means that each civil society group or organization was playing an active role in being a
part forming the sub-structure within a given network structure.

Reciprocity
The density concept that we have discussed earlier emphasized on the mere
existence of ties or relationships among actors but did not specifically take into
consideration the direction of ties. From a structuration point of view, if actor A reaches
out to actor B to establish a communication relationship, then, the tie direction in the
network graph will be from A to B with an arrow pointing to actor B. This is only a
primary state of existence of a social structure, which is in the form of one tie. The stage
after this is that if actor B responded and also sends out a tie to actor A showing its
interests in the communication relationship, then, the connection between A and B is
called a “reciprocated” tie. Rather than looking at the picture from an ego-centered point
of view, reciprocation brings in the other communication partner into the process. A
reciprocated tie signals the interest of both parties and the beginning of a development of
trust on both sides of the relationship. The foundation of a social structure builds on top
of this type of ties. Without reciprocation between pairs of actors, information will not be

315
able to pass along across the network. Moreover, the strength of such a bond forged
through a tie will be less likely to develop further if one of the parties is not responding to
the other’s relationship-building initiative. A network structure with more reciprocated
ties will tend to be more “stable” than the one with more un-reciprocated ones
(Hanneman and Riddle 2005). From a long term perspective, this is similar to the
argument of “sustainability” and “endurance” of ties. If the two actors involved do not
even have shared interests and trusts that could promote their actions in reciprocation, it
is also less likely that the tie between them will withstand the test of time. Since one main
goal for this study is to depict an explicit process from which the structuration, or in other
words, institutionalization of a sustainable social environment can occur despite of a time
of extreme uncertainty, the starting point of examining the reciprocity measures will be
necessary.
One type of reciprocity measure is calculated by the dyad method. It focuses on
the proportion of pairs of actors that have reciprocated ties between them. The degree of
reciprocity is calculated among the pairs of actors that have any connection in the first
place. The reason that I chose this particular measure was that the basis of calculation is
concentrated in the actors themselves rather than on the ties. This was consistent with the

316
essential aim of this research, which was putting a primary emphasis on the capability of
actors’ agency in initiating change
61
.
Looking at the communication network results for the period before the
earthquake (see table 5.1.5), we find that 9% of pairs have reciprocated connection
among all pairs of actors that have any connection.
Table 5.1.5. Reciprocity Measures (Communication Network)
Time period Hybrid
62
Reciprocity (Communication)
t1 0.09
t2 0.0925
t3 0.1067
Essentially, not nearly one out of ten pairs of actors that had connections was
reciprocated. In other words, the actors at this time were very much “content” with the
state of existence as “closed” functioning entities without much connection to others.
When a process of institutionalization is defined to start off with building up reciprocal
relationships among groups and organizations, the communication network at this stage

61
Also needed to be taken into consideration was the inclusion of the non-responsive actors. The calculation of
reciprocity did treat the ties that were being initiated toward them as non-reciprocate consistently throughout the three
time stages. I defined that the measures were comparable across these periods by understanding the concept as the
prevalence of reciprocity in a given network environment (Hanneman and Riddle 2005).
62
When the data is not partitioned according to certain groups according to pre-defined attributes, this is the same as
the dyad method.

317
was far from having the signs towards being institutionalized across the pairs of actors.
The majority of the pairs of connections were based on the direct action of one of the
actors rather than two-way connections. It could also be inferred that there were not
enough interests or much trust among actors and there seemed to be no need for mutually
sharing of information. In this case, some of the actors were eager to communicate and
build information sharing ties with others but without much response from the other side.
For the emergency response stage after the earthquake, the measure went up to
9.25%, which is a 0.25% increase from the previous period. This means that although
slowly, more pairs of actors were having reciprocated connections. The noticeable point
here was the condition under which such a growth occurred. Recall that at this time, the
network became increasingly inclusive as more actors emerged to establish themselves in
the field. And this dynamic was also accompanied by an increasing connectivity of those
who were already in existence in the network before the disaster. An increase in
reciprocity for a growing network represented a surge of “openness” in accepting others
and willingness for actors to be connected and known by others at the same time. As the
actors became more embedded in the network structure over the long term recovery
period, reciprocity measure kept its gradual increase to a level that 10.67% of the pairs of
actors with ties have reciprocated connections. This showed that during the recovery

318
period, relationships among actors beginning to show preliminary signs of
institutionalization in terms of the degree of reciprocation in the network. For example,
when actor A reaches out to actor B without B reciprocating in the form of
communication, the relationship is unbalanced because the amount of information is
passed along across the structure are tilted towards similar sources and there are few
channels of facilitation pathways that information can flow in both directions. As a result,
certain actors will tend to withhold information from the rest of the network or act alone.
In cases when a disaster strikes, such a network structure will be more prone to be
dysfunctional as ties are more one-directional with the other party not getting sufficient
feedback or needed information to carry on further actions. A balanced relationship, on
the other hand, features both actors accept and trust each other and are both willing to
engage in sharing pieces of information. Such a reciprocated network provides a
“nurturing environment” for each pair of actors to receive support and develop a
stabilized connection with each other over time.

319
Transitivity
Relationships can show signs of institutionalization through the reciprocation
process between one pair of actors. This is the most primary stage. The next level of
examining how institutions of a structure can be formed is through the analysis of the
network closure activities, or the creation of “triads”. A triad involves three actors, such
as one among actor A, B, and C, or actor B, C, and D illustrated in figure 5.1.2 discussed
earlier. This type of basic structure among actors in this study is important to investigate
because it looks at situations of how communication network grew and expand at the
micro level as “friends of friends became friends”.
There are two main types of triads in network analysis that are critical when
relating to the context of this study. One is the called the “transitive triplet” and the other
called “three-cycle”. In this paragraph, I will briefly compare and contrast the technical
nature of these two triadic relations and the later paragraphs will be following up with a
discussion on the outcome applications to the communication networks in this study.
First of all, figure 5.1.3 below provided a visual presentation of the two types of triads.

320

Figure 5.1.3. Triad Representations as in Transitive Triplet and Three-cycle
Relationships
Source: (Snijders et al. 2010, 11)
Basically, both of these social structures represent the particular ways of expressing
network closure. One other similarity between the two is the existence of a “friend of
friend” relationship from actor i to actor h through j:
 
i j h 
. Starting from here,
it is the direction of the tie from i to h that essentially differentiates of what constitute as a
transitive triplet structure and a three-cycle structure among the three actors. For
transitive triplet network closure structure is one when given the two-path of:
 
i j h 
, it is actor i then reach out to h to close the tie structure (
ih 
). Note that
there is one other possible way to examine this structure. It is also true that transitive
triplets can take effect when there is:
 
; i h j i j   
, not just in circumstances like:
 
; i j h i h   
. Essentially, if actor i is the focal point of interest, the transitivity of
actor i can be calculated by counting the number of pairs j, h such that there is the
transitive triplet structure of Figure 5.1.3a.

321
For three-cycle network closure structure on the other hand, the two-path
 
i j h 
is closed by the tie
hi 
, as is shown in Figure 5.1.3b. Here, the critical
difference from the transitive triplet mentioned earlier is that the way triadic structure
closes itself. When actor h initiates a tie towards i, the network structure becomes a
complete cycle with each of the three parties willingly participating in the general
exchange of information or resources. In the network analysis literature, this type of
structure can also be interpreted as “generalized reciprocity”. Looking at Figure 5.1.3a,
we can see that actor j is the medium between i and h. But in the transitive triplet
situation, actor i not only initiates a tie to j but also reached out to h. The fact that actor h
can but did not close the relationship by initiating communication with i proactively
shows that actor h probably is more prominent possibly in terms of information or other
types of resources that might be helpful from actor i’s point of view. Further comparing
the two structures in terms of information exchange and communication type of
relationships, information can spread evenly or in a more balanced way for three-cycle
triad while the transitive triplet triad tends to create a dynamic where information flow is
concentrated in favor of certain actors but not others. While this is logical from the
theoretical network analysis perspective, as in the later sections when examining these
structures in the context of this study, I argue that for rapidly growing social structures

322
such as the communication and collaboration networks after the earthquake disaster in
China, transitive triplet triadic relationships can also be a representation of active agency
on the part of focal actors out of their desire to build ties at a given point of time and
context.
As defined earlier, a triadic structure is considered being “transitive” when it
depicts a type of relationship like that in Figure 5.1.3a. The primary formula being used
to calculate the transitive ties is by counting the number of times that
ih 
occurs, if we
see
ij 
and
jh 
. Table 5.1.6 provided a direct summary of the different transitivity
measures that will be used to take an in-depth look at the formation of the triadic type of
institutions among actors in the network.
Table 5.1.6. Triadic Relationships in Communication Networks (Transitivity
Measures)

  ,, AB BC AC

 
,, AB BC anything

Transitivity
t1 283 902 31.37%
t2 4328 12458 34.74%
t3 6101 16067 37.97%

323
For the period before the earthquake, there were 283 transitive triples where, if AB and
BC present, then AC is also present. The second column of the table provided a way of
seeing triadic relationships that is less constraining than the one used in the first column.
It relaxes the direction within which A must send a tie to C, given the condition that AB
and BC present. In this case, as long as there is a tie between A and C to close the triadic
relationship among the three actors, whether the link direction is AC or CA is not taken
into consideration. At this stage, there were a total of 902 this kind of triads in the
communication network. When dividing the number of
  ,, AB BC AC
by the number
of
 
,, AB BC anything
, we get the “transitivity” measure illustrated in the last column.
Therefore, 31.37% of all the communication relationships that could be transitive,
actually were. For the time stage before the disaster, from all the cases that provided
opportunities for actor A to reach to actor C, 31.37% of the actual relations turned out to
be A initiating a tie towards C. For the short term emergency response stage, the sheer
quantity of the number of
  ,, AB BC AC
relationships jumped to 4328, which were
more than 15 folds of the measure from the previous period. At the same time, the
 
,, AB BC anything
relationship increased in quantity even more from its earlier
measures. The transitivity thus increased to 34.74%, meaning that of all the possible
incidences where ties can be transitive, this percentage of relations actually came into

324
being. Both the number of transitive triples and the transitivity measure increased steadily
in the recovery period. There were a total of 6101 transitive triads at this last stage and
they counted 37.97% out of all cases that could easily be transitive with one link to
complete the triad.
To take a closer look at how the network structure in terms of transitivity has
changed before and after the earthquake, I now examine the first two columns of table
5.1.6 in an in-depth manner.
First of all, the number of cases, where a single tie between A and C regardless of
directions could complete a triadic relationship, experienced a 92.75% increase from the
period before the disaster to shortly afterwards. And this type of triadic case had a
22.46% increase when it was into the recovery period. This meant that as a result of
agency action in establishing relationships on the part of each actor’s out-degree
connection, the opportunities for actors to get connected through mutual acquaintances
increased dramatically. The key point here is that actors were not presented with these
opportunities from an “outside” agent by remaining inactive, but such was the result of
actors themselves being proactive in putting their ambition and desire into practice
particularly during the emergency response period after the disaster. Now faced with

325
further possibilities to get to communicate with others through intermediate actors, such
as cases with ties from AB and to BC, actors such as A were also willing to reach out to
those like C to close the triadic relationship among the three actors. From the measures
presented in the first column, the actual cases of transitive triplets increased by 93.46%
from before the earthquake to emergency response stage. And it was followed by a
29.06% increase afterwards when going into the long term disaster recovery period.
Transitivity structures such as these have generally been regarded as an indicator that the
network has a tendency for hierarchical structural relationships. As was discussed earlier,
this is due to the fact that there are certain actors when embedded in a triadic relationship,
such as actor A in our example, are more likely to reach out more to initiate ties with the
“friend of a friend” and passing information towards the other party.
To some extent, this is a reasonable interpretation of the network because when
understanding from the perspective of communication and information exchange, if one
party (such as actor A) kept being the one sending out signals to the other two actors in
cases such as:
 
,, AB BC anything
, it is possible that actor C could have more prestige,
“fame”, or “power” in terms of their field of expertise that others might be more willing
to connect with. However, within the context of this research, I would like to argue that

326
this type of transitive structure can also be viewed as a way to represent an act of agency
to deal with the disastrous impacts and changes in times of crisis.
When interpreting the meanings of the network measures to understand the
structures and actions of civil society in this study, it is always helpful to keep in mind
the particular context from which the social structures at hand had emerged. As the
response to the disaster triggered the act of voluntarism among civil society actors, it also
helped these actors to realize their functioning opportunities to reach out to others. This
could be a sign of actively recognizing their independent role as initiators in
communicating and relating to others who share similar values. On the part of those
actors who were like actor A, especially if they were newly emerged groups and were
also “new” to the civil society domain, the dramatic increase in the number of cases for
them to be the ones to close a transitive triad relationship could have another important
layer of interpretation. It signified a type of behavioral choice available when these actors
perceived to have more opportunities presented as cases of
 
,, AB BC anything
.
In other words, the structure itself provided a certain kind of freedom of
functioning for action and once these functionings were perceived as beneficial for
relational growth or communication-building on the part of civil society actors, they

327
would act it out by initiating ties towards others through mediums. The synergy or
motives behind the decisions to make such choices could, I argue, be captured by
examining the dynamics of transitivity structures. The measures of the cases where actor
A was willing to reach out to actor C rather than wait reactively for C to reach itself, were
key indicators that civil society actors in this context valued a spirit of agency to take
matters into their own hands. For times after a catastrophic disaster when both the
physical and social systems were shaken and damaged, this type of behavior was more
valuable for newly emerged social actors and getting them linked into the network.
Therefore, as the other side of the “hierarchical tendency” perspective, the emergence of
transitive triple structures could offer a relatively new picture to understand the society
side of the story.
In the field of planning and policy-making, the perspectives of the society are
often being assumed or even overlooked in the decision-making process. This is mainly
because of the difficulties in making the actions taken by the civil society actors
explicitly through which their behaviors and motivations can be revealed. The
interpretation process of the transitivity measures in this section can function as one of
the steps for state actors to begin to understand civil society through its actions. For
network analysis in general, the lesson here is that close attention needs to be paid to the

328
context within which the measures are represented. Just as in this study, switching the
lenses from looking at the structure through a top-down perspective to a bottom up
perspective--by allowing the stories of the civil society actors to reveal itself over time,
the other sides of the implications are equally important for policy-making.
One last point that I would like to make here is regarding the paradox of
institutional structure and actions of change mentioned in the theoretical section. In the
discussion of transitivity, one needs to recognize that the very first stage of institutional-
building process in this case originated from the agency actions on the part of civil
society actors. As each actor extended its immediate communication relationships with
the remaining actors in the network, the number of intermediary increases and thus the
cases of triadic incidences that actors would face. The kinds of relations that could easily
be triadic, such as the structure like
 
,, AB BC anything
, thus provided the embedded
actors two choices to complete the triadic closure. One is actor A reached out to C
generating a complete transitive triad and the other option is C reached out to A
generating a three-cycle relationship. From a network analysis perspective, the structure
itself does postulate a certain degree of local “constraints” on the action choices of actors
A and C in terms of creating triadic relationships among one another. However, if
looking at the situation from the actors’ perspective, especially the civil society actors in

329
the disaster response and recovery context, the existence of such cases where a single link
could complete the creation of a transitive triplet can be perceived as an opportunity to
build further connections and make themselves known. In particular, when an actor was a
newly formed grassroots group in response to the earthquake, it would be more willingly
to seize such an opportunity to make the first move in completing the triadic closure. In
the Chinese disaster recovery context, with the mass emergence of Sichuan-based
grassroots social groups and the coming-together of those nonprofit organizations from
across the country in Sichuan Province shortly after the earthquake, the possibility of
each one of these civil society actors knowing the other well was relatively low. The
relationships among them, for most of the cases, had to be built from scratch. During the
emergency response period, when responding to the needs of the impacted areas in time
became the top priority, the existence of transitive triplet triads represented one way of
how civil society actors could proactively deal with change to alleviate the disastrous
impact of the catastrophe. The more they initiated connections with others, the more
useful information they would be able to obtain, and thus the higher chances that they
would provide assistance to locations that needed the most. From this line of logic, I
argue that the transitive triplet triadic relationships could be one of the representative
social structures to examine disaster response and recovery efforts on the part of civil

330
society. Note that this is a derived position taking into consideration of the civil society
actors’ point of view of the structure. And when such a lens is equipped to examine the
transitivity measures, alternative explanations of the existence of the structure are
possible in order to advance the theories in understanding network structures in different
types of contexts.

Clustering
So far, I examined some of the basic embedding structures that actors can be
institutionalized at their local environment. Two of the most important ones are
reciprocal and triadic relationships. The emergence process of institutional structures for
these two types of connections can be specified and understood through the examination
of reciprocity and transitivity measures. A third concept that is critical on this journey to
unravel the growth process of Chinese civil society after the crisis is called clustering. It
is a measure that considers the immediate neighborhoods of actors, which includes all the
actors that the focal actor is connected to. Essentially, it is measured by calculating the
density of every actor’s local neighborhood. The size of each of the actor’s immediate
neighborhood can be a useful indicator in taking a closer look at the actor-by-actor level

331
of agency activity through communication networks and the level of motivation in being
committed through collaboration networks. The measures can be informative in
describing the tendency of each of the structures towards dense local neighborhoods.
In table 5.1.7, three categories of measures for clustering in communication
networks are listed here for investigation. Because of the high level of integration and
compactness of the information exchange network after the earthquake, I use the
communication network as a representation of the motivational drives for actors to
actively move toward each other.
Table 5.1.7. Clustering Coefficient (Communication Network)-Motivational
Network
Overall graph
clustering coefficient
Weighted overall
graph clustering
coefficient
Overall Density
t1 0.341 0.129 0.0122
t2 0.455 0.139 0.0544
t3 0.496 0.167 0.0631
First, the overall graph clustering coefficient is the average density of all the
neighborhoods of all the actors. Secondly, the weighted version is shown in the second
column. It is the coefficient actually takes into account of the differing sizes of the
neighborhood densities and the average density in this case is calculated proportional to

332
the sizes of the neighborhoods. The third column shows the overall density of the entire
network graph. The reason to bring back the overall density measures here is to establish
a point of comparison when making a statement regarding the level of clustering at
different time periods.
Let’s start with the communication network clustering before the earthquake. The
overall coefficient is 0.341, which is much higher than the overall density at this period
of time (0.0122). However, after weighing across the sizes of neighborhoods, the
coefficient measure became much less dense (0.129), but still higher than the overall
density. This means that there were certain degrees of clustering in different actors’
immediate neighborhoods.
In order to depict the disparities of clustering among actors, I further examined
the node level clustering coefficients. Each actor’s level of embeddedness was derived
from two measures. One of them represented the size of an actor’s local neighborhood by
examining all the other actors that have direct connections with the focal actor. This was
calculated by counting the number of pairs of actors inside the immediate neighborhood
of the focal actor. The other one represented the percentage of all the possible ties among

333
these neighbors were actually present. The higher this percentage, the higher the level of
neighborhood clustering within which the focal actor would be embedded in.
Let me use a graphic example to illustrate a process of clustering formation in
detail. The graph below (Figure 5.1.4) shows the communication structure before the
earthquake event.

Figure 5.1.4. Communication Network with Selected Actor Traits (Pre-
earthquake
63
)
Actor #115, visually, seemed to be engaged in a “radiation-like” local
neighborhood, which means that it sent out many one directional ties without getting

63
For graphs corresponding to emergency and recovery periods, please see Appendix 5.1.3.
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
*Color: establishment
Yellow: before EQ; Blue: after EQ
*Shape: location
Circle: Sichuan based; Square: Non-Sichuan
*Size: Registration status
Small: registered; Large: non-registered

334
many reciprocations from others. Nor did the actors within its immediate neighborhood
were well connected with each other. For some of them, such as actor # 136, #90, #92,
#113, and #124, their ties with actor #115 were the only communication channels they
had during the period before the disaster. The information pathways for these actors were
very much constrained by and dependent on the actions of #115. Given this immediate
set of neighborhood conditions, what the node level coefficient first calculated was the
number of pairs of neighbors or the possible ties among these pairs of actors. For actor
115’s neighborhood, there were a total of 153 possible pairs of actors, hence ties that
could exist among them. Compared with some of the other actors in the network, this was
a relatively large cluster if all of these ties were realized. However, only 3.3% of all these
possible ties were actually realized at this period of time. The conclusion we can draw for
actor #115 is that it was not embedded in a highly clustered neighborhood before the
earthquake. Graphically, this could be demonstrated by looking at a very loosely
connected neighborhood for the focal actor.
Compare the situation of #115 with actor #119, the latter exemplifies a quite
different picture. First of all, there were 136 pairs of possible ties in its immediate
neighborhood, which was comparable to that of actor #115. But this time, 18.8% of all
these were actualized for #119. The actors in its local neighborhood were well connected

335
to each other. For communication networks, one of the advantages of being embedded in
a relatively higher clustered neighborhood is that the information exchange pathways will
be able to flow more “smoothly” from one actor to another. From the focal actor’s point
of view, embeddedness in a highly clustered local neighborhood will also open up the
opportunities for it to reach out for information and get itself known by more actors in the
network.
Observing the two measures across all the actors, it can be concluded that the
largest size of neighborhood that one actor had during the period before the earthquake
was surrounding actor #51. Given the number of actors that it directly connected to, the
total number of pairs of neighbors that were possible for this actor to be embedded in is
703. The potential to be well connected in its local neighborhood was high. But only
5.2% of all the possible ties were actually present. Actor #61 had the next largest size of
immediate neighborhood consisting of 210 possible pairs of actors and out of these,
10.7% of the ties were actually being realized. The largest node level cluster coefficient is
at 100%. However, the actors whose local neighborhood has a level 1.0 clustering were
only connected to one pair of neighbors. The presence of a tie between these two
neighbors generated the most basic triadic relationship among three actors. But this does
not mean that the focal actor was embedded in a highly clustered neighborhood because it

336
did not have a large number of direct connections to others to form a dense enough
clustered neighborhood. In general, this communication network is relatively less
clustered. In other words, actors had few chances to become embedded in highly
clustered local neighborhoods because of the low density of the overall network. Civil
society actors in existence at this stage of time were not so well acquainted with each
other. Even though an actor had higher level of opportunities to be embedded in
neighborhood resulting in high clustering, the lack of agency action further reduced such
possibilities to be realized. This resulted in the communication channels generally being
sparse for the period before the earthquake.
As the overall communication network became more clustered shortly after the
earthquake, the number of possible pairs of neighbors surrounding actor #115 decreased
from 153 to 78. However, among these 78 ties among its neighbors, 35.3% were actually
in presence. While the number of pairs of neighbors for this actor stayed the same over
the long term recovery stage, the realization of these ties went up to 42.9%. This was an
indication that the actor gradually became more woven into the connections among its
neighbors, thus making the flow of information within this particular cluster easier.
Observing the measures of the entire network for the two periods after the earthquake,
actors were generally embedded in higher clustered neighborhoods over time.

337
There are two lessons to be learned from the civil society actors’ perspective in
this part of the investigation. One is that it is important to recognize the agency action on
the part of each one of the focal actor to actively seek out the opportunities to establish
initial ties with the remaining others in the network. This type of behavior yielded the
following results. On the one hand, action itself helped building large local neighborhood
for a particular focal actor. On the other hand, the tie possibilities among the actors in the
immediate neighborhood, whether actually presented or not, could be seen as one form of
“capability set” that the focal actor could choose to utilize for information exchange or
collaboration purposes. This leads us to the second lesson. The realization of such
“capability sets” for each focal actor also depended on the actual tie formation among the
neighborhood actors. Such interdependency became another illustration of the
relationship between action and structure. When institutions were understood as the
enduring connections among actors in consideration, the institutionalization of ties can
also be a process generating a breeding ground for change. In an information exchange
network environment, a position that the focal actor is embedded in a highly clustered
neighborhood is likely to be desirable because such a structure facilitates information
flow through the available different channels of pathways. Once these possible pathways

338
were being built and endured over time, this opened up the opportunity for the focal actor
to communicate with as many “friends of friends” in its local neighborhood.
For the period before the earthquake, there was still certain degree of clustering in
the communication network because the weighted overall clustering coefficient is about
10% higher than the overall density measure. Actors could have large size of immediate
neighborhood but very low node level clustering coefficient. Very few of the possible ties
between pairs of neighbors yielded into actual connections. There were also numerous
pockets of empty pairs for a number of actors throughout the calculation of their
clustering coefficient. This was either due to the actor in existence being disconnected
from the network or due to the non-existence of the actors at this stage of time.
During the period immediately after the earthquake, the overall degree of
clustering increased from 0.341 to 0.455. After taking into account of the proportions of
different neighborhood sizes, the weighted overall clustering coefficient increased from
0.129 to 0.139. This weighted coefficient was still higher than the overall density
measure, which also jumped from 0.0122 to 0.0544. Reflected in the node level
clustering coefficient measures, one can observe a general trend in terms of more actors
being embedded in highly clustered local neighborhoods. As more previously isolated or

339
non-existent actors became actively engaged in the emergency response network, the
pockets of “emptiness” in pairs of neighbors for these actors generally disappeared.
Compared to the previous period, civil society actors were more actively engaged in
initiating ties and hence, increasing the creation of more relatively highly clustered
neighborhoods available for each individual actor across the network.

Registration Status Group-external and Group-internal Ties
So far, I have examined the embedding processes of actors, each as a single node
immersed within their communication networks. It is now useful to conduct a preliminary
investigation to see whether one of the actor attributes affect the embedding structures. In
network analysis, each category within a particular actor attribute is named as one group.
The number of categories that an attribute differentiates is equal to the number of groups
that will be investigated at each stage. In this section, I focused on actor registration
status. For this attribute, two general groups were defined to represent either registered or
unregistered status. The original impetus for making such a differentiation on actors’
status was that in the Chinese context, registration for nonprofit groups and organizations
often faced with issues of getting the formal institutionalized status with both the

340
Ministry of Civil Affairs and sponsorships with local governments. The level of
interaction between informal groups and formally registered nonprofit organizations is a
critical component in looking at the nature of information flow and projects collaboration
among actors. From a disaster response and recovery policy design point of view, it is
more desirable for non-registered actors and registered actors to reach out for each other
as the latter are usually more experienced ones in the field and can particularly provide
information or resource assistances for the former to develop in the long term. More
cross-groups ties can also signify a sense of openness and embracing willingness to help
each other out after the earthquake.
In order to evaluate this social structure by looking closer into the group “fabrics”
within which actors were embedded, I used a measure called the E-I Index. The external-
internal index (E-I Index) is one way in network analysis to examine the connections that
are made inside a group and across different groups. The measure is calculated first by
subtracting the number of ties group members made inside the group from the number of
ties of group members made towards outsiders. The result from this subtraction will then
be divided by the total number of ties. Therefore, the measure for the index will be
between -1, with all connections being made inside of the group, and +1, with all
connections being made outside the group. A point to note here is that the directions of

341
the ties do not matter in the calculation of this measure. As long as there is a connection
made between two actors, regardless of who initiated it, the relationship will be counted
as one tie. With these definitions in hand, I will now explore the different aspects of the
E-I Index for registration status and location attributes from the current dataset.

Pre-Earthquake
I defined two groups in the category of registration attribute. Group 1 represented
the registered actors and group 2 represented the un-registered actors. The results listed in
the table below (see table 5.1.8) provided a general picture of actor activities within and
between groups. The density for communications made within the registered nonprofit
organizations was 0.017 and 0.027 for information exchanges made among the non-
registered actors. General communication across the registered and nonregistered groups
was 0.031, which was slightly higher than within group densities.

342
Table 5.1.8. Communication Network Pre-Earthquake Within-Group and Cross-
Group Density
64
Measures based on Registration Actor Attribute
Group 1 (registered) Group 2 (non-registered)
Group 1 (registered) 0.017 0.031
Group 2 (non-registered) 0.031 0.027
Therefore, more ties were made among actors who belonged to the un-registered group.
Off-diagonally, it seemed that out-group ties were more prevalent than in-group ties.
Overall, during the period before the earthquake, the in-group connections made by
registered civil society actors are not as many as those made among non-registered actors.
The actors also made more cross-group ties rather than constraining themselves to
connecting to only those who have similar registration traits. This was a preliminary
indicator showing that registration status did not seem to be a barrier for civil society
actors to develop communication relationships among each other even before the
earthquake event in 2008.
From Table 5.1.9, we can observe the in-group and between-group ties for the
network as a whole. The total number of internal ties being made, regardless of group
numbers, amounts to 228 and that was 54% of all ties existing in the network. The

64
These are block densities measures. The ratios off the diagonal represent out-group densities. The ratios on the main
diagonal represent in-group densities.

343
number of external connections was slightly smaller than that of the internal ties, and this
amounted to 194 ties (46%). In general, the behavior of civil society actors did not
exemplify a distinguished bounded-ness structure of sub-population, when considering in
terms of registration status at this period of time.
Table 5.1.9. Communication Network Pre-Earthquake Whole Network Results of
Group Internal and Group External Ties Based on Registration Actor Attribute
Frequency Percentage Possible Density
65

Group Internal Ties
228.000 0.540 12584.000 0.018
Group External Ties
194.000 0.460 6322.000 0.031
Continuing further examination of results from the whole network perspective, we
can see that the rescaled value of the E-I index for the communication network before the
earthquake is negative 0.081. Since this measure takes into account of the group sizes and
the density of the graph, the re-scaled measure is taken to be more reliable as the primary
indicator for group embeddedness. Also recall that a negative 1 index measure represents
that all ties are internal to the group. An index value of -0.081 suggested that there is a
very weak tendency towards group closure.

65
The overall density measure presented here are calculated as the ratio of the number of group internal ties (or
external ties) that are present divided by the number of pairs—all possible dyadic connections.

344
The next level of analysis was the group-level E-I Index. Table 5.1.10 illustrated
the variations across groups in terms of their degrees of closure.
Table 5.1.10. Communication Network Pre-Earthquake Group level E-I Index
Based on Registration Actor Attribute
Internal External Total E-I
Group1
(registered)
206.000 97.000 303.000 -0.360
Group2
(non-registered)
22.000 97.000 119.000 0.630
First of all, actors in the registered group made a total number of 206 group internal ties
and 97 group external ties. The non-registered actors made significantly less number of
internal ties (22). But at the same time, these same actors made exactly the same number
of outside-group information exchange connections (97) as that of its registered
counterparts. What this suggests is that the non-registered actors, or informal grassroots
groups, tended to build relationships across institutional boundaries during the period
before the earthquake. Although actors in the registered group also made the same
amount of external ties, compared to their efforts in making connections inside the group,
they showed a quite strong tendency towards group closure. The structure overall, when
thinking in terms of the roles of sub-groups, the informal grassroots appeared to be more

345
likely to communicate outside their group boundary than their registered counterparts
did
66
.
In order to look closely at the actor-level of connections, I developed a way in
understanding the group embeddedness by examining the variability at the actor-level.
Table 5.1.11 showed some of the actors who ranked relatively high in having internal and
external ties.
Table 5.1.11. Communication Network Pre-Earthquake Ranking of Variability
across Actors with Group Trait Based on Registration Actor Attribute
Internal #115, #137, #119, #6
External #51, #61, #100
Actors #115, #137, #119, and #6 tended towards group closure concerning registration
status. Before the earthquake event, each of them made higher level of communication
with those with similar registration status with themselves. Among them, #115 and #137
were registered domestic nonprofit organizations. Looking back at the actors who were
highly active in initiating ties towards others, both #137 and #115 were among those with
relatively high out-degrees. We now have a more precise picture of the embeddedness of
these actors. They were not only proactive in building communication relationships with

66
One caution when using the measures in table 5.1.10 is that the E-I index results were not in the “re-scaled” format,
which is taking into consideration of the group sizes and the density of the connected network. Therefore, it is safe to
look at these measures here in corroboration with all the other results developed through this analysis.

346
others in the network, many of their connections, regardless of which party initiated the
ties, ended up to be with registered nonprofit organizations. Both actors #119 and #6
were non-registered foreign nonprofit organizations operating in China since before the
earthquake. During this period of time, they were all active in constructing
communication ties with others as well, which ranked them high in terms of the out-
degree. Contrary to the behavior propensity for actors #137 and #115, the fact that actors
#6 and #119 had a tendency to have group closures with non-registered actors in the
communication network suggested that these foreign-based civil society actors were less
bounded by registration status of the other actors they were in touch with and were more
willingly to be open towards others, especially Chinese domestic grassroots.
Actors #51, #61, and #100 ranked the top among those who had a tendency to
have ties outside of their own registration groups. From earlier investigation, I showed
that all three actors also ranked high in sending out communication ties towards others. I
have shown the nature of the actors
67
and now I have information on the characteristics of
the kind of ties they made. These actors were all registered nonprofit organizations at this

67
Recall that actor #61 originally entered China as a foreign-based NGO, but over the years of practicing in the field
across different regions to alleviate poverty inside China, it established its formal field offices and obtained its
registration status before the 2008 Wenchuan earthquake. Actors #51 and #100 are both Chinese domestic nonprofits.

347
time period. The degree of external ties for these three actors suggested that they tended
to develop ties that cross the registration status divide in the network. When capacity
development for civil society actors is the policy goal for planning not just for disaster
preparedness and mitigation but also for times of uncertainty, the development of the
newly emerged actors in the field or those smaller grassroots informal groups is an
important aspect of policy design process. Aside from the policy measures prescribed and
mandated by the state, one critical concern is how to increase the capacity of civil society
actors themselves to take actions to cope and adapt when facing crisis. In the Chinese
context, the ability and willingness of the more experienced and established civil society
actors to extend a hand to those smaller informal ones could be a case in point. Actors
like #51, #61, and #100 that exemplify high out-degree and a tendency for high external-
group ties can be the key starting point in effectively engaging civil society actors.

Post-Earthquake
I now move on to look at the changes of the communication network after the
earthquake. First of all, let me compare the measures showed in the density tables (5.1.12
and 5.1.13).

348
Table 5.1.12. Communication Network Emergency Response Within-Group and
Cross-Group Density Measures based on Registration Actor Attribute
Group 1 (registered) Group 2 (non-registered)
Group 1 (registered) 0.098 0.107
Group 2 (non-registered) 0.107 0.074

Table 5.1.13. Communication Network Long-term Recovery Within-Group and
Cross-Group Density Measures based on Registration Actor Attribute
Group 1 (registered) Group 2 (non-registered)
Group 1 (registered) 0.108 0.128
Group 2 (non-registered) 0.128 0.089
We can see that shortly after the disaster, the density of ties built within each of the two
groups increased dramatically. The overall density within the registered group actors
increased from 0.017 to 0.098, while the density among nonregistered group actors went
up from 0.027 to 0.074. This means that actors inside each of the groups were
communicating and sharing information with significantly higher number of others of the
same registration status. The within-group information exchange relationships became
increasingly prevalent for both the emergency response and the recovery periods as
compared to before the earthquake. In addition, the registration status did not seem to be

349
a barrier for information sharing post-earthquake. In other words, civil society actors
were more willing to take cross-boundary initiatives. When comparing the in-group and
out-group ties together, the densities for out-group ties were still more prevalent than the
densities of in-group ties. This pattern of density increase remained throughout the long
term recovery stage, with the measures for both within-group and between-group
connections continued to grow steadily.
Compared to the emergency response stage, the recovery period experienced a
less dramatic climb in all density measures, for both within group and across group
communication ties. Note that the nature of the change pattern was not one that the
increasing trend stopped or reversed into decreasing trend, instead, actors continued
building ties both in and between groups with the process turned to be steady. This
primarily indicated that actors, both registered and non-registered were willingly to
maintain the ties that were built during period shortly after the earthquake. The change in
agency actions over the emergency response stage was not merely an instinctive
temporary surge of kind-heartedness and voluntarism on the part of those ordinary
citizens who participated in the response activities. It was more importantly, a collective
intentional journey for the Chinese civil society to cope and search for its own identity
and growth when facing a crisis. A process of institutionalization of connections among

350
the group and organizational actors had been an inherent pattern revealed by the
sustained efforts in tie-building over time. This pattern can also be demonstrated by
examining the communication network as a whole, shown in table 5.1.14 and table
5.1.15. The number of internal ties jumped up from 228 to 1208 (64.2%), and the number
of external ties went up from 194 to 674 (35.8%).
Table 5.1.14. Communication Network Emergency Response Whole Network
Results of Group Internal and Group External Ties Based on Registration Actor
Attribute
Frequency Percentage Possible Density
Group Internal 1208.000 0.642 12584.000 0.096
Group External 674.000 0.358 6322.000 0.107
Note: There are a total number of 1882 ties.

Table 5.1.15. Communication Network Long-term Recovery Whole Network
Results of Group Internal and Group External Ties Based on Registration Actor
Attribute
Frequency Percentage Possible Density
Group Internal 1348.000 0.625 12584.000 0.107
Group External 808.000 0.375 6322.000 0.128
Note: There are a total number of 2156 ties.

351
After adjusting the different group sizes and the density of the graph, the re-scaled
E-I index for the emergency response stage was -0.284. We can say that, given the
demographic constraints in terms of the maximum possible ties that could be made inside
each group (shown in fourth column of table 5.1.14) and the overall density (shown in the
last column of table 5.1.14), the communication network during the emergency response
stage had a higher tendency, when compared to before the earthquake, to in-group
bounded-ness and closure in terms of registration status. For a social structure that
focuses on communication and information exchange, such a slight tendency towards
sub-population closure in terms of registration status increased the possibility that there
would be barriers in information flow, especially for registration-related cross-group ties.
One explanation for such a tendency is, with the increasing network integration with the
emergence of informal grassroots groups at this period of time, it could be a phase where
civil society actors trying to get to know each other and starting to be aware of the
existence of those just came to disaster response and recovery practices in the field. The
tendency to go towards those with similar registration attribute could be the beginning
stage for actors to explore the possibility of developing communication partners. In the
short term after the disaster, the time constraints for actors to take action promoted a

352
temporary kind of “near-sightedness” behavior. This resulted in being drawn to those
with similar traits as themselves.
Some parts of the emergency registration sub-group structure pattern continued
into the long term recovery period. The total number of ties increased from 1882 to 2156.
Out of all these existing ties, the percentage of internal group ties remained to be larger
than those made externally. While the number of connections inside each group kept
increasing, the percentage for group internal ties decreased slightly to 62.5% with the
group external ties increased to 37.5%. There was a slight change in the grouping
structures for the immediate stage after the earthquake. The overall internal group
communication decreased by 1.7%, while the external group ties increased by the same
percentage points. This could be an encouraging sign because the nature of the
communication network structure appeared to become less segregated in terms of
registration status. Alternatively, if we look at the overall density of those ties made
internally and externally, the latter (0.107) turned out to be higher than the former
(0.096). One point worth noting here is that these measures were calculated based on the
consideration of the maximum possible number of internal and external ties, which was a
different formula from using the actual existence of ties to calculate the percentages. The
density measure, therefore, represented a more comprehensive picture of the

353
concentration of ties for each of the group. For the long term recovery period, the speed
of the information diffusion across the two groups was more likely to have increased,
given the consistent increase in density measures of both group internal and group
external ties.
In terms of calculating the E-I index when considering the differentiation in group
sizes and densities as discussed earlier, the re-scaled E-I index for this period is -0.250,
which turned out to be lower than that from the emergency response period. Over time,
there happened to be lower degree of tendency for civil society actors’ communication
behavior to be segregated by registration status and there were more information
pathways for cross-group communication to take place. This piece of evidence also
suggested that the behavioral tendency for group closure had been slowly dissolved as the
structure itself showed signs of institutionalization over time. As actors became more
acquainted and having had more information about others through the embedding
structures such as dyadic, triadic, and the clustering, they gradually revealed their
willingness to take actions reaching out regardless of the registration status concerns.
Going further in-depth with the group level of the analysis, tables 5.1.16 and
5.1.17 illustrated the group level E-I index for the two periods after the earthquake.

354
Table 5.1.16. Communication Network Emergency Response Group level E-I Index
Based on Registration Actor Attribute
Internal External Total E-I
1(registered) 1148.000 337.000 1485.000 -0.546
2(non-registered) 60.000 337.000 397.000 0.698

Table 5.1.17. Communication Network Long-term Recovery Group level E-I Index
Based on Registration Actor Attribute
Internal External Total E-I
1(registered) 1276.000 404.000 1680.000 -0.519
2(non-registered) 72.000 404.000 476.000 0.697
Comparing before and immediately after the earthquake, the number of internal ties made
by both groups of actors climbed up during the emergency response stage, but with a
more dramatic increase inside the registered group from 206 to 1148 ties. The rate of
growth slowed down when it went into the disaster recovery period. On the one hand,
both registered and non-registered group members continued to exchange more
information inside each of their own group. On the other hand, one of the most
significant structural development at this point could be reflected through the number
changes in cross group information exchanges. Recall that before the earthquake, there

355
were a total of only 97 cross-group ties. Shortly after the disaster, this number jumped up
to 337 and kept increasing to 404 during the long term recovery. From the E-I index
column, the two groups also showed different tendencies in terms group closure. At the
immediate response stage, the registered actors seemed to be more likely to have in-group
ties while the non-registered group actors was more likely to build ties outside of their
group. Over the long run, the tendency for registered actors to have in-group closure
waned while the non-registered actors remained consistent in their efforts in reaching out
to form cross-group communication ties.

Civil Society Actor Level Variability
I now briefly examine the variability across actors. This will provide a zoomed-in
picture of what happened to the agency actions within the civil society domain. Tables
5.1.18, 5.1.19, and 5.1.20 illustrate some of the top-ranking actors that exhibited high
level efforts in building internal and external group ties before and after the earthquake.
Table 5.1.18. Communication Network Pre-Earthquake Ranking of Variability
across Individual Actors with Group Trait Based on Registration Actor Attribute
Internal #115, #137, #119, #6
External #51, #61, #100

356

Table 5.1.19. Communication Network Emergency Response Ranking of Variability
across Individual Actors with Group Trait Based on Registration Actor Attribute
Internal #24, #3, #32, #34, #6, #57, #107, #137
External #51, #61, #135, #24, #93, #34, #7, #100

Table 5.1.20. Communication Network Long-term Recovery Ranking of Variability
across Individual Actors with Group Trait Based on Registration Actor Attribute
Internal #3, #32, #6, #64, #76, #24, #12, (#123, #19)
External #93, #51, #135, #61, #3, #7, #100, #101
Overall, there was a general increase in the number of both internal and external ties
across the communication network shortly after the earthquake. The trend continued into
the long term recovery stage. This could be another piece of evidence that suggested the
endurance of the communication network structure over time. In particular, actor #137
and #6 remained high ranking in terms of having internal group ties for the emergency
stage while actor #6 was consistently having relatively high level of internal ties even
through the recovery period. Among the newly emerged grassroots actors, group #3 rose
up towards the top by having an increasing number of ties with non-registered actors and
maintained its position over the long term. The following qualitative examinations were

357
conducted to understand the sources of this kind of within civil society action and
institutional development.

The Case of Social Group Actor #3 (SG3
68
)
There were two types of relationship-building activities being experienced by
group actor SG3 after the earthquake event. As a non-registered entity, one type is its
interaction with registered civil society actors. The other type is the interaction with the
state sector. Regarding the former type of within sector relationship, the factor of
institutional formality became less of an emphasize aspect of consideration among the
civil society actors as they interact with each other. When asked about the group’s non-
registered status and its influence on communicating with others inside the civil society
domain, the young participant recalled:
In fact, throughout the times of our interaction, we don’t really have this concept
(of who is registered and who is not) at work. Whether it is our work relationships
or other types of interactions, or even personal communications, this won’t be an

68
For non-registered actors, I used the code “SG” (meaning Social Group) followed by their unique number given in
the network analysis. For registered organizations, I used the code “NGO” followed by their unique number given in
the network analysis.

358
issue at all. To be honest, I myself wasn’t even clear of all these terms, like
‘volunteer teams’, ‘NGOs’, etc. (SG3-01-11
69
)
First, note that she referred to the formality issue in terms of the registration status as a
concept that became open to interpretation from the perspective of different civil society
actors. When it came to communication relationship-building after the earthquake event,
the formality of an actor was not understood as a merely an attribute status that simply
distinguishes the actions of an informal social group or a formal organization, instead,
what mattered was how the actor experienced its own actions in relation to those whom it
provided services towards or with other civil society actors. In fact, at the early stages of
communication action shortly after the earthquake, the young participant herself was not
even aware of the differences between “volunteer groups” and “NGOs”. In other words,
communication connections among group/organizational actors were clearly not so much
based on institutional formality in terms of registration status.
It is important to note the phrase that the young participant in SG3 referred to the
group’s interactions with civil society actors. She consistently used the term “ 大家”
(Dajia), which has a direct translation in English means “us”. But when communicated in
this context, the phrase purveys a meaning of “togetherness”, or more precisely, a type of

69
For original Chinese script please refer to Appendix 5.1.5.02.

359
relationship structure that is perceived as “cohesive” in such a way that it identifies
group/organizational actors as belonging to a larger sphere of framework. First of all,
this sense of “belonging” did not derive from a “top-down” hierarchical sense of
connection, but was conveyed through a presentation of civil society actors “standing” on
a level ground as sharing a common aspiration in their devotion in disaster relief and long
term recovery.
…We have never thought about who belongs to whom, or any related concepts.
Maybe because, for one, through such a process of getting to know each other,
either by means of work or personal relations…and whenever they had a need, if
not related to finances, but in terms of information or other types of consultation
services, our organizers SG3-03 and SG3-02 would provide some suggestions on
their projects. In other words, whenever they have some kind of needs, they
would come to us. And when we organize activities, be it training or study
seminars, we would send out to them. They can come if they are interested.
It was not as if we were a specific bounded group, or a platform whose center is
us, we have not thought of it that way. In fact, shortly after the earthquake, some
of us also signed a collaboration memorandum. By that time, we were not an
independent concept, the center itself is a platform. All of us are the main actors
on it. That’s why we signed such a memorandum. If that is so, then, the center
would be more precisely in a nature of ‘league’, and this entity exists on
performing as a platform. And there are many groups/organizations functioning
on it. They had connections among each other, but not in an intensive way. Then,
as people had to go back to their own work schedule after the immediate response
period, the communication relationships among us needed someone to maintain
their existence. Based on such concerns, an independent actor with substantive

360
entity-like features gradually emerged, but still functions like an intermediary
connection point. (SG3-01-12
70
)
Three aspects in terms of the nature of actor SG3 can be related to its institutional
dynamics with the other civil society actors. First of all, the identification of “belonging
together” is exemplified through the group’s providing services for others not just in
sharing information, but also in consultation and training. Most importantly, these
interaction activities between the group and other civil society actors were also
intertwined with other types of relationships among the representatives of groups and
organizations. And these connections included friendships among representatives of
groups/organizations built at the emotional level throughout the periods after the
earthquake event.
Secondly, the emergence of SG3 from short term to long term after the disaster
cannot be seen as a stand-along process. It co-evolved with the institutional development
and interactions of all the civil society actors that the group was trying to support over
time. Note that there was a critical difference between the characteristic of SG3 during
the emergency response period and the long term recovery period. At the most primal
stage of its own institutionalization process, the group was perceived less of a formal

70
For original Chinese script please refer to Appendix 5.1.5.03.

361
institutional entity, but rather, could be more precisely described as a “virtual” platform
that functioned to provide a bonding-friendly environment for actors to interact with each
other. As time went by, especially when coming into the long term recovery stage when
representatives of voluntary social groups went back to their daily work, an independent
entity of SG3 slowly emerged basically to serve the purpose of maintaining the
connections among all the civil society actors. The term “concept” was being used in the
young participant’s account as a way of representing the emergence of SG3. Essentially,
the institutionalization process of SG3 started out by it being conceptualized through a
relational perspective, one that characterized as being an “intermediary” performing the
role of connecting and promote interaction among other civil society actors. The group is
in itself an enabler facilitating a “growth-oriented” institutional climate inside the civil
society domain. Its eventual establishment as an informal social group began with an
emphasis on building a platform of relationships and later transformed into being
willingly embedded as an actor whose identity evolution tend to be further intertwined
with the connection decision-making of others. From here, we can see the importance of
looking at the institutional development of civil society from a relational perspective
because it could be the source of how actors identify themselves at the prime stage of
emergence, and thus as a foundation for their evolution over time.

362
From this kind of institutional transformation, it is important to emphasize the
group’s identity transition after the disaster also became a stepping stone for it to sustain
itself as a civil society actor practicing long term in the field of nonprofit sector. As was
described by the young participant:
At the beginning we see our entity as a center, like I mentioned earlier, meaning
in the form of a joint entity (“commonwealth”) and needed the participation of
many member organizations. We even collaboratively signed some kind of an
agreement. At the earliest times, there were 21 organizations, and then evolved
until the end of 2008. After that, you could observe that the joint-entity nature of
the center has gradually been less emphasized, either intentionally or un-
intentionally, or due to some objective reasons. Then, we have not actively raised
the concept of “member organizations. (SG3-01-12
71
)
We are servicing other organizations, this ‘service’ identity originated from our
intention to provide assistances to those other civil society groups/organizations
that were conducting disaster related works. But in fact, we would provide
services to all of those civil society groups/organizations that come to us for help
or assistances. Therefore, we gradually would no longer talk about the concept of
‘member groups/organizations. (SG3-01-13
72
)
As we can see, the evolution of the institutional identification for the group itself was also
accompanied by an expansion of services towards the development of civil society actors
in general rather than bounded to particularly those focused on disaster recovery. What
was being exemplified here could be referred to as the beginning stage from which the
actor SG3 originated from a crisis situation and gradually emerged into a non-disaster-

71
For original Chinese script please refer to Appendix 5.1.5.04.
72
For original Chinese script please refer to Appendix 5.1.5.05.

363
related nonprofit group functioning as a long term coordinating and servicing partner for
other civil society actors. This stage was essentially characterized by civil society actors
functioning independently from each other without a hierarchical orientation by
categorizing themselves as members within a larger organizational entity. What can be
inferred was that over time, especially after the emergency response period, actors inside
the civil society domain indeed valued independent growth from a grass-root “bottom-up”
approach. Such a tendency could be characterized as a starting phase that signifies a
“standing up” Chinese civil society in the context of after a catastrophic disaster. This
standing-up character can also be demonstrated by the ability of civil society actors to
form organizational network relationships among themselves.
We have not intentionally count the exact number, but roughly speaking, those
who have had connections with us, either in the type of communication or
collaboration, can sum up to at least 300. This is probably one of the biggest
resource assets of the center since its establishment three years ago. It has created
these relationships with other groups/organizations in pretty good terms. (SG3-01-
14
73
)
In terms of the types of interactions within the immediate network neighborhood
of actor SG3, the relationships can be categorized into the following based on the account
of the young participant (SG3-01). The first type of communication relationship arose

73
For original Chinese script please refer to Appendix 5.1.5.06.

364
from the need of group SG3, functioning as an information coordinator, to gather
information from other civil society actors performing disaster recovery works in Sichuan
Province at the time. In these incidences, SG3 will reach out to them first by phone or
email. The requested information will then be posted as “work briefing” on the website of
SG3. As the person in charge of this section of the duties, the young participant also
enthusiastically planned to design the website in such a way that “when you get online,
you’ll be able to click open a map of Sichuan Province, and at the same time you will see
the location of each group/organization’s recovery related projects, as well as which
project was completed and which ones are ongoing…”. The second type of
communication tie arose from the needs of other civil society actors towards the services
provided by SG3. In these cases, other groups/organizations will initiate the
communication connection with the group. In terms of work contacts, these will be
incidences when they would have needs for information support or other kinds of
assistances. During the transitioning period from emergency response to the recovery
phase, these types of support including arranging temporary housing and helping in
purchasing tickets for traveling were strongly demanded particularly from those civil
society groups/organizations coming from outside Sichuan Province (SG3-03). Another
sub-type of the incoming nominations arose from the need for coordinating available

365
temporary spaces or logistic assistance when civil society actors were interested in
organizing trainings or activities. Different from these two types of work-related
communications, the third kind of connection can be categorized as emotionally-related
personal friendships. “As more young people are joining in this ‘circle’, we more or less
share some of the same hobbies or interests together, particularly when we were spending
our leisure times hanging out with each other. At other times, (some work-related
connections) might be sparked by us young people sharing off times” (SG3-01). From
here, the nuances of how work-related communication networks among civil society
actors were actually intertwined with other types of network environments. The last type
of communication means was through an online interactive medium called “QQ group”
where all kinds of information can be posted by civil society actors. For example, group
SG3 could send out invitations towards others to inform them of the opportunities for
participating in training programs. Postings could also be in the form of actors searching
for volunteers and posting their requests through this interaction medium online. This
would usually lead to faster responses.
One other point that deserves further elaboration is the decision-making processes
behind the institutional structural change of the group SG3 through emergency response
to long term recovery phases. For this group actor in particular, the time when the entity

366
as a whole had to decide whether to continue the function of the original information
platform came in June 2008, which is about a month after the earthquake event.
According to the reflection of one of the senior participant, there were two sides of
opinions among the participating civil society actors. One side argued that since the
emergency response period has already passed and the involved groups/organization
could establish their own local branches in the local communities that they intend to
provide assistance towards. Therefore, there were no needs of the continuation of such a
platform. Other civil society actors, particularly those smaller grass-roots coming from
outside the Sichuan Province, strongly raised their voices in preferring the sustaining of
the function of the platform.
If this platform had not existed, they would face significant difficulties, especially
those coming from outside the Sichuan Province. This is because when it was just
one month past the earthquake event, NGOs from all over the country would
gather there, and they would have special needs for assistance. As they would
often say that they weren’t familiar with the local culture. (SG3-02-03
74
)
The result of the discussion was that the group as a whole decided to continue its
existence with a more sustainable form toward an institutionalized structure instead of
being simply a group of volunteers coming together helping out with the emergency
response efforts. Such a transitioning mode was defined as the group performing a type

74
For original Chinese script please refer to Appendix 5.1.5.07.

367
of project support in order to preserve its identity and functioning
75
. In summary, the
tendency for actor #3 to be increasingly active in building communication relationships
with other non-registered actors could be attributed to three factors. One is its pursuit in
developing platforms for information exchange thinking beyond the stages of disaster
response and recovery. Secondly, it became increasingly aware of the co-evolution of its
own functioning with all the other civil society actors. Thirdly, recognizing the
“togetherness” with others working in the field and sharing a common aspiration for long
term social development also contributed to its continued connections with those newly
emerged non-registered social groups.
Alongside with other civil society actors that emerged as key players promoting
communication inside and between groups, actor #24, also became one of the most active
agents in both fields of having internal and external ties shortly after the earthquake. This
type of action cannot be separated from how the actor cared and perceived the
institutional environment of Chinese civil society actors in general.
In fact, the informant of the organizational actor #24 (NGO24-01) held a rather
positive outlook to the general institutional environment for the future development of

75
For detailed account, refer to Appendix 5.1.CaseSG3.4

368
grassroots groups and NGOs. He perceived that the government actually was “slowly
starting to realize the importance of NGOs and was gradually re-directing its attitude
towards us”. According to the recall of the director, it had stated clearly in year 2011
twelfth fifth year plan of the important role of social development in disaster mitigation
and preparedness. And the need to discover a process incorporating the collaboration
efforts between the government and NGOs, and thus the development of a “big society”
was stressed among all the main tasks that the government needs to perform. As he was
describing these observed progresses in government policy orientations, the director
finally stated that “I believe that the development of existing (institutional) environment
will be in more and more favor of the survival of NGOs” (NGO24-01-02).
LU ：Then, how should one raise the government awareness of the importance of
NGOs?
NGO24-01: In fact, there is a gradual change in government’s attitudes towards
NGOs and they started to recognize our importance. For example, in the twelfth
five-year plan, it raised the importance of social development and its role in the
process of disaster mitigation and preparedness. In the plan, one of the critical
tasks raised for the government is for it to manage well-functioning collaborations
with NGOs and nurturing the development of ‘big society’. So I believe the
surviving institutional environment for NGOs will keep getting better. (NGO24-
01-02
76
)
Aside from a positive attitude toward the institutional environment for the future
development of civil society actors in building up the social capacity for disaster

76
For original Chinese script please refer to Appendix 5.1.5.08.

369
mitigation, there was one other driving factor that could explain the higher concentration
of within-group and cross-group communication activities after the earthquake. The
actor not only cared for the development of indigenous sources and strength on the part
of Chinese civil society for disaster recovery purposes, it also had a long-term vision in
the general trajectory of social capacity-building for China. For example, the informant
talked about several internal difficulties arose within grassroots groups and NGOs
themselves which contributed to the additional burdens for these civil society actors to
grow over time. One was the lack of the funding sources and the shortage of it could
further hinder the “professional capacity-building”. The other internal factor was
regarding the types of activities particularly chosen by the voluntary groups after the
earthquake event. To the informant, many of these informal voluntary groups rushed into
the disaster hit areas to focus more of their attention on emergency response activities
alone. And less attention was being paid to the long term recovery aspect of activities
being established at that period of time. Therefore, the actor’s thinking not only went
beyond the care for its own post-earthquake activities but also went toward the growth
process of other civil society actors in the field in terms of building up the social capacity
to withstand future crisis. These motivations promoted the shift of expansion in both
within and cross-group actions after the earthquake.

370
One other distinguished feature of the ranking of external group ties across actors
was that actors #51 maintained to occupy highly active positions in communicating with
non-registered actors. It established the highest number of group external ties for both the
response and recovery periods. Because of its persistence in engaging non-registered
actors with information exchange, it would play a significant role in developing policy
interventions that aim at breaking down the tendency for group segregations based on
actors’ attributes and at enhancing diversity composition of the communication network,
particularly in the case of disaster preparedness planning.

The Case of Actor #51 (NGO51)
In this research, I distinguished three types of civil society platform-building
initiatives. Recall that one source of motivation in platform-formation was to provide an
environment that welcomed and facilitated the communication and information sharing
among civil society actors in general. The primary example for this kind of role
formation was revealed through the actions of actor #3. Its platform initiative was also
being sustained through the transitioning from emergency response to long term recovery.
The second type was one that emerged for facilitating not just for connections among

371
civil society actors but also for cross-sector communication and collaboration
relationships through building a particular field of practice. For some actors for instance,
the field specialty was chosen to be in youth education and development. Thirdly, from
the example of actor #51
77
, I found another type of platform initiative in the form of
facilitating capacity building for civil society actors. Through the emergence of various
community development programs throughout the earthquake-impacted areas, an
institutional environment that could enhance an in-depth understanding among actors
across the civil society, the state, and the market domains was gradually being cultivated.
This thus became a collaborative learning experience not just for the growth of civil
society actors but also for enhancing the strength of cross-sector initiative in field
practices. In later sections, I will provide a detailed account of how this process was
being initiated and conducted through one field project implemented by actor #51.
For now, let me focus the attention on finding out the driving forces for actor #51
to engage in cross registration group activities both before and persistently after the
earthquake. After establishing a formal field office and gained an official registration
status in Chengdu, the actor’s original motivation in participating in the earthquake
recovery gradually expanded towards functioning areas related to long term social

77
For interview accounts on this topic, refer to NGO51-01-03 in Appendix 5.1.5.09.

372
development. Some of them included being an “incubator” for emerging civil society
actors, establishing community service platform, as well as civil society capacity building.
The common thread these areas of specializations had was a devoted concentration in the
institutional development of the Chinese civil society domain. By “institutional
development” in this context, it incorporates three levels of interpretations. One is at the
level of individual civil society actor, incorporating the meaning of an emergence of a
formal type of organizational structure from informal social or volunteer groups. The
other one is at the relational level, incorporating the establishment of long term
communication and collaboration connections among civil society actors. The third level
of understanding is a combination of the first and the second, meaning that the capacity
of one civil society actor is developed through a process of interacting with others. The
following illustrated how the case of actor #51 exemplified this third type of institutional
development through its field actions
78
.
The civil society groups/organizations that we aim to develop are like this. One is
that they participate into our own programs. Aside from our N programs, we also
will introduce some groups/organizations into other programs established in the
high-tech district in Chengdu. For these kinds of groups/organizations that have
collaborative relationships with us, we will do our best to nurture their growth.
This is firm. The other one is our “incubator” program. This is particularly
designed for assisting the growth of civil society groups/organizations. So as long

78
For further detailed account, refer to Appendix 5.1.CaseNGO51.2.

373
as they have communication relationships with us, we will be responsible to train
and ‘bringing them up’. (NGO51-01-04
79
)
From the informant’s perspective, a “collaborative” relationship incorporated two types
of actions. One was through the participation of its community development programs by
being a “third-party” civil society group or organization assisting in running the local
community centers. The other route was by being “incubated” through one of the civil
society capacity development programs. Both types of relationships counted towards
“collaborative” connections between other civil society actors and actor #51, while at the
same time they inherently provided a nurturing and enabling environment for the growth
of each partnering civil society actor by nature of each of the program goals
80
.
It can be concluded that the high level of communication between actor #51 and
non-registered social groups mainly originated from its goal in supporting the
institutional development of civil society actors through the nurturing of collaborative
relationships. This type of relationships in-turn promoted the construction of a
communication platform for the involved parties to further share and exchange
information, thus providing possibilities for facilitating capacity development among
actors across the civil society, the state, and the market domains.

79
For original Chinese script please refer to Appendix 5.1.5.10.
80
See NGO51-01-05 in Appendix 5.1.5.11.

374
Different from actor #24, #3 and #51, actor #123 first emerged to become one of
those actors that had higher level of group internal ties only during the long term
recovery stage. Since actor #123 was a non-registered social group, the higher level for
its internal group ties represented that it had been particularly active to communicate with
more informal civil society actors especially during the long term recovery period.
Therefore, I further examine the origin of its emergence at the group level in order to
qualitatively trace the origin of such actions.

The Case of Actor #123 (SG123)
When inquired regarding the motivations for forming both communication and
collaboration connections that these two actors, the motivation to seek appropriate
funders became an important issue.
The organizer distinguished two types of foundations from which the group could
seek funding from at the time. One is called the “private foundations” (私募基金会)
and the other is “public foundations” ( 公募基金会). While both could possibly provide
funding for the group, the preference is given towards the former. The reason is that, in
comparison to the “private foundations”, funding from the official channels through the

375
public foundations will often provide less “maneuver room” for the fund seeker to
perform stand-alone tasks that are independent of the possible “influences” from those
operating the public foundations. On the other hand, funds coming from private
foundations would provide more freedom for the seeker to manage tasks in a way that fits
the need of the field activities themselves.
Knowing (actor #125) was due to my own effort to find funding and had to travel
to Beijing to meet them. I personally prefer the private foundations (as compared
to public foundations), like ND foundation, and TX, and many other
nongovernment-related foundations. This is because many of the government-
related public foundations are more being controlled by bureaucrats. Once you
join and apply for their money, you will be skimmed off and used for their
purposes and will instead become one of their tools to draw money for themselves.
Even if you were successful in your application, a lot of the money will be
deducted and go to their own pockets. What’s left-over for us will be very slim.
But for private foundations like (actor #125), their focus is on getting some real
work done, and the source of their employees’ salary was provided by (the
founder himself). (SG123-01-03
81
)
Essentially, what the organizer valued the most with its funding relationships with the
two private foundations was that their provision of not only the funds needed perform
group’s desired tasks, but most importantly offered an opportunity for the group to
develop its own capability in functioning. Just as expressed in his interpretation of the
“standing up” role of NGOs in relation to the general mass and the state, the ability and

81
For original Chinese script please refer to Appendix 5.1.5.12.

376
the freedom to do what the group values in the best interests of those whom it serves, is
critical in the process that the group built its relationship partners over time.
Another within sector connection the group established during the emergency
response stage was with actor #3. At the time immediately and short-term after the
earthquake, actor #3 was perceived as an important “mediator” connecting the needs of
information across civil society actors. Note that it was indeed the primary goal set by the
organizers of actor #3 at the time of its establishment. However, over time, when it came
to the training programs coordinated and provided by actor #3, the organizer of group
123 described his frustration in the materials that had been taught and passed on. One
comment he made was the gap between the theoretical content of the training and what
was happening in the field. This discovery eventually led the group 123 to drop its
communication nomination towards actor #3 during the recovery stage. From a policy-
making perspective, what this process revealed is that as civil society actors grew and
matured in different stages, the tools that were needed for them to conduct activities and
practice in the field diverge. Although theoretical background learning experiences were
essential in developing a way of thinking in terms of a “bigger picture” in social
development, they seemed to be afar from what the group faced in reality. However, this
does not cancel out the contributions of actor #3 in performing a role of assisting civil

377
society actors building up opportunities to learn and share experiences among each other.
From the prior investigations, recall that the actor itself was also embedded in this
process of learning and adjusting its role over time. It is therefore a “collaborative
learning” experience for all the participating civil society actors to develop a type of
dynamic that not only enhances their relationship with each other but also the effective
tools they are able to select when providing services towards the local communities. If a
policy measure could facilitate such a process of relationship development while at the
same time providing guidance in civil society actors’ capability development for
initiating further innovations in tool-searching, a policy design for social “resilience”
could emerge. But the key here is to let the “agency action” side of the civil society
reveal itself over time
82
.
In general, the changes in both external and internal group ties listed in tables
5.1.16 and 5.1.17 can be seen as important components of structural fabrics that
contributed to the social resilience especially at time of catastrophic changes brought
upon by nature or man-made. On the one hand, an increasing number of both registered
and non-registered civil society actors quickly emerged in assuming the role of promoting
communication flow both internally and externally. Identifying them and understand the

82
For further account, refer to Appendix 5.1.CaseSG123.4

378
factors that contributed to their participation in the network could be critical for policies
that are designed to facilitate the speed of emergency responsiveness on the part of the
civil society. On the other hand, some of these actors were also able to sustain their role
in terms of maintaining the number of group internal and external ties throughout the
different stages of time. When a policy measure is designed to enhance the capacity-
building of the civil society actors for long term disaster recovery and mitigation, it will
be important to identify actors whose embeddedness in terms of internal and external
group behavior endured beyond just the emergency stage after the disaster event.

Structural Foundations of Agency Action
Overview
In the last section, I examined how communication structures were being
constructed from a micro-level perspective starting at reciprocation stage to the group
interaction stage. Another way to investigate the impact of action on the structuration
process is through a macro-level perspective by directly examining the larger sub-
structures.

379
In this section, I will examine the possible kinds of groupings that civil society
actors exemplify throughout the periods before and after the earthquake. Investigating the
changes in macro group structures deciphers how civil society actors whose relationships
constituting the network environment as a whole evolved and was likely to behave. The
findings can serve as the pre-conditions within which micro-level interactions can occur.
Furthermore, it is important to look at such sub-structure activities because these
structural formation behaviors tend to have critical policy implications in the context of
disaster response and recovery. For communication networks, the formation and
development of micro sub-structures at different levels can be used to identify key actors
who can perform the role of facilitating the information flow among groups in order to
enhance the overall cohesiveness of the network. For collaboration networks, actors’
attribute characteristics and their areas of specialization can be used in combination of
their structural grouping behaviors so as to design possible projects that encourage
collaborations across sub-structural boundaries. Here I will focus on the communication
network environment and next chapter will be devoted to closely examine the
collaboration networks.
In sub-structure analysis, there are generally two levels of perspectives to
investigate the different stages of existence and composition. One is at the macro-level by

380
looking at the connectivity of the whole network. The other is at the micro-level by
examining the smaller and denser communities within the larger structure and how they
combine together to form the larger network environment. In the following section, I will
start out with a macro-outlook and then further break down the investigation into smaller
sub-groups.

Top-down Approach
Component
The primary indicator of identifying the macro-grouping activities is the concept
called “component”. The identification of weak components in a network does not take
into account of the direction of a tie. As long as there is a set of ties connecting and
linking the actors together, this set of actors are named into one group of weak
component. For information exchange and communication networks, this perspective
allows one to observe whether there is a general connectedness among the actors. Those
that are not included in the main component regardless of directions of the ties will be
listed as “isolates”. Figure 5.1.5 illustrates the weak component communication structure
at the time stage before the earthquake.

381

Figure 5.1.5. Communication Network Pre-earthquake Weak Component Structure
We can see that all the connected actors were colored in red as one large
component. The isolated actors are colored in blue and being listed on the left side of the
graph. The analysis output of the weak component structure indicated that there were 63
of them found at the time stage before the disaster. Clearly, those 76 actors who
happened to be connected with each other were being identified as in the same
component. Each of the rest of the 62 actors was being categorized into a separate
component. Among them, recall that some were civil society actors that were not in
existence at this stage of time. The dramatic change in the weak component structure
came immediately after the earthquake. The first column of table 5.1.21 below shows that
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Color Red: weak component
Color Blue: isolates

382
all of the previously non-existent and non-active actors were being drawn into the main
component and as a result, there turned out to be only one large weak component
structure.
Table 5.1.21. Development Stages of Weak and Strong Components in
Communication Networks

Weak Component
(Action)
Strong Component
(Condition for
Institutionalization)
Pre-earthquake 63 121
Emergency Response 1 91
Long-term Recovery 1 91
From the whole network perspective, this dynamic indicated that the actions inside the
civil society domain exemplified a tendency to enhance the cohesiveness of the
communication network. And such an overall integration process was able to persist over
time through the long term recovery stage. The network remained to be intact with all the
actors being connected to one or the other regardless of who initiated the contact. Figure
5.1.6 and figure 5.1.7 below graphically illustrated the changes toward such integration in
the connection structure short term and long term after the earthquake.

383

Figure 5.1.6. Communication Network Emergency Response Weak Component
Structure

Figure 5.1.7. Communication Long-term Recovery Weak Component Structure

1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129 13
130
131
132
133
134
135
136 137
138
14
15
16
17
18
19
2
20
21
22
23
24
25 26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124 125
126
127
128 129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62 63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Component 1: Color Red

Component 1: Color Red

384
When direction of the communication ties were taken into consideration, another
way in distinguishing actors is by imposing a constraint requirement in deciding the
inclusiveness of the structure. A strong component is implemented by only counting
those with a directed path from one to the other in order to be listed into the same
component. Essentially, this measure takes into account of the tie direction among actors
and those listed in a strong component were actually able to reach out and communicate
among each other. This kind of communication bonding is stronger than those identified
in the weak component. As far as the strong component is concerned, it would be ideal
for all actors to belong to one large communication group. This is because in the context
of disaster response and recovery, information will not be as easily transferred across
different components as compared to among those within the same component.
Therefore, the more the break up the network in terms of the number of strong
components, the more segregated actors across groups will be, as it can be expected.
From table 5.1.21, we can see that there were 121 strong components found during the
time before the earthquake and that was nearly two times more than the number of weak
components during the same period of time.
This showed that when directions of ties were taken into account, the
communication network structure before the earthquake was divided into more sets of

385
actors among whom had directed paths among each other. With the emergence and
actions of newly formed groups and formal nonprofit organizations shortly after the
earthquake, the formation of network inclusiveness was accompanied by a decrease in the
number of strong communication components. What this indicated was that the
information exchange patterns in the civil society domain became more integrated as a
whole with less strong component grouping behaviors. Such a trend remained stable
during the long term recovery period suggesting that there was no further general
“segregation” in terms of the communication actions among civil society actors. Thinking
in terms of the role of an information exchange network after a catastrophic disaster, the
turning point in dynamic patterns of change for both weak and strong component-creation
behaviors shortly after the earthquake signified two things regarding the overall structural
development of the civil society domain. One is regarding how the function of agency
action in times of crisis was to be perceived for the Chinese case. If the agency action
alone was being considered, the changes in the numbers of weak components revealed an
important characteristic of action in the civil society domain. And this can be summarized
as a kind of energy or force that pulled communication activities into an integrated whole
where all actors were linked together regardless of who initiated the contacts in the first
place. In other words, action, in the first place, promoted completeness in the connection

386
structure. The other point is that compared to the period before, the communication
structural environment moved in such a way that created fewer barriers for actors to
access and distribute information. Given the consideration of the direction of an initiated
action, actors also created opportunities for themselves to exercise the choice discretions
on whom they aim to reach out towards. In order to depict the sources and emergence of
the available choice sets among which actors were able to choose from, it is necessary to
examine the grouping activities from a “bottom-up” perspective and I will provide further
detailed discussions in the next section.

K-core Analysis
In the previous section, I identified the “component” structures as the primary
way to understand the formation of sub-groups. The concept was used here to emphasize
on the general “connectedness” of structural development and it was helpful in depicting
how actors were being drawn into the network structure over time. As more actors were
being connected to each other either through an incoming or an outgoing tie after the
earthquake, we saw that the main component structures in the communication network

387
expanded dramatically after the earthquake and maintained its integrated-ness all the way
throughout the long term recovery period.
However, when closely observing the graphic display of the connectedness
transformations of the communication network (Figures 5.1.5, 5.1.6, and 5.1.7), it
became clear that there were actors positioned themselves on the periphery of the
network, such as being “hangers”, “bridgers”, “isolated trees”, and “isolates”
83
. When the
interest of the study is to identify the “core” set of actors that tightly knit the network
structure together, I look at the k-core structures of a graph. This method probes into the
inner-most areas of the component structures and finds a core with intensively cohesive
and connected actors. In terms of understanding the emergent structures of civil society
resilience, it is critical to identify these core set of actors and observe any changes in their
“nested-ness” inside the larger component over time.
During the period before the earthquake, the k-core output results
84
showed that
the core set of actors was found at the 5k-core level. This can be interpreted in the
following manner. At the lowest level of cohesion, a 1k-core simply represents all of
those actors that had at least one connection with each other and the boundary of

83
For definitions of these concepts, refer to Network Concept table in Appendix 4.3.
84
Please see Appendix 5.1.5B1 for K-core UCINET output results.

388
inclusion can be rather extended such as the weak component structure I identified in the
last section. As we increase the level of connection intensity to 2k-core, which means that
all actors with a degree of 1 are ignored, the number of actors inside this boundary
became less. As the cohesiveness level is being defined “stronger” by requiring only
those with higher degree measures to remain inside the core boundaries, the findings
eventually showed set of core actors at the 5k-core level. This means that actors with
degrees (regardless of directions) of 4 or less were being excluded from this core
structure. In other words, only those with a connection with 5 or more others will be
counted as part of the core. Therefore, Figure 5.1.8 illustrates this identification of the
different levels of k-core groups by node colors.

389

Figure 5.1.8. Communication Network Pre-earthquake K-core Structure
The core set of actors were being identified with red color located near the center
of the graph. By definition, these actors were the most cohesively and intensively
connected with each other during the time before the earthquake. As cut-off level of
cohesiveness is gradually being weakened, the boundaries are being stretched to include
more actors and eventually include the single component.
What is worth noting here is that the state aggregate represented by actor #1 was
inside the core structure at the highest 5k-core level. This evidence further demonstrated
that the government agencies, from the perspective of civil society actors, did play an
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118 119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Color: K-core identification
Node Size: Out-degree

390
important part in participating in the most intensely connected section of the information
sharing component structure. The market sector joined the 4k-core as the boundary is
being relaxed.
For the emergency response period, the core set of actors was being found at a
dramatically higher level of k-core85. The most stringent level of cohesiveness was found
by excluding those actors with a degree of 12 or less. Therefore, those actors who were in
the core had at least 13 communication connections among themselves. What this means
is that the core structure became more tightly-woven- together right after the earthquake
as compared to before the disaster. The nodes with red color in figure 5-1-9 represented
those actors who were counted as part of the 13k-core structure.

85
Please see 5.1.5B2 for sample K-core results.

391

Figure 5.1.9. Communication Network Emergency Response K-core Structure
Visually observing the graph, these core actors were all located in an area with higher
density where communication actions were intensely embedded among the actors. Also
note that at this period of time, the private sector represented as actor #2 joined the core
structure. This signified that the role of the market system in disaster response was being
particularly recognized by the civil society actors. At this time stage, both the state and
the market domains were part of the core structure. This was a primary piece of evidence
showing the intensity and cohesiveness of bonding among the three domains—civil
society, state, and the market—at a time of crisis. What remains to be seen was whether
such a bonding mechanism can be sustained over time as a sign of institutionalization of
the relationship structure among these three domains.
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124 125
126
127
128 129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62 63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Color: K-core identification
Node Size: Out-degree

392
For the long term disaster recovery stage, the core structure was found at an even
higher k-core level than that of the emergency response period. The core set of actors
engaged in information exchange at the 14k-core level
86
and this means that only those
actors that had 13 or more communication ties with others will be included in the core
structure. The intensity of relationship was one degree higher than the boundary defined
in the emergency response stage. This structural development showed that the intensity of
connection among the core members not only sustained but also rose to a higher level
over the long term. This point can be demonstrated graphically by observing figure 5.1.10
below.

Figure 5.1.10. Communication Network Long-term Recovery K-core Structure

86
Please see Appendix 5.1.5B3 for sample K-core output results.
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129 13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25 26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Color: K-core identification
Node Size: Out-degree

393
The state actor still played a critical role in contributing to the connectedness at the 14k-
core level. The market actor, on the other hand, did not join the state to be counted as part
of the core at this level of cohesiveness. But it made it to the core when relaxing the
constraint to 13k-core. In figure 5.1.10, the actors in the 13k-core were shown by nodes
with color red and the color of light blue. The light blue ones further expanded the
boundary of the core set with 14k-core intensity.
Overall, actors inside each of the component, especially those within the core
structures remained to be tightly bonded together over the long term. Rather than
becoming disintegrated, the evidences presented in this analysis showed an increasing
intensiveness and cohesiveness of connection from the emergency response period to the
long term recovery period. The state and the market domain became intensely bonded
with the civil society actors shortly after the disaster, and such “attraction” dynamic
remained throughout the recovery stage. What can be inferred from this type of core
identification analysis was that the cohesiveness among actors with high out-degree
measures had not been a temporary structural change. The increasing tightness among the
k-core actors can be interpreted as that the communication agency structure not only
among the civil society actors themselves, but also among domains of state, market and

394
civil society showed signs of being institutionally transformed towards long-term
cohesiveness and connectedness.

Community Structuration in Information Exchange
The Concept of Community structure (Girvan-Newman Modularity)
In order to look further into the structuration process, it is necessary to go one step
further to examine the existence of possible community structures. By “community”, I
interpreted it as groups of actors that are tightly knit together among themselves but only
loosely connected between groups (Girvan and Newman, 2002). The general purpose of
this part of the investigation was to develop a lens through which to understand social
grouping formation and change process before and after a crisis situation.
The Girvan-Neman community structure detection method focuses its attention on
the importance of tie connections rather than actors. It calculates the betweenness
centrality of a tie, which is defined as the number of shortest paths between pairs of
actors that run along it. The difference between the betweenness measures that I have
discussed earlier and the one being used here is that the former pays attention to the role
of actors while the latter focuses on the role of connections. Essentially, the interpretation

395
of such a role played by one communication or a collaboration tie is that the higher the
betweenness measure, the more loosely connected communities had to go through this
connection in order to reach to other communities at the shortest paths. In other words, if
these ties are being removed, various groups will be singled out and separated from each
other. This way, different communities are being detected. The discovery of any
particular connection that played a critical role in forging relationships between groups
that would otherwise be separated would indicate that in the emergence process of the
communication structure after a disaster, it was not only the actions themselves mattered,
to whom a tie was connected to and the direction of the tie also mattered in holding up
and maintaining the connectedness of the structure of the network.

Communities in Information Exchange Network (Girvan-Newman Detection Method)
Using this method, I examined the changes in community structure for
communication networks before and after the earthquake. Besides observing the changes
in grouping structures, the out-degree measures and the actor attribute of registration
status were also being added in the graphical representations for each period of time. This

396
was done to depict any patterns of community structuration that can be related to actors’
outreach intensity and registration status.
Starting from the period before the disaster (see figure 5.1.11), the color
representations in the following graph illustrated the different communities detected.

Figure 5.1.11. Pre-earthquake Girvan and Newman Community Structure-
Communication Network (Highest Q Modularity Value=0.136; 4 Communities
Found)
Note that within the main connected structure, four groups were being discovered. The
largest community in conducting information exchange was among those nodes with red
color. Most of those who were active in initiating communication ties were being
detected in this largest community. Three other smaller ones colored in grey, pink and
yellow were being singled out at this point of time. Note that actor #32 would be cut off
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118 119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Size: Out-degree
Color: communities
Circle: registered
Square: non-registered

397
from the rest of the network if it had not initiated a contact with actor #5. In a structure
like this, the only way for those in other communities to have access to the information
that actor #32 offered was through its connection towards actor #5. The functioning role
of the latter in this case was similar to one of a “messenger” in connecting actor #32 to
the more connected section of the network. Such a role can also be found for actor #115
and actor #20 inside their own groupings. In general, there was certain degree of
separation inside the communication network structure and various communities were
being detected for the time stage before the earthquake.
Figure 5.1.12 showed the result of community detection during the period shortly
after the earthquake.

398

Figure 5.1.12. Emergency Response Girvan and Newman Community Structure-
Communication Network (Highest Q Modularity Value=-0.000; 5 Communities
Found
87
)
With more previously isolated actors joining the network and the emergence of new
groups and organizations, this time period was characterized by a pattern of full
integration of actors regardless of the level of outreach activities and the registration
status. One major community was being detected with the majority of the actors being
included except for three actors on the periphery of the network structure. What could be
inferred was that the happening of the disaster event became a turning point for civil
society actors to be tied together by communication actions. For connections that are

87
Actor #26, #30, #16, and #23 were found to be 4 separate communities each by themselves. The rest of the actors
were all in one community.
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123 124 125
126
127
128 129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62 63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Size: Out-degree
Color: communities
Circle: registered
Square: non-registered

399
being built for information exchange, such integration dynamics would be working in
favor of getting most important pieces of information through the structure within the
constraint of time, especially for providing timely assistances in allocating resources for
the emergency response efforts. Surprisingly, based on the Girvan-Newman detection
method, this “one-ness” of community structure was also being maintained all the way
through the long term recovery period (see figure 5.1.13).

Figure 5.1.13. Long-term Recovery Girvan and Newman Community Structure-
Communication Network (Highest Q Modularity Value=-0.000; 2 Communities
Found
88
)

88
Actor #80 was found to be a stand-alone community by itself. The rest of the actors in the connected network were
in one community.
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129 13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25 26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Size: Out-degree
Color: communities
Circle: registered
Square: non-registered

400
All actors consistently fell into one community structure in terms of information
exchange except actor #80 because it had only one connection toward the main structure
of the network. Removing that tie will essentially isolate this actor from the rest of others.
As for the 137 other actors, the network cannot be separated into more loosely connected
groupings that actors must go through certain links to get to others. In general, after the
earthquake, the communication ties among actors became so dense that by removing any
one connection from one actor to another would not separate them into different
groupings. This was because the agency actions to engage in information exchange was
so strong in intensiveness and breadth that the relationship bond was able withstand the
cutoff of any one tie while maintaining closely attached in the same community structure.
When a community of actors was able to be bonded through such strength of togetherness
not just for the short term but also over the long term, further development of
commitment in various kinds of activities were made possible, such as forging
collaboration relationships.
The relatively low Q modularity levels identified across the three time periods
indicated that there might be more nuanced changes in the community structures that the
Girvan-Newman method had not detected. As a preliminary effort in showing what such
changes might involve, I therefore adopted an alternative method (Blondel et al., 2008)

401
that could take a closer consideration of the communication networks at hand and was
able to distinguish further communities, particularly for the emergency and recovery
stages. Figures 5.1.14 to 5.1.16 graphically illustrated the detection of communities along
with the depiction of actor out-degrees before and after the disaster event. Note that
further communities were indeed being identified with different colors, particularly for
the emergency response period (see figure 5.1.15). Comparing the community structures
before and immediately after the disaster event, one can easily find that there was first of
all, a drastic expansion in the number of actors inside each of the information exchange
communities. The second feature was that the intensity both within and across
communities clearly became denser. Such a pattern of diversification remained through
the long term recovery as the density of the overall network increased over time. This can
be demonstrated by the density of information exchange connections represented by the
different colors of the lines connecting the actors in figure 5.1.16. These graphic
demonstrations presented an alternative way in understanding the changes in community
structure of the communication network. Note that during the emergency as well as the
recovery stages, the network remained to be consisted of five different communities but
with increasing number of actors inside each one. Furthermore, the intensity of
connections among actors within each community and across communities became

402
increasingly dense, thus showing the insurgence and persistency of agency action after
the earthquake. Therefore, the nuances of these community structures are not to be
ignored.

403

Figure 5.1.14. Pre-earthquake
89
Community Structure Detection-Communication
Network (Detection Method by Blondel, et al., 2008)

89
Please see Figure 5.1.4A for the original network graph in Appendix 5.1.4.
Size: Out-degree; Color: Communities

404

Figure 5.1.15. Emergency Response
90
Community Structure Detection-
Communication Network (Detection Method by Blondel, et al., 2008)

90
Please see Figure 5.1.4B for the original network graph in Appendix 5.1.4.

Size: Out-degree; Color: Communities

405

Figure 5.1.16. Recovery
91
Community Structure Detection-Communication Network
(Detection Method by Blondel, et al., 2008)

91
Please see Figure 5.1.4C for the original network graph in Appendix 5.1.4.

Size: Out-degree; Color: Communities

406
Bottom-up Approach
Cliques
In the previous sections, I explored the process of agency structure formation and
development inside the civil society domain from a macro perspective. Such endeavor
was executed through the examination of weak components, strong components, and
community detections by using the Girvan-Newman method. The development of these
macro-level sub-structures provided insights in terms of understanding the general
behavior of civil society actors in constructing the network as a whole at different points
of time.
One needs to go a step further to discover the process of how general behavioral
trends could emerge from smaller scale of relationships. Therefore, in the following
sections, the discussion will be directed towards a micro perspective to look at how civil
society actors organized themselves in building up network sub-structures and how these
smaller groupings together construct the dynamics of structural evolutions.
There are different ways of examining the micro-level structures in network
analysis. One lens I chose to examine the structure is called the “cliques”. From a
structural analysis point of view, a sub-set of actors can be defined as part of a clique

407
only when the relationships among them satisfy the following conditions: 1) every
possible pair of actors is directly connected by a tie; 2) the clique is not contained in any
other clique (Scott, 2001). In other words, the process in finding cliques is a way in
discovering the existence of groups whose members are “maximally” connected and all
of them are adjacent (directly connected) to one another. Since the networks that I am
investigating took into consideration of who initiated or received a tie, the cliques found
would essentially only take into consideration of the reciprocated connections. Thus, all
actors in a clique were able to reach out to another through one direct step while every
single one of such initiation was being reciprocated by the actor on the other end of the
tie. Recall the concept of “component” that I have discussed earlier, the boundary was
defined by counting all actors that were connected by direct paths or indirect ones
through intermediaries. The point of emphasis for weak components was on the
connectedness of the network structure. Cliques, on the other hand, focus on the
completeness of a structure. I chose to put a stricter constraint in identifying the tightness
of a group of actors in order to understand how denser neighborhoods in the
communication structure were formed from the bottom-up over time. This kind of
“tightness” incorporates a condition that not only all actors in a clique are directly

408
reciprocating to one another’s tie initiating efforts, but also each actor is immediately
connected to one another without the need of intermediaries.
Overall, I examined the existence and the actor composition of this kind of micro
sub-structure for the following reasons. First, it looked further into the more densely
connected areas of a component and depicted the specifics of the sources of an emerging
process of cohesiveness within the central areas of a connected graph. Second, it helped
in understanding how information was being exchanged or projects are being
collaborated among actors with different attribute characteristics.

Pre-Earthquake Stage
During the period before the earthquake, a total of four cliques were being
detected and each of them was composed of three actors. There was certain degree of
overlapping among these cliques. For example, actor #51 was a member who showed up
in all four cliques. Such was a signal that this actor played a “central” role in building
“close” information exchange relationships and was able to gather information from
actors across cliques. Each of the actors #119, #137 and #70 was a member of two
different cliques. One distinguishing characteristic of the clique structure at this period of

409
time was that close clique relationships among civil society actors were mainly triadic
and was strongly diffused by the central role of one actor. Actors also tended to share
memberships across cliques.
The “hierarchical clustering of overlap matrix”
92
from UCINET output at the
actor-by-actor level could be used to depict the numbers of clique memberships actors
have in common. During the period before the earthquake, actor #51 and #119 were first
“joined” together in being closely attached to each other because the two of them shared
two clique memberships together. This joined attachment was further expanded to
include actor #70 as it shared one clique with #51 and #119, and had two common
memberships with actor #51. As the restraining level of the hierarchical clustering
relaxes, actors #137, #20, and #61are being gradually attached to the other clique
members consecutively.
I also examined the level of adjacency of each actor in the communication
network to the clique members. Actor #100 was adjacent to 1/3 of the members in cliques
1 and 2. Actor #19 was adjacent to 1/3 of the members in cliques 2, 3, and 4. Actor #20
was also closely tied to the rest of the cliques. However, when looking across the entire

92
Please see Appendix 5.1.5C2 for UCINET clique output for communication network at pre-earthquake stage.

410
clique participation score table as a whole, one can easily find that it was rare incidence
for actors in the pre-earthquake communication network to have “close” attachment to
the members inside the four cliques found. By “close”, I mean the incidences when other
actors have direct reciprocated connections to the clique members. In general, although
several maximally complete cliques were found in the communication structure, they
seemed to be distant from the actions of others in the network.
A closer look at the specific overlapping patterns among the four cliques can be
examined by using the “clique-by-clique Co-membership matrix”. The occurrence of
overlapping memberships across maximally complete sub-structures offers insights in
determining whether information can be transferred from one group to the other. An
overlapping clique structure also signified actor behaviors that tended to work against
information “blocks” and separation among actors. In this network specifically, each
clique had certain degree of overlapping with another and the highest number of actors
that two cliques can share together turned out to be two. Such a result revealed that small
tightly “joined-together” groups did exist but the information being shared within one
group is also able to be diffused into another group, due to the central role of actor #51.

411
I will now extend the discussion into a graphical context. The purpose here is to
descriptively examine two main aspects of the clique members. One is the distribution of
actors inside each of the cliques in relation to their attribute characteristics, such as
registration status and geographical location. The other aspect is to look at the
distribution of agency action in terms of actor out-degree for actors inside the different
cliques found. Both of these aspects of investigation will also be displayed along with the
communities detected by the Girvan-Neman method. This is to show how the macro and
micro level of structuration can be integrated in light of actors’ relationship-building
behavior and attribute characters. Depicting the structural dynamics in terms of these
perspectives is constructive step towards understanding the sources of change agent that
directs the behavior of the network as a whole over time. Figures 5.1.17 depicts the
graphics of pre-earthquake communication network.

412

Figure 5.1.17. Combined Attributes of Communication Network (Pre-earthquake)
Recall that actors #51 and #119 were the first two being “attached together” based
on the number of clique memberships they had in common. From figure 5.1.17, we can
observe that both of their practices at this period time were based in Sichuan Province.
Clearly, operating within the similar location provided more opportunities for the two
actors to interact and communicate with one another to build stronger ties. Actor #70 and
#137 further joined to be close with #51 and #119. The graphics showed that although
both were registered organizations, none of them turned out to have operating
headquarters in Sichuan Province. Actors #20 and #61were the last two in joining the
sequence of “closeness” in terms of the number of common clique membership with
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117 118
119
12
120
121
122
123 124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95 96
97
98
99
Size: Registration
Color: Community (Partition=4 in the
connected network)
Shape: Location
 circle: Sichuan-based
 square: non-Sichuan

413
others. In this case, neither of them were members of another clique. Thinking in terms of
communication and information exchange functioning, it was clear that the role of actor
#51 cannot be ignored. The fact that it appeared as a member of all four cliques
demonstrated that it held the power in terms of facilitating communication and transfer
information among members not just within but also across clique boundaries. This actor
could also be seen as a “change agent” in tiding the “threads” among actors in different
cliques and thus playing a role in promoting and maintaining the cohesiveness of these
core actions inside the main community of the structure. Observing the distribution of
attributes across the members in the four cliques, note that #119 was the only un-
registered actor. It was a Hong-Kong based nonprofit organization that had dedicated
itself in community development in Sichuan before the 2008 earthquake. It also became
the first civil society actor in jointly “attaching to” the key actor #51 in sharing
memberships in different cliques. Thus, the role of foreign-based (but domestically
registered) nonprofit organizations operating in mainland China in terms of facilitating
information exchange cannot be ignored.
Figure 5.1.18 graphically illustrated the agency action in terms of out-degree
alongside with community structures during the time before the earthquake.

414

Figure 5.1.18. Pre-earthquake Communication Network (Out-degree and
Communities)
The size of the nodes represented the level of out-degree of actors while the color showed
the community detection from the Girvan-Neman method. Again, we can see that all
clique members belonged to the largest community sub-structure except for actor #20. In
fact, this actor was a grassroots nonprofit organization based in Beijing. Its
communication activities were situated in the same clique with #51 and #137. While the
former was a formal organization with established field office in Sichuan, the latter was
one that operated in Beijing as with actor #20. The structural composition of this clique
(#51, #20, and #137) had two critical implications in discovering the behavior of civil
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119 12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Size: Out-degree
Color: Community (Partition=4, in the
connected network)

415
society actors before the earthquake. The first implication was that it further
demonstrated the facilitating role of #51 not only in connecting actors across institutional
boundaries in terms of registration status but also across geographic locations. Since actor
#20 was not sharing membership with any other clique members and was the only one
from a peripheral community (colored in black in figure 5.1.18) that “made it” into one of
the clique structures, the existence of #51 and #137 would act as mediators to promote
the passing of information from #20 to other clique members. From this perspective,
there was a sense of “dependency” of #20 on #51 and #137. However, this structural
characteristic of #20 should not be confused with its role and being interpreted as its
weakness for such “dependency”. From the out-degree graphic display, we can see that
the communication tie-building action of #20 was critical in connecting those in its own
community (colored black) to the larger community (colored in red) and thus to the
central actor of #51. In a sense, actor #20 can also be categorized as a “change agent” in
facilitating certain information exchange ties that would otherwise not be possible among
actors across community boundaries. A further observation in relating out-degree with
clique members was the fact that all of clique members occupied relatively high level of
communication outreach activity positions. Actor #51 has the highest out-degree measure
at this period of time.

416
When a policy measure is aimed at spreading important pieces of information
among closely related actors and thus generating further “rippling effect” of diffusion
across the network, the particular structural embeddedness thus discussed for actor #51 as
well as those that it shared clique membership with are critical factors to be taken into
consideration.
Emergency Response
The focus of this section is on the period shortly after the disaster and the primary
purpose is to observe the changes in the actors’ structural behavior at the micro-level and
how it relates to the macro-level sub-groupings. I first discuss the output results generated
from the clique analysis. Then I will relate such findings to attribute characters and
agency action from a graphical point of view.
During the emergency response period, the number of maximally complete sub-
structures drastically increased from only 4 to 25. The number of actors inside each
clique also increased from a maximum of three to four. This indicated that civil society
actors became active not only in expanding clique boundaries but also in developing the
intensity of their ties towards increasing number of others through direct and reciprocated
communication relationships. Such behavior also demonstrated a tendency towards

417
building expansive long-term relationships at the time of crisis. On the other hand, the
increasing number of cliques also indicated a type of behavior that favored
communication activity and diversification of ties across the entire network. Compared to
the clique structure during the period before the earthquake, the short term response stage
is characterized by the tendency that close ties were no longer being built around a
limited number of groupings. Civil society actors activated their communication
capabilities across various kinds of close relationships. In other words, the agency
sources of passing over and receiving information were no longer being dependent on the
action of only a few closely connected actors. Different channels of information
exchange were being unleashed. Such awareness of active participation in building
available opportunities on the actors’ part can be argued to be the primary structural
foundation showing the coping behavior of Chinese civil society in times of extreme
uncertainty and stress.
The “hierarchical clustering of overlap matrix”
93
at the actor level provided
insights in terms of the “joining sequence” of how actors were being closely attached to
one another through sharing clique memberships. It is apparent from the results that
actors #3 and #51 shared the most number of cliques together. This meant that they

93
Please see Appendix 5.1.5D2 for UCINET clique output for communication network at emergency response stage.

418
played key role in facilitating communication and information flow across the clique
boundaries. At the next “joining” level, actors #24 and #119 became further attached to
#3 and #51. Among these four actors, note that #51 and #119 were also the first two
actors to share the most number of clique memberships together during the pre-
earthquake stage. After the disaster, they were able to sustain their role in tying up the
linkage across different cliques. Also note that actor #3 formally came into existence as a
social group after the earthquake. Together with actor #24, both originated and operated
as grassroots civil society actors with a geographic focus in Sichuan Province. This
signified that “locally-grown” groups and nonprofits started to emerge in playing an
important part in communication relationship-building and prevention of grouping
segregation among closely knit-together sub-structures.
Examining the results showing the overlapping clustering level, I found that this
was the stage where actors showed certain “distances” in their behavior for clique
membership sharing. The first set of split behavior among the three sets of actors: 1) #109
and #115; 2) #119, #24, #3, #51; 3) #123 and #57. As the stringency level of clustering
was being relaxed, the twenty five clique members eventually joined together as a whole.
In terms of information sharing, this type of clique formation showed some primary
diversification in clique membership choices and could be categorized as one sub-

419
grouping trait for civil society action behavior shortly after the earthquake. However, the
story had another side to it. This kind of choice-expansion activity was also accompanied
by a “disruption” in the continuity of “joining sequence” when one closely observing the
cluster overlapping structure shown in Appendix 5.1.5D2. As the constraint level of
clustering overlapping started to relax, we can see that the clique co-membership
participation behavior initially separated the actors into three types of “closeness”. And
as actor #100, #37 and #7 gradually joined with the main common membership group,
there was still a divide between two groups of actors whose information sharing activities
were basically not reachable to each other. It was not until the attachment of actor #76 to
the clique composed of #137 and #24 that the information pathway across the divide was
being bridged. In other words, pieces of information might not be as easily passed
through across cliques as the case when the “joining sequence” was more integrated
smoothly as a whole when before the earthquake.
Moving on to the next level of analysis illustrated in the “clique-by-clique actor
co-membership matrix”. This matrix depicted the number of members in common
between a pair of cliques. Compared to the period before the earthquake, the matrix
94
for

94
Please see Appendix 5.1.5C3 for the pre-earthquake matrix and Appendix 5.1.5D3 for the emergency response
matrix.

420
the emergency response stage not only turned out to be larger in size but also in terms of
the way that actors shared their membership among cliques. Note that there appeared to
be a large number of zeros between pairs of cliques and this meant that there were
incidences when cliques did not share any members together. However, for members
from clique 1 to clique 14, all pairs of cliques were connected to each other through
sharing at least one actor’s participation. For the rest of the 11 cliques, information could
at least be transferred from a selected number of others, thus leaving none of the cliques
completely isolated. What could be generalized here is that although there were
“disruptions” in information sharing among particular pairs of cliques, communication
contacts from one clique member to another can be reached throughout the 25 cliques,
either directly or indirectly. This means that on the one hand, actors at this period of time
tended to pick and choose others to develop closer communication ties. On the other
hand, the line of information sharing was not completely broken throughout all the
cliques. No one clique became isolate as a result of such sub-structure formation.
Let’s now graphically examine some of the characteristics of the distribution of
actor attributes and out-degree activities among some of the key clique members. From
the hierarchical clustering analysis at the actor level discussed earlier, I found that actors
#119, #24, #3 and #51 were among the first four in the attachment sequence to share the

421
most number of cliques together. Therefore, in the examination here, I will mainly focus
on how their micro-structuration behavior descriptively related to actor attributes and
agency action represented by out-degree measures. Figure 5.1.19 depicts the actors’
registration status and location within the short term response communication network
structure.

Figure 5.1.19. Emergency Response Communication Network (Combined Actor
Attributes of Registration, Community, and Geographic Location)
The nodes colored in red were all detected to belong to the same community. As we can
see, the majority of the actors became part of this larger community, with the exception
of a few “hangers” located on the outer periphery of the structure. First of all, from the
Size: Registration
Color: Community (Partition=10)
Shape: Location
 circle: Sichuan-based
 square: non-Sichuan

1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128 129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99

422
shape of the nodes illustrated in the graph, all four actors (#119, #24, #51, and #3) had
field offices established and operated within the earthquake impacted area in Sichuan
Province. This indicates that shortly after the earthquake, some of the most important key
players in facilitating information flow among tightly connected sub-structures were
either “locally-grown” or had been practicing in the region for a long period of time and
developed culturally-rooted ties with local communities. Secondly, the size of the nodes
tells us that #119 and #3 were non-registered actors (larger in size). What is encouraging
regarding the emergence aspect of civil society agency structure lies in the existence of
actor #3. As a newly formed social group immediately following the disaster, the actor
quickly became a critical member across multiple numbers of cliques in terms of
developing direct and reciprocated communication relationships with others in the
network. Its non-registered status did not seem to prevent it’s proactivity in reaching out
towards others. At the same time, the institutional status difference was not perceived as a
barrier for others in the network to build strong clique connections with the actor #3.
In fact, the sharing of membership activity was consistent with the execution of
agency freedom in terms of out-degree activity. From figure 5.1.20, we can see that those
who first joined together to be close in terms of sharing clique memberships together
were also the ones with higher level of out-degree measures.

423

Figure 5.1.20. Emergency Response Communication Network (Out-degree and
Communities)
Two implications can be drawn from this finding. One is that shortly after the
disaster event, those actors who were engaged in close communication connections with
others through clique participation did value their agency freedom by actively initiating
ties with others both within and across clique boundaries. Secondly, such active behavior
was also being reciprocated or being “treasured” by others, and the result was a drastic
expansion of maximally complete sub-groups within which civil society actors could
share information directly with one another. These two action processes built up the
foundation for the rise of the key actors in maintaining the dynamics of cross-clique
communication and information sharing.
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124 125
126
127
128 129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62 63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Size: Out-degree
Color: Community (Partition=10)

424
Long-term Recovery
So far, I have shown that there was a dramatic change in terms of the number of
cliques when comparing the pre-earthquake stage and the emergency response stage. One
of the key conclusions being drawn from earlier discussions was the tendency for actors
to diversify choices in developing close communication relationships and at the same
time reaching out across the entire network rather than focusing on their immediate
neighborhoods. In order to determine if there were any signs of institutionalized
structural changes based on such dynamics revealed from a short term response, I further
examine what happened over time during the long term recovery stage.
From the output results, the number of cliques increased from 25 in the previous
stage to 47 during the recovery stage. First of all, with the exception of the first clique,
the number of actors inside each of the cliques remained consistently around three to
four. The largest clique being found was composed of five members. Compared to the
maximum member composition from the emergency response period, there was a small
scale of expansion within cliques. Secondly, the sheer increase in the number of cliques
in general indicated that civil society actors kept becoming intensely engaged in “close”
communication tie-building with others across the network. This shows that not only the

425
activity level of information exchange has not been winding downwards over time, but
the amount of agency “energy” in making close connections was also sustaining. This can
be indicated as a sign of tendency or willingness to institutionalize relationships being
built and role maintenance as perceived by the civil society actors themselves.
In order to depict the key actors through an attachment sequence determined by
the number of cliques members shared together, I look at the “hierarchical clustering of
overlap matrix
95
” at the actor level. Note that actor #3 and #24 became “closest” in terms
of sharing membership in 8 cliques altogether. Actor #51 and #61 further joined the
sequence in being close as the clustering level is relaxed at the next two levels. As the
clustering level being further relaxed to less stringent levels, more separate cliques can be
found. For example, at the clustering level of 1.000, four groups of actors tended to be
loosely connected to one another but at the same time, closely joined together in terms of
the number of clique memberships they share. Compared to the emergency response
stage, at this same level of clustering constraint, I found that there were three such groups
in the joining sequence. This indicated that when given a particular level of overlapping
clustering, there were slightly less incidences of common membership sharing across
actors during the long term recovery stage. Actors at this point of time behaved in a way

95
Please see Appendix 5.1.5E2 for UCINET output.

426
that exemplified a sense of “boundary-protection” when it came to communicating across
cliques. This can be further demonstrated by the results shown in the “clique-by-clique
actor co-membership matrix”. Examining the general distribution of degree of
overlapping between pairs of cliques, there is a clear pattern showing differing levels of
disconnectedness among different cliques. The group with the most connectedness
contained those from clique #1 to clique #8, all of which share at least two actors
together. This means that information being distributed throughout these cliques would
have a higher chance of reaching across with more efficiency. Actor participation among
these cliques focused comparatively more on both diversification and boundary
expansion. These efforts tended to work against the tendencies to divide and factionalize
the network. At the next co-membership sharing level, the group contained those from
clique #9 to clique #34. Most of the pairs of cliques in this group shared at least one actor
together, but not as many as those in the first group. Information transferring pathways
across cliques were still open but there were less alternative actors to turn to if one failed
to share a piece of information with members at another clique. Therefore, compared to
the “closeness” of cliques in the first group, members the second group tended to have
more distant and estranged relationships among each other. The third group contained
those from clique #36 to clique 47. The most distinguishing feature of the connections

427
among these cliques was that a significant number of cliques were without any common
memberships. This indicated that information sharing across members of these cliques
would encounter barriers. This is because while members within cliques were closely
connected to each other, no ties were being built across cliques. Information could be
circling inside one clique but less likely to be transferred over to another clique domain.
The level of separateness for this third group of cliques was even higher than the previous
two groups.
However, regardless of the emergence of certain degree of “estrangement” during
this period of time, the connections among all the 47 cliques remained open. No isolated
cliques were being found across the co-membership matrix. Therefore, information
exchange and communication in general was still possible across cliques, but with the
need of intermediary cliques, thus not as easily accomplished as the cases in the earlier
two periods.
Moving on to relate the clique formation behavior with actor attributes and out-
degree activities, similar patterns can be discovered when comparing to the emergency
response period. Take some of the key actors in facilitating communication across cliques
for example, such as actor #3, #24, #51, and #61, all of them turned out to have

428
established locally-rooted offices and field practices. In terms of registration status, actor
#3 was the only non-registered one. Actor #61 originally came into China as a foreign-
based nonprofit organization but gained registration status only after the 2008 Wenchuan
earthquake. Over the years leading up until the 2008 earthquake, the actor developed
culturally rooted practices in the area of poverty alleviation within the local communities
in Sichuan Province. The organization further expanded its area of practices to disaster
response and long term recovery in the form of assisting community development for the
rural villages that were significantly impacted by the earthquake event
96
.
Figure 5.1.21 shows the out-degree activities of civil society actors during the
long term recovery stage, the four key actors in information exchange across cliques (#3,
#24, #51, and #61) also maintained high levels of agency actions in initiating
communication relationships. One can see that actor #3 had the highest number of out-
degree among those identified inside the large community (colored in red). It was further
followed by actor #51, #24, and #61 consecutively in terms of the level of agency action.

96
Chen, Taiyong. “The Actions and Models of ‘#61’ during the Wenchuan earthquake response and recovery”. (This
internal document was provided to me by the informant of actor #61 upon my visit to the organization’s field office in
Chengdu, Sichuan Province.

429

Figure 5.1.21. Long-term Recovery Communication Network (Out-degree and
Communities)
Therefore, the findings in this section can be summarized as follows, with regard
to the bottom-up structural formation in the communication network. First, there was a
tendency towards high level of micro-structure expansion. The first stage was illustrated
by a tendency for diversification and boundary extension in forming different cliques for
information exchange. The second stage was characterized by continued expansion of
closely tied communication relationships and a development of certain degree of
stratification across the micro-structures. Information could be easily transferred
throughout certain set of cliques than others. Participation and involvement in the local
communities was a valuable asset for key actors who played facilitating role in bridging
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129 13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25 26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Size: Out-degree
Color: Community (Partition=10)

430
the boundaries across cliques. This can be demonstrated by observing figure 5.1.22
shown below. Note that all of the four actors who shared the most number of cliques
during the recovery stage (#3, #24, #51, and #61) were all found in one community with
the same circle shape which represented those who conducted activities inside the
Sichuan Province.

Figure 5.1.22. Long-term Recovery Communication Network (Combined Actor
Attributes of Registration, Community, and Geographic Location)
Secondly, outreach communication tie-building behavior was being sustained
among key actors who became highly active in being mediums for cross-clique
information exchange. Two unique cases were worth looking into when observing the
clique formation results across the three time periods. One is the emergence of actor #4
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129 13
130
131
132
133
134
135
136 137
138
14
15
16
17
18
19
2
20
21
22
23
24
25 26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Size: Registration
Color: Community (Partition=10)
Shape: Location
 circle: Sichuan-based
 square: non-Sichuan

431
started sharing clique membership with two other newly emerged civil society actors,
including actor #123. While actor #123 remained to have tight communication
connections with others during the recovery stage, the involvement of actor #4 in cliques
was only short-lived during the emergency response period after the earthquake. Another
unique case was the emergence of actor #97 in participating in two cliques during the
recovery period, which involved actors #3 and #51. The actor was not sharing
membership in any of the cliques found for the emergency response stage. It is therefore
important to investigate the driving factors for the actions of #97 in terms of forming
close communication ties within the domain of civil society.

The Case of Actor #97
When it comes to within-sector institutional development, the original emergence
of the actor #97 could be separated from its connection with actor #51, a formal domestic
NGO specializing in supporting and funding projects for the development of newly
formed civil society groups/organizations. The particular type of interaction between the
two actors was one that can more precisely characterized as project support through
“collaboration”. The initial connection was made by actor #97 applying as one of the

432
“incubation organizations” that would be eligible for a one-year funding from actor #51
to execute its pre-designed project at the local community level. As was recalled by the
organizer of #97, this opportunity also came along with the intention of actor #51 to
develop Sichuan-based nonprofit groups/organizations after the 2008 earthquake event.
This way, the collaborative relationship between the two actors was motivated by mutual
interests in the long term recovery at the local scale. In addition to this type of project-
supporting collaboration relationship, another type of interaction between actor #97 and
other civil society actors involved assisting others in their health-related projects at the
community level. One example provided by the organizer was its assistance in building
self-help strategies within community networks for elders suffering from diabetes
97
.
Regarding the aspect of long term growth for the actor #97 as a whole, its
interactions with actor #3, particularly in the form of participating in the latter’s training
platforms built during the long term recovery stage played a key role. This role contains
two aspects. On the one hand, the training platform provided by actor #3 facilitated
further communications and exchanges of information between #97 and other civil
society actors. On the other hand, the training programs themselves were also perceived

97
For detailed account, refer to Appendix 5-1-CaseSG97-4 and interview account in SG97-01-07 in Appendix 5-1-5-13.

433
as an educational tool for participants to further understand the “idea of civil society” and
related concepts, such as “citizenship”.
But the positioning of (actor SG3) is different, maybe in terms of a different
functioning field and group of people, and what they are doing is related to
trainings on the topics of civil society and civic awareness. I also participated in a
few of their trainings and I felt that they have a particular focus on the concept of
citizens. But they won’t emphasize too much on it because there is this practical
side that we all have to face in the field. So they will also provide us with some
case studies or focus group discussions. They are still trying to find a good way to
do this as well. (SG97-01-08
98
)
This reflected one particular characteristic of the actions taken by actor #3 when it came
to long term recovery stage. And this was the transitioning towards training platforms not
only for the purposes of mediating “bonding relationships” among civil society actors,
but most importantly, for building up a common conceptual foundation specifically
understood as Chinese “civil society” within which groups/NGOs field practices were
based upon. This way, the emergence process of civil society in the post-crisis context
maintained its “momentum” in both the specific devotions of each group/organization
actor over time and in being reinforced through the functioning of actor #3 providing a
“nurturing environment” for the sustainability of civil society actors at the level of
practice and awareness.

98
For original Chinese script please refer to Appendix 5.1.5.14

434
LU: So what impressions would you have for the information platform initiated
by Actor #3, or what kind of platforms would you hope to see?
SG97-01: It depends on how they position themselves. I also talked to (NGO3-03)
about this and their current positioning might be restricted by the program itself.
They might not have a particular focus, and the topics are all scattered around.
But their goal is to construct a communication platform. That’s why the way they
arrange certain topics is less systematic. But when we participated in the training
provided by (actor #134) on topics related to action research, that one is relatively
systematic. (SG97-01-09
99
)

Interactions between Domestic and International NGOs
Another trait worth noticing regarding the interactions of actors within the civil
society domain was reflected through the organizer’s (NGO97-01) account of the
activities and interactions between domestic and foreign NGOs after the earthquake event.
One of the most recognizable differences between the actions of the two types of civil
society actors can be traced back to their diverging ways of operation and focus of
practice. The organizer openly admitted that most of the interactions the actor made
inside the civil society “circle” were with other domestic NGOs, rather than with foreign
ones. It was mainly due to the latter’s unique “operating strategies and management”
apart from the functioning of their domestic counterparts. However, when a foreign NGO
became formally registered inside China established its field offices, such as the case of

99
For original Chinese script please refer to Appendix 5.1.5.15.

435
actor #61, the operation procedures would be “different” in terms of being adjusted to be
more customized towards the local cultural and social conditions.
Another difference between the functioning of domestic and foreign NGOs is
revealed through the motivations for sustainability over time.
There were many foreign NGOs who mainly focused on response assistances or
aid, and those were inherently short-term. They will leave as long as their job was
done because that was how they positioned themselves in doing. For our domestic
NGOs, especially those working in Sichuan Province, are all localized and long
term. Therefore, they are different. Many of us are in the transitioning period now,
basically heading towards focusing on local development. The local NGOs care
more about local people, so they must be transitioning towards works related to
long term development. The earthquake itself was just an emergency event, and
because of this, many of the foreign NGOs were drawn here. Whether or not they
are domestic or foreign NGOs, they all had to leave eventually. It’s been a long
time since the event first happened, the main focus is different from back then.
You can see that there are fewer NGOs remaining now, which is necessary,
because the response period is over. There was no need for them to stay. So the
ones who chose to stay were mainly those devoted to long term development at
the local level. They were either Sichuan local NGOs or domestic ones, either
way, their point of emphasis must be related to long term human and social
development in the area, rather than on emergency assistances. (SG97-01-10
100
)
Thus, the source lies in the functioning orientations after the disaster event. Foreign
NGOs are perceived to perform emergency response and short term-oriented disaster
relief tasks to alleviate the immediate impacts of the earthquake. Domestic NGOs,
especially those established in Sichuan Province after the event, were more grass-roots in

100
For original Chinese script please refer to Appendix 5.1.5.16.

436
nature and also aimed at long term involvement in the area. Again, the source for those
domestic civil society actors to remain beyond short term relief period and maintaining
recovery actions arose from their motivations in transforming into development-oriented
organizations while devoting their focus on the social and human aspect of the recovery.
Therefore, the fundamental positioning of how foreign and domestic actors saw their
roles through the short term into the long term periods was perceived to be a contributing
factor in their divergence in action. This also provided a primary explanation for the
active participation of actor #97 in different clique structures over the long-term recovery
stage. The increasing awareness of the inter-dependency among the civil society actors
toward capacity development of grassroots actors also contributed to its sharing of
membership with actors such as #3 and #51 during the long term recovery stage.

Communication Platform Building (SG97)
One factor that needs to be noticed by the illustration of the emergence stories of
different civil society actors is the insurgence of the function of communication platform
building through disaster response and recovery. Such a functioning in the midst of great
uncertainty and stressful conditions could perform as a shield from a tendency for the

437
society to lose its strength to “stand-up” against the troubles brought about by great
disasters. This is because the platform often provides a level ground for relationships
among actors to emerge, develop, flourish, and sustain over time. I now use the example
of actor NGOLF to qualitatively illustrate how this point can be applied to understand the
origin of clique expansion in the communication network.

The Case of Actor NGOLF
The long term recovery activities chosen by the actor NGOLF in general, from an
outset, resembled those performed by actor #3, such as the envisioned role of functioning
as a communication platform, provisioning of training programs and resources for
communication. When it came to the actual practices of field activities, differences did
emerge in terms of ways that each actor positioning itself in the civil society field. First
of all, actor NGOLF focused its attention specifically in “Youth and family education”.
And over time, the actor had already accumulated some skilled “teachers and volunteers”
who could manage the educational training programs. For example, in one of the
collaborative projects implemented with another local NGO, some of the teachers for
their youth-related classes were volunteers from NGOLF. Or sometimes the actor would

438
also provide training for teachers from the former. At the same time, its partner’s role
was to administer and manage the overall functioning of the program.
Thus, in terms of field of specialization, actions of NGOLF were oriented towards
servicing a particular sub-group of the local population and the nature of its activities
were more concentrated in one area of disaster recovery. The area of focus defined by
actor #3 (SG3), on the contrary, was relatively broad in terms of its service targets. Recall
that one of the primary goals defined by actor SG3 at the beginning stage of its formation
was to service all nonprofit groups and organizations that came to Sichuan performing
disaster recovery related activities. In this regard, the elaborated account by the
participant of NGOLF illustrated the specifics of its action concentration:
We have one key point of focus regarding the groups of people we intend to serve,
and our direct service targets are youth and adolescent. But the group of people
that we encountered the most was those volunteers and teachers, more in touch
with teaching resources, including in the disaster areas. What we have done the
most was in making contacts with their schools or their educational unit inside the
government, making connections for them. And of course there will be times that
we have to get in touch with local civil society groups that are conducting such
kind of services, as well as schools in the area. This is because our focus of
service is on youth. Although we ourselves would organize some activities for
this group of population, we feel more towards playing the role of ‘propeller’ (in
encouraging all these other entities to engage in communication and
collaboration). After all, we don’t have too much resources to do too many things
at the same time. But what we can do…we can create a platform by making

439
connections with many different systems, to facilitate communication among
them. (NGOLF-01-04
101
)
This account depicted clearly of what consisted the actor’s functioning of
“communication platform” as compared to that performed by SG3. Two points are worth
elaborating here. First, the communication platform formed by NGOLF was primarily
centered on the issue of youth education. Thus, the actor facilitated the communication
not just among civil society actors, but also across schools, other local social service
entities, as well as those in the government, such as branches in the educational system.
The functional purpose of building such a platform originated from a specific focus on
youth development. For actor SG3 on the other hand, its platform was less issue-oriented
and was not built on targeting a particular sub-group of people in the society. Rather, the
function of actor SG3, as described by the participant of NGOLF, incorporated a broader
and characteristically different population as its servicing targets. Its goal was primarily
in facilitating the communication ties among civil society actors in general. Figure 5.1.23
below provides a graphical illustration of the role performed by the two actors in relation
to their position with all other civil society actors in the field.

101
For original Chinese script please refer to Appendix 5.1.5.17.

440

Figure 5.1.23. Types of Communication Platforms (Long-term Recovery)

NGOLF
Volunteer
s

NGO33
SG4
NGO124
Youth Development
Communication
Platform
Schools
Social
service
providers
NGO#

441
Understood from the perspective of the participant NGOLF-01, his way of
defining the function and role of itself was one characterized by promoting and
supporting the communication connections among the different parties involved in youth
development and education. While its own emergence was supported by its
communication and collaboration relationships with schools, volunteers, and other social
service providers at the local level, its action to expand the platform participation started
with its connection toward other civil society actors that functioned in the same area of
practice, such as its communication ties with actors NGO33 and SG4. In other words, its
actions entailed a background motivation of raising the social awareness of the issue at
hand and performing as a “mediator” among the different social and state actors to enable
them to recognize their role in making a contribution to the capacity development of civil
society actors. To some extent, another way to look at the action of NGOLF is its
function in enhancing one form of social capability, which particularly entailed youth
development during the disaster recovery stage.
As a general reflection on the development of Chinese civil society after the
earthquake event, the participant held a firm view that the disaster performed as a catalyst
particularly for the NGO development in the country. Before the earthquake, “none of us
had heard of the term ‘NGOs’…and it would be unimaginable if we were to go into the

442
rural communities to help them with something by introducing ourselves with this title
and identity”. After the earthquake, however, things changed drastically, especially in
terms of the level of general acceptance of the name of “NGOs” and the functions being
performed by them. “No matter how remote a place is”, as was recalled by the participant,
“when we say we are volunteers coming here to provide (certain kind of) services, it is
started to become relatively easy for the local people to accept us now as they have at
least heard of who we are and what we do”. Therefore, the disaster event could be
thought of as an impetus for not only the growth of civil society groups/organizations, it
also performed as an accelerator in promoting the social awareness of what civil society
actors are and what they do in relation to the ordinary lives of the general public
102
. Such
impetus was demonstrated by the increasing amount of civil society actors engaged in
building close information exchange ties with each other through the formation of cliques.
The sustenance of these close relationships came from a spirit of dedication in each of the
actor’s chosen areas of practice as well as the persistence in their agency action to keep
reaching out toward others.

102
Detailed account refer to Appendix 5.1.CaseNGOLF.2

443
Chapter 5

(Part II)

Collaboration and Sustainability of Agency Structures
Formation of the Sustenance Structure
In this second part of this chapter, I examine the collaboration network
environment, paying special attention to investigating it alongside with its
communication counterpart. This is because the project collaboration behavior
represented a level of agency action that works together with communication behavior
helping to enhance the strength of civil society to explore its capability with dedication of
efforts and time. In the context of encountering extreme distressed situations such as the
2008 Wenchuan earthquake, I have shown that it was the agency freedom to initiate
communication that built up the foundational structure of connection. It is therefore
necessary to find out if there was any sustenance for such agency structures, thus setting
up the conditions for the institutionalization process. In this study, I investigate the
collaboration efforts that led to building up of the sustenance structure.

444
Reciprocity
Comparing the reciprocity measures for communication and collaboration
networks (Table 5.2.1), we can see that of all the pairs of actors that had connections, the
percentages of the pairs had reciprocated connections across the three periods of time
were generally less for collaboration ties than that of communication ties.
Table 5.2.1. Comparison of Reciprocity Measures before and after the Wenchuan
Earthquake (Communication and Collaboration)
Time Period Communication Reciprocity Collaboration Reciprocity
t1 (Pre-earthquake) 0.0900 0.0556
t2 (Emergency Response) 0.0925 0.0856
t3 (Recovery) 0.1067 0.0989
Before the earthquake, the percentage of reciprocated project collaboration ties were
about 4% less than that of the communication reciprocation. Shortly after the disaster,
while the reciprocity for communication network experienced 0.25% increase, the
reciprocity for collaboration networks experienced 3% increase from 5.56% to 8.56%.
Although the measures in themselves are still lower in quantity than that of those for the
communication ties, the higher amount of increase again indicates the high level of
motivation and commitment of civil society actors to make contributions to assisting the
response process. This is because the activity of project collaboration inherently suggests

445
a higher level of dedication on resources, time, and possibly intended field-expertise
development on the participating actors’ part (Gazley 2008).
This kind of increase in reciprocated collaboration connections in such a short
period of time after the earthquake primarily signified two factors regarding the nature of
civil society as a whole in the Chinese context. One is the agency nature on the part of
civil society actors not only to establish themselves as informal self-organized groups or
formal organizations, communicating with each other through times of crisis but also
their capability in building collaborative connections. I argue that this characteristic is
unique to the Chinese case. In circumstances with social structures that were more
vulnerable when reciprocity was rare and actors were “sparsely” connected such as the
network structures before the 2008 Wenchuan earthquake, the standard expectation is the
structure as a whole would be less likely to withstand an outside “shock” such as a
disaster event. On the contrary, this not only did not happen in the Chinese case but most
importantly, the event explicitly brought forth a sense of self-awareness of the agency
role on the civil society actors’ side. The enactment of such agency further established a
foundation for “resilient” structures characterized by increasing percentage of
reciprocation in collaboration ties that were more likely to be enduring over time.

446
A second nature of the Chinese civil society can also be revealed. It is the actors’
growth-oriented focus even during the short term response period. Immediately after a
disaster, attentions are normally diverted to emergency assistance duties and the long
term life-span of the actions and interactions among group/organization actors are rarely
being investigated in an in-depth manner. In the Chinese case, the persistence of
reciprocal communication and collaboration behavior presented some of the primary
evidences showing the existence of an increasing willingness for Chinese civil society
actors to engage each other with a long-term perspective. There was a sense of
persistence through the actions of these actors. Clearly, this commitment has been carried
on to long term recovery stage where the reciprocity measure increased to 9.89%.
So far, I have investigated the idea of “institution” as the emergence and
development of a set of mutually accepted relationships that together constructed the
social structure within which group or organizations reside. So far, what this study
suggests in particular is that the actors’ decision-making behaviors are not only based on
their embeddedness in the structure but also their active contribution in the formation of
the structure itself. In the field of urban planning, the concept of “institutions” has often
regarded as a set of rules that will constrain the “freedom” of actors, but at the same time,
providing stability of functioning for actors’ social life. Such is essentially an equilibrium

447
perspective of what constitute as institutions. And this point of view may have been
remiss of the factor of change into the theoretical picture. A paradox thus arises when
theorists are trying to explain or make sense of how a structure based in stability or
equilibrium can at the same time provides opportunities for change (Giddens 1979). The
key to resolve this seemingly contradictory “paradox”, I would argue, is by shifting the
focus instead on the dynamics of decision-making of actors within the structure, thus
making the feedback loop between structure and action an explicitly identifiable process.
It is from this line of logic that I bring in the concept of “institutionalization”. In this
context, it is defined as a process from which institutions emerges, thus putting an
emphasis on both procedural side and the structural outcome sides of the “paradox”. If
one act of initiation of a unidirectional tie or connection from one actor to another can be
understood as the primary starting point from which institutional structure arises, the
process from which the relationship become reciprocal, I argue, will be essential in
examining the actual steps of institutionalization, thus integrating “change” and stability”
in an explicitly observable way.
The current results of this part of the study revealed that there are different
observable types of institutionalization that can be utilized to understand the dynamic
state between “change” and “stability” of the paradox of institutions. And within these

448
various types, there are degrees of institutionalization. The reciprocation of ties in this
context can show us the different stages of institutionalization of communication and
collaboration relationships among actors at various points of time. First of all,
communication relationships in this study, were defined as information exchange
connections among actors and generally did not involve a deeper commitment on the side
of actors. Collaboration relationships, on the other hand, involved further binding
agreement on the sides of both actors, such as jointly managing and carrying out field
projects or programs that aim to provide long term benefit for particular local
communities after the earthquake. Therefore, the two types of connections yield
structures that could be qualitatively different in terms of the functions of the intended
performance.
The process of institutionalization, accordingly, can be categorized as these two
different types and each with varied degrees of evolution. As the reciprocity measure
increases for both networks, the pace of an initial stage of institutionalization can be
made observable. Differentiating such paces of change occurring in the targeting
structural processes may be necessary to link institutional change to policy
implementation for disaster preparedness and mitigation purposes. This necessity can be
exemplified in two consecutive stages of planning policy making and implementation.

449
The first step is to determine whether the policy aims to increase the efficiency of
information flow and communication, or to facilitate the opportunities for project
collaborations among the actors in the civil society. Once the types of targeting
institutional structure are set, the second step will involve using the reciprocity measures
for each type to understand what exactly has been going on in the evolution process. For
example, the increase in reciprocity from before the earthquake to shortly after the
earthquake for collaboration networks would suggest that actors were not only at a stage
of engaging high level institutional-building activities but also at a higher level of
motivation to disaster recovery. When more attention is paid to the social aspect of
recovery in the longer term, policy can be formulated to facilitate the opportunities and
enhance the capabilities of civil society actors to be involved in the local communities
that experienced weakened social support systems as a result of the disaster event. The
timing of each stage that pointed to more dramatic levels of change, such as the periods
before and right after the earthquake, can also be informative in providing signals to
policymakers in when to implement certain interventions for reaching the intended
outcomes.

450
Transitivity
Pre-Earthquake
Now, I will look at the transitive triplet formation for project collaboration
networks over time. Table 5.2.2 shows the comparison between the results for
communication and collaboration networks over time.
Table 5.2.2. Triadic Relationships in Communication and Collaboration Networks
(Transitivity Measures)
Communication Collaboration

  ,, AB BC AC

 
,, AB BC anything

Transitivity
  ,, AB BC AC

 
,, AB BC anything

Transitivity
t1 283 902 31.37% 19 84 22.62%
t2 4328 12458 34.74% 148 765 19.35%
t3 6101 16067 37.97% 300 1538 19.51%
Before the earthquake, there were only a total of 19 transitive triads in the collaboration
network. And the number of cases where a single link could complete the triad was 84.
The pre-earthquake transitivity for the collaboration network is 22.62%. Out of all the
relations that easily could be transitive, only 22.62% actually were. When the measures
are compared with those from the communication networks, there were clearly less
transitive triplets structures when it comes to project collaboration, particularly during the
time before the disaster. This can be explained by the fact that the required commitment,

451
efforts, and resources are often much greater for forming and carrying out a collaboration
project than that for communication with information exchange purposes. Thus, the lower
measures for transitive triplets and transitivity percentages by themselves at each one of
the time periods should not be taken as the sole indicator to conclude that actors at a
particular point of time were less active or engaged. Collaboration ties are considered to
have directions meaning that when one actor reaches out to another to develop a possible
collaborative project, it does not necessarily mean that the other party will accept the
initiative. Due to the possible amount of long term dedication and resources that are
being involved, the decision to initiate a collaborative connection on either party is
equally “valuable” in understanding the formation of this type of social structure. The tie
that made the transitive triplet complete, regardless of directions, indicated the actor’s
actively choose to be committed in the field of practice. Therefore, it is more meaningful
to compare the measures across time periods in order to examine such changes in
motivation, especially before the after the disaster.

452
Emergency Response and Long Term Recovery
During the emergency response period, the number of transitive triplets increased
from 19 to 148 by 87.16% and this was followed by another increase to 300 during the
recovery stage, which amounted to 50.67% increase from the level in emergency
response stage.
The cases when a single link could complete the triads went from 84 to 765,
which was a 89.02% increase from before the earthquake event to the emergency
response period. The quantity of transitive triadic cases such as
 
,, AB BC anything
also
nearly doubled to 1538 during the recovery stage. Therefore, after the earthquake, at the
same time actors enacted their agency activity in initiating communication relationships
with others, their behaviors were also accompanied by high level of motivation in terms
of the drive to commitment. This is demonstrated by the fact that the number of
incidences of
  ,, AB BC AC
maintained a 50.67% increase from emergency response to
the long term recovery stage.
As more actors starting to build collaborative relationships with the remaining
others in the network, the cases that could easily become transitive went up by 89.02%.
Perceived as opportunities from the civil society actors’ perspective, they actively

453
recognized the possibilities within the boundaries of
 
,, AB BC anything
structure.
More actors were then being enabled through such a structural environment and were
willing to complete the institutionalization process of a triadic structure. The reason that I
used the term “enabled” here is because there was a kind of force that triggered further
action of completing the transitive closures. Such force was not generated by any one
actor but resulted from the collaborative synergy generated by the outreach activities
across the entire network inside the civil society domain. As a result, the increased
availability of intermediaries between actors also provided actors’ opportunities to look
for projects collaborators that formerly would not be possible without the existence of
medium actors. In other words, the sheer size of the “capability set” that the actors could
exercise made it possible for their functionings to expand through the availability of
 
,, AB BC anything
structures. Here, I am using the transitive triplet measures of
collaboration networks to illustrate these concepts such as capability set and functionings.
And this is because compared to the motivations for information exchange, the intensity
of the level of commitment for each collaboration tie represents a choice that is more
enduring and better reflects the opportunities in terms of long-term trajectories that actors
can act out their commitment and devotions.

454
It is apparent from the very last column of collaboration transitivity figures (see
table 5.2.2), the percentage of transitive triplets by all the relations that are easily to be
completed with one tie, there is a decrease of approximately 3% when comparing before
and shortly after the disaster. One interesting aspect of the development of transitivity ties
for collaboration networks is that it experienced an initial shock going down from
22.62% to 19.35%, then it slowly picked up its pace to 19.51% over the long term
recovery period. This type of trend could be due to the following reasons. One is the
nature of commitment and level of devotion for collaboration transitive ties. After the
earthquake, as more actors were being increasingly embedded into the network structure,
the cases that could easily become transitive such as structures of
 
,, AB BC anything
also increased dramatically as showed earlier. Thus, the opportunities for those such as
actor A to reach out to actor C to actively close the transitive triplet structure among A,
B, and C increased accordingly. However, the expansion in choices within their
“capability set” was one thing, but the execution to actually establish a collaborative
project by reaching out to the “friend of a friend” is a different matter altogether. In the
case of
  ,, AB BC AC
, for actor A to actually make a decision to initiate a connection
that is based on potential project collaboration, A has to take into account of factors such
as the match of interests between A and its target C, commitment to the field of practice

455
and expertise development, project location, as well as the intended duration of the
possible project. Compared to building communication ties, collaborative structure
formation required additional steps for projects to come into being and each step taken
along the way represented the higher level devotion on the part of the civil society actors.
Therefore, shortly after the disaster, although both the number of
  ,, AB BC AC
structures and
 
,, AB BC anything
structures increased significantly at the same time,
the latter increased by a higher percentage than the former measure. The result of this was
a temporary fall of transitivity for the emergency response period. Over the long run,
more collaboration ties were being established as actors took the time to get to know each
other. Therefore, in the discussion of transitivity structural changes over time, it would be
premature to simply conclude that actors were less committed to collaboration with each
other right after the earthquake. In the context of understanding institutional-building
within the civil society domain at times of catastrophic changes such as disasters or crisis,
it is useful to take into account of the specific types of structures under consideration and
weighing the different factors that might be affecting the institutionalization process.

456
Clustering
Moving on to the formation of clustering structures in the collaboration networks
over time, let us compare the clustering coefficient measures of the communication
networks to the project collaboration networks. As shown in table 5.2.3, the first
featuring difference is that the overall graph clustering coefficients are consistently lower
in quantitative measures for collaboration networks than that of the communication
network.
Table 5.2.3. Cross-Network Comparison of Clustering Coefficient
Communication
(Motivational)
Network

Collaboration
(Commitment)
Network
Overall
clustering
coefficient
Weighted
clustering
coefficient
Overall
Density
Overall graph
clustering
coefficient
Weighted
clustering
coefficient
Overall
Density
t1 0.341 0.129 0.0122 0.114 0.066 0.0040
t2 0.455 0.139 0.0544 0.236 0.084 0.0127
t3 0.496 0.167 0.0631 0.223 0.101 0.0164
For the more commitment-oriented collaboration networks, while the clustering shortly
after the earthquake increased from 0.114 to 0.236, the measure dropped to 0.223 during
the long term recovery period. However, after weighing the differences of neighborhood
sizes inside the network structure, the measures of clustering coefficients kept a tendency

457
for increased clustering. The level increased from 0.084 to 0.101 for the two periods after
the disaster. One unique feature of this change process is how quickly the incidences of
clustering actions in project collaboration caught up with the agency that built up the
communication clustering structures. Note that the commitment-oriented collaboration
clustering for both before and shortly after the earthquake turned out to be lower than
those of the motivation-oriented communication networks. However, during the longer-
term recovery period, the commitment level in terms of clustering through collaboration
ties (0.101) became almost comparable with the actors’ drive to exercise agency freedom
observed through the level of communication clustering (0.167).
This is an encouraging sign in formulating theories that provide possible
alternatives in understanding the process of emergence of civil society in crisis situations.
In the context of Wenchuan recovery, it had been a consistent character for the civil
society actors to be more willingly to execute their agency for relationship-building
through communication networks. This tendency can be demonstrated by a couple of
network measures that I have looked at. They include cohesiveness measures such as
density, reciprocity, and transitivity. The higher clustering levels for communication
networks across all time periods again illustrated the tendency for actors to act out their
agency freedom in the information exchange type of networks. For the emergency

458
response period, information exchange was more densely clustered as compared to before
the earthquake. This is understandable because the primary intention at that time was to
focus on providing immediate assistances to alleviate the catastrophic impact of the
disaster. The situation itself prompted a sense of urgency for actions to communicate
mainly due to the underlying motivation of the actors to help local communities with
whatever the resources were available at the time. This way, the motivation itself was
quickly revealed through the information exchange behaviors. For the collaboration
networks however, motivation itself alone would not guarantee a successful emergence of
a project collaboration tie between two actors. Collaboration relationships represent a
level of commitment that would take into account all aspects of development in which
actors were engaged. As stated earlier, these include the field of expertise, location of
practices, resources available, etc.
In the context of this study, the fact that the level of clustering for both
communication and collaboration networks experienced growth throughout the same time
frames is another indicator of the nature of civil society emergence in the Chinese
context. This informs us that civil society actors possess both motivation for original
action and commitment for sustained institutional-building over time. The fact that
clustering coefficient for the collaboration network during the recovery period almost

459
caught up with the level of communication network at the same time stage is one of the
most significant indicators showing the actors were being increasingly committed. This
indicated a tendency for the ties within the collaboration structure to be further
institutionalized. Since this research study was conducted to reveal the reality from civil
society actors’ point of view, the stories of structure formation and sustainability were in
turn designed to emphasize the active side of the nature of civil society. At this stage of
the analysis, I look next at the communication networks as a representation of
motivational social structure and the collaboration networks as a representation of
commitment social structure.

Pre-Earthquake
Let’s now visually examine the collaboration network in order to further illustrate
the changes of clustering coefficients over time. At the time stage before the earthquake,
the weighted clustering coefficient for the collaboration network is 0.066, which is higher
than the overall density of the whole network (0.004). This demonstrates that there was
already certain degree of clustering at this period. Looking at the collaboration network
graph for the stage before the earthquake (see figure 5.2.1), the first visually stunning

460
feature of this network is that it is separated by two distinct clusters, aside from the
“isolates” listed on the very left hand side of the picture. These “isolates” are the actors
that did not engage in any collaboration projects with others. Both of their in-degree and
out-degree were zero.

Figure 5.2.1. Pre-earthquake Collaboration Network (Combined Actor Attributes of
Date of Establishment, Geographic Location, and Registration Status)
Those who were active in project collaborations can be further divided into two
types of connection structures. One is the larger connected network with the majority of
the actors inside of it. The other one was structured as an “isolated tree” composed of
only four actors. It is called the “isolated tree” because none of them were connected to
1
10
100
101
102
103
104
105
106
107
108
109
11
110 111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
*Color: establishment
Yellow: before EQ; Blue: after EQ
*Shape: location
Circle: Sichuan based; Square: Non-Sichuan
*Size: Registration status
Small: registered; Large: non-registered

461
the majority of the actors and nor were they collaborating extensively with each other.
Actor #107 actively reached out to #8, #27, and #56, thus it had a local neighborhood size
of three. But none of these neighbors were connected to each other. So when calculating
the density for this particular neighborhood by leaving out the focal actor itself (#107),
the measure would give the clustering coefficient for actor #107 of zero because all of the
three others would become isolated without the outreach effort of #107 in maintaining
their collaborative relationships.
The cluster containing the majority of the actors was relatively more loosely
connected compared to the structure of communication network at the same period
(shown in figure 5.2.2). The collaboration network also had fewer reciprocated
connections (ties colored in red) as compared its communication counterpart.

462
Communication Collaboration

Figure 5.2.2. Comparison of Pre-earthquake Communication and Collaboration
Networks (Combined Actor Attributes of Date of Establishment, Geographic
Location, and Registration Status)
The overall pattern of the output results for node level cluster coefficients
103

demonstrated this point. The average size of the local neighborhoods for those included
in the network was smaller than its communication counterpart. Out of all the possible
ties in the immediate neighborhood of each actor, fewer of those actually turned out to be
present. In other words, the percentage that pairs of neighbors were connected out of all

103
Please see Appendix 5.2.3A1 for sample comparison from the UCINET output.
1
10
100
101
102
103
104
105
106
107
108
109
11
110 111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
*Color: establishment
Yellow: before EQ; Blue: after EQ
*Shape: location
Circle: Sichuan based; Square: Non-Sichuan
*Size: Registration status
Small: registered; Large: non-registered

463
the possibilities also turned out to be relatively low. Most importantly, the disjointed
nature of the collaboration network before the earthquake indicated that actors in the civil
society were less aware of each other’s existence and areas of specialization. They were
also relatively more self-focused and dependent on the state sector to make their
commitment in project formation. For example, the state actor (#1) had one of the highest
numbers of local neighbors at this period of time. Out of its 45 pairs of neighbors, 6.7%
turned out to be present. The next highest level of neighborhood size is 36, appeared for
actors #137 and #61. For actor #137, none of its neighbors actually had collaborations
with each other. For actor #61, only 5.6% of all these possible collaboration ties actually
were present at the time. In summary, the general clustering feature for this network was
that civil society actors tended to develop collaboration projects with only limited number
of others rather than having expansive and diversified collaborative partners. Also, the
local neighbors for one particular actor did not tend to have collaborative ties with each
other. The general synergy of behaviors for civil society actors in this network was one
characterized by limitation, disconnectedness, and commitment actions that were
relatively inward-looking.

464
Emergency Response
The situation changed immediately after the earthquake event. The dynamics of
change was most dramatic and obvious during the emergency response period. A visual
inspection of figure 5.2.3 below showed that the collaboration network became
increasingly more integrated with more actors being embedded into the different clusters
of neighborhoods.
Communication Collaboration

Figure 5.2.3. Comparison of Emergency Response Communication and
Collaboration Networks (Combined Actor Attributes of Date of Establishment,
Geographic Location, and Registration Status)
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128 129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
*Color: establishment
Yellow: before EQ; Blue: after EQ
*Shape: location
Circle: Sichuan based; Square: Non-Sichuan
*Size: Registration status
Small: registered; Large: non-registered

465
Indeed, the weighted overall graph coefficient for project collaborations went up from
0.066 to 0.084, which is about eight times higher than the overall density of the graph
(0.0127). With increased level of clustering
104
, there were still actors not included in the
collaboration network and acted out by either being “isolates” listed on the left-hand side
of collaboration graph or being an “isolated tree” composed of actor #42 and #103. The
collaboration between these two actors was completely disconnected with the main
structure of the network. For those who were already inside the main structure, the actors
were being more enthusiastic and open about reaching out to others to develop
collaboration projects. This is reflected in both an increase in focal actor’s local
neighborhood size and higher level of clustering coefficients when more collaboration
connections were actually being realized out of all the possibilities around the different
neighborhoods. Overall, despite of the continuing existence of some non-participating
actors, the clustering measures of emergency response period revealed that the civil
society actors became more open-minded in diversifying their collaboration partners and
were willing to take actions towards making commitments. From figure 5.2.3, we can
note that this attraction force to build collaboration ties was accompanied by the complete
integration of the communication network at this period of time. We can also recognize

104
Please see Appendix 5.2.3A2 for sample comparison of the two types of networks during emergency response.

466
that during this time, more newly emerged grassroots actors, both formal and informal,
became embedded in the highly clustered section of the graph. These include actors such
as #3, #123, and #49.

Recovery
The most distinguishing development of the formation of collaboration networks
for the long term recovery period was the disappearance of “isolated trees” in the graph
(shown in figure 5.2.4). Actors #107, #27, #8, and #56 that used to be acting in their own
“enclave” before the earthquake, along with the collaboration between actor #42 and
#103 that were being separated from the main connection structure during the emergency
response period, all joined together into the collaboration network. The density of the
whole graph is 0.0164. And the clustering density of local neighborhoods (0.101) turned
out to be much higher than the overall density, thus indicating high clustering activities
being maintained through the long-term recovery stage. More actors were being
embedded in relatively highly clustered neighborhoods
105
.

105
Please see Appendix 5-2-3 (a3) for sample comparison of the two types of networks during recovery stage.

467
Communication Collaboration

Figure 5.2.4. Comparison of Recovery Communication and Collaboration Networks
(Combined Actor Attributes of Date of Establishment, Geographic Location, and
Registration Status)
At the same time that agency actions in making primary connections through information
exchange remained to be further integrated and cohesive (see communication network
graph in Figure 5.2.4), the synergy of openness and activity in developing collaboration
projects endured and enhanced as compared to the previous period. Civil society actors
also perceived that the state and the business sectors as critical partners in developing
collaborative efforts for long term earthquake recovery. The state actor had an expanded
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19 2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84 85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129 13
130
131
132
133
134
135
136 137
138
14
15
16
17
18
19
2
20
21
22
23
24
25 26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
*Color: establishment
Yellow: before EQ; Blue: after EQ
*Shape: location
Circle: Sichuan based; Square: Non-Sichuan
*Size: Registration status
Small: registered; Large: non-registered

468
neighborhood size of 325 possible ties with 6% of them being realized into actual
collaborations. Recall that during the period before the earthquake, it only had 45
possible ties with the potential to develop into collaborations. When the percentage of
actual realization of those ties for both periods stayed around the same level, it means that
more of the civil society actors actually engaged the state to cultivate collaboration ties
over time. For the private sector, it had 91 possible collaboration ties in its immediate
neighborhood at the recovery time. Therefore, the business sector was also perceived as
an important player in the collaboration network and it was increasingly being embedded
in highly clustered neighborhoods. At the same time, 12.1% of these ties turned out to be
present. As compared to the measures before the earthquake when only 15 pairs of actors
were in its local neighborhood, there was a rather drastic change in the collaboration
structure providing increasingly more opportunities for private sector in the market
system to engage civil society actors in project collaboration efforts. Evidence also
indicated that this built-up potential was not fully utilized as out of all the possible ties
among its neighbors, only 3.3% of the potential collaboration relationships were actually
present.
In retrospection, the examination of the above measures of clustering
demonstrated a self-evolving process of structure clustering among actors with different

469
sizes of local neighborhoods. Originally “awakened” by the catastrophic impact of the
earthquake, the agency spirit of the civil society actors was “stirred up” through the
coming-together efforts for emergency response across sector boundaries. The motivation
and willingness to connect and build-relationships with other groups and organizations
were revealed in the fact that more actors were being increasingly embedded in highly
clustered local communication neighborhoods. At the same time, the clustering actions
also showed high level of determination for civil society actors to be committed in the
field practice for the long term. Such results are in general in accordance with what I
have found from the formation of dyadic and triadic communication and collaboration
relationships.

Registration Status Group-external and Group-internal Ties
In the last chapter, recall that I focused on the analysis of group closure based on
registration trait for the communication network structures. Note that in this study, the
communication networks represent a type of social structure that reveals the underlying
motivations of actors. In order to look deeper into the commitment side of the story, we
need to investigate further to see if there were variations of group bounded-ness resulting

470
from actor traits in the project collaboration networks. It is also important to examine the
structure of cross-sector collaborations among the state, the market, and the civil society
in the context of sub-groups, which will be dealt with separately in one of the later
chapters. The possible in-group and between-group collaborations categorized by
registration status will also have important policy implications for civil society capacity-
building from a disaster response and recovery perspective.
One point worth mentioning here is regarding the treatment of the nature of
collaboration ties for calculating the measures in this section. When considering group
internal and external ties in the current network analysis, the directions of ties, whether
incoming or outreach, were not taken into consideration. If one actor nominated the other
actor to be its project collaborator, regardless of whether such a tie was being
reciprocated by the other, the connection was still counted as the existence of one tie
between the two actors. On the one hand, it is certainly possible for collaboration ties to
have directions in this research context and one factor that could have contributed to this
phenomenon is due to a set of possible “mismatched” understanding of collaboration
projects among the participating actors. While it is not possible for current network
analysis program to capture the nuances of distinguishing all the possible definitions from
the actors’ perspective, it does allow the study to treat ties that could partially take this

471
factor into consideration. And this is the case for all the other analyzing sections that are
being discussed. On the other hand, the reason that the ignorance of tie directions is
allowed in this particular section is because of the nature of the structure in consideration
here. The emphasis of the discussion on group internal and external ties is based on actor
attributes such as registration status in this case. Thus, whenever a tie is being named by
one actor, it is the grouping of these ties based on a particular category of attribute that
are the key consideration in this part of the analysis. As a result, the following analysis is
conducted when taking these arguments into account so as to move forward with the
available interpretations.

Density Matrix
Let’s start out by examining the changes of the density measures for the three
consecutive periods. I will first compare and contrast the patterns of change in the overall
density within and across registration status groups for collaborative networks both
before and after the earthquake. As shown in table 5.2.4, for the time period before the
disaster, the densities for cross-group collaboration (0.009) are higher than within group
collaboration (0.007).

472

Table 5.2.4. Communication and Collaboration Network Pre-Earthquake Within-
Group and Cross-Group Density Measures (Registration Actor Attribute)

Communication Collaboration

Group 1
(registered)
Group 2
(non-registered)
Group 1
(registered)
Group 2
(non-
registered)
Group 1
(registered)
0.017 0.031 0.007 0.009
Group 2
(non-registered)
0.031 0.017 0.009 0.007
This means that before the earthquake, project collaborations across registered and non-
registered group actors appeared to be more prevalent than within group actions.
However, the in-group and cross-group densities were significantly lower for
collaboration ties than those in the communication network at the same period of time.
Cross-group communication was at a density level of 0.031 while the collaboration
density level was at 0.009. Immediately after the earthquake, as shown in table 5.2.5, the
density measures for the registered in-group collaboration ties went up from 0.007 to
0.024. The measure for the non-registered in-group collaboration ties went up from 0.007
to 0.017. However, compared to the communication network, the within group and cross-
group initiatives in building collaboration ties was still not as high in intensity.

473

Table 5.2.5. Communication and Collaboration Network Emergency Response
Within-Group and Cross-Group Density Measures (Registration Actor Attribute)
Communication Collaboration
Group 1
(registered)
Group 2
(non-registered)
Group 1
(registered)
Group 2
(non-
registered)
Group 1
(registered)
0.098 0.107 0.024 0.023
Group 2
(non-registered)
0.107 0.074 0.023 0.017
Also, project collaborations among registered actors became more prevalent than those
made among non-registered actors at this stage. At the same time, the cross-group
collaborations increased from 0.009 to 0.023. On the one hand, this suggested that not
only more actors became participants of the collaboration network but also more
collaborative projects were being established among them. On the other hand, the density
of cross-group project collaboration was as high as the within-group collaborations
among the registered actors (0.024).
Over the long term recovery, from the results shown in table 5.2.6, one of the
most noticeable changes in the group closure structure was that the density for cross-
group collaborations (0.034) became higher than its within-group counterparts (0.028 for
registered group and 0.025 for non-registered group). When comparing the measures

474
before and after the earthquake, actors clearly became more engaged in collaboration not
only within their own registration status groups but also reached further across the
institutional status divide.
Table 5.2.6. Communication and Collaboration Network Recovery Within-Group
and Cross-Group Density Measures (Registration Actor Attribute)
Communication Collaboration
Group 1
(registered)
Group 2
(non-registered)
Group 1
(registered)
Group 2
(non-
registered)
Group 1
(registered)
0.108 0.128 0.028 0.034
Group 2
(non-registered)
0.128 0.089 0.034 0.025
At the same time that cross-group collaborations were gaining prevalence over time, the
process was accompanied by the consistent increase in the total number of collaboration
ties. Tables 5.2.7, 5.2.8, and 5.2.9 showed that collaboration efforts within registration
status groups increased from 86 to 300 from before the earthquake to emergency
response. It maintained its upward trend to 350 during the recovery stage. Collaborations
made across registration status groups experienced a change from 58 ties to 144
immediately after the disaster and such efforts kept going all the way into the longer term
with 216 project collaboration ties made during this time stage.

475
Table 5.2.7. Collaboration Network Pre-earthquake Whole Network Results of
Group Internal and Group External Ties Based on Registration Actor Attribute
Frequency Percentage Possible Density
Internal 86.000 0.597 12584.000 0.007
External 58.000 0.403 6322.000 0.009
Note: Total ties=144

Table 5.2.8. Collaboration Network Emergency Response Whole Network Results of
Group Internal and Group External Ties Based on Registration Actor Attribute
Frequency Percentage Possible Density
Internal 300.000 0.676 12584.000 0.024
External 144.000 0.324 6322.000 0.023
Note: Total ties=444

Table 5.2.9. Collaboration Network Recovery Whole Network Results of Group
Internal and Group External Ties Based on Registration Actor Attribute
Frequency Percentage Possible Density
Internal 350.000 0.618 12584.000 0.028
External 216.000 0.382 6322.000 0.034
Note: Total ties=566
This is an encouraging piece of evidence signaling a gradual institutionalization
process represented by the commitment-oriented collaboration structure between
registered and non-registered actors. As more emerging grassroots voluntary groups
started out with a non-registered status during the immediate short term response period,

476
the fact that over time, these actors became increasingly embedded in collaborating
relationships with registered actors reveals the following. First, this kind of behavior
pattern showed that actors in the network were being increasingly committed to cross-
group collaborations. Second, the informal grassroots groups that were smaller in size
and less in experience would have higher chances to grow and learn from being involved
in the same projects with those registered actors, who often had more expertise and
experience in the field. Thirdly, the dominating prevalence towards cross-group
collaboration over the longer-term signified a primary stage of civil society emergence
through the institutionalization of a collaborative structure that was based on actor
commitment, long term devotion, and most importantly climbing willingness and agency
efforts to break the boundaries of separation.

Rescaled E-I Index
After taking into consideration of the different group sizes and the density
measures, I examine the re-scaled E-I index for collaboration networks over time (see
table 5.2.10).

477
Table 5.2.10. Collaboration Network Re-scaled E-I Index
t1 t2 t3
Re-scaled E-I
Index
-0.194 -0.351 -0.237
The E-I Index measure for the period before the earthquake is -0.194, which can be
interpreted as there was a slight tendency towards group closure at this time stage. After
the event of the disaster, the index was temporarily up to -0.351 and this suggests that
there was a further tendency for collaboration ties to move towards group closure at the
period of short term response. However, the trend made a reverse turn during the long
term recovery stage by decreasing to a level of -0.237. From these results, what we can
conclude regarding the change pattern of collaborative behavioral tendencies throughout
the three time periods are three folds. Firstly, immediately after the earthquake, there was
a surge of tendency towards in-group closure, given the group size constraint and the
overall density at the particularly time stage. Secondly, the tendency towards group
closure waned over time into the long term recovery stage. Thirdly, as civil society actors
became increasingly committed to the field by establishing project collaboration
initiatives among each other, they also tended to take less consideration of the registration
status of their project partners. What this suggests is that the focus of efforts in
establishing a collaboration tie was status-based according to registration formality or
informality. Taking such a process into an understanding of the emergence of civil

478
society, the changes in group embeddeness also indicated that at the last stage of
observation, civil society actors, driven by commitment, were actually able to self-
organize and orient towards a collaborative structure that tends to be less segregated
based on registration attribute. Such characteristics of institutional change inside the
domain of civil society can be informative when it comes to understanding the “fabric”
and “texture” of institutional change of disaster response and recovery networks.

Group Level E-I Index
Having examined some of the general group closure trends from the whole
network perspective, we now turn to the results from group level analysis. During the
period before the earthquake, we can see that more internal ties were being made among
registered civil society actors than internal ties made among non-registered actors. At the
same time, each group made 29 external ties. With only 6 internal collaborative
connections, the non-registered group seemed to be more active in having collaborations
with registered actors. Such tendency can be more precisely demonstrated by the group
level E-I indexes shown in table 5.2.11.

479
Table 5.2.11. Collaboration Network Pre-Earthquake Group level E-I Index Based
on Registration Actor Attribute
Internal External Total E-I
Group1
(registered)
80.000 29.000 109.000 -0.468
Group2
(non-registered)
6.000 29.000 35.000 0.657
The group index measure for registered actors is -0.468 and 0.657 for non-registered
actors. Since -1 represents that all ties are internal to the group and 1 represents all ties
are external, we can conclude that non-registered actors had a higher tendency for making
cross-group collaboration connections. While the degree for registered actors to make
group internal connections was relatively high, but in absolute value terms, the non-
registered actors showed greater tendency towards building collaborations with those
possessing different traits in terms of registration status. What can be inferred here is that
the collaboration network during the period before the earthquake already had a set of
“endowed” tendency for informal grassroots groups to have cross-group project
collaborations with formal organizations in the field. Although the number of external
ties quantitatively was small in size, when taking into consideration of the internal ties for
both groups, the unique character of the informal grassroots groups at this stage of
structure development became apparent.

480
Immediately after the disaster, the project collaboration structure experienced a
dramatic change in terms of the number of internal and external ties for both groups (see
table 5.2.12).
Table 5.2.12. Collaboration Network Emergency Response Period Group level E-I
Index Based on Registration Actor Attribute
Internal External Total E-I
Group1
(registered)
286.000 72.000 358.000 -0.598
Group2
(non-registered)
14.000 72.000 86.000 0.674
First of all, the total number of collaboration ties inside the registered and the non-
registered groups jumped up and this was accompanied by almost three-fold increase
from 29 ties to 72 ties in external ties across the two groups. Also, the registered actors
showed a greater tendency towards group closure with its index measure increased from -
0.468 to -0.598. The non-registered actors, on the other hand, were more inclined to build
collaborations outside of their own group. In general, the nature of group embeddeness
for both the registered and non-registered actors during the immediate disaster response
period remained to have similar trends in their collaboration behavior as before the
earthquake. There was a continuation of collaboration tendencies for both groups but
with opposite directions in terms of group closure. Non-registered actors continued to be

481
more open to collaborate with those registered actors in the network while the later tend
to have more relationships within group.
Over time, when it was into the long term disaster recovery period, one of the
most distinguishing changes occurred was a decrease in group closure tendency for
registered actors when comparing the E-I Index columns of table 5.2.12 and table 5.2.13.
Table 5.2.13. Collaboration Network Recovery Period Group level E-I Index Based
on Registration Actor Attribute
Internal External Total E-I
1(registered) 330.000 108.000 438.000 -0.507
2(non-registered) 20.000 108.000 128.000 0.688
And this happened at the same time when the non-registered actors kept an even higher
tendency for having outside group ties. This piece of evidence corroborates with the
conclusion drawn earlier when interpreting the decreasing re-scaled E-I index for the
overall collaboration network. The results at the group level revealed that the change
mechanisms inside the registered group might help explain the declining tendency for
group closure from the overall network perspective. More specifically, as the network
became denser with actors building more collaboration projects among each other over
time, the tendency for within-group project collaboration among the registered actors
seemed to be dissolving.

482
Another point to notice is regarding the timing when this change occurred. On the
one hand, like the communication network, collaboration network reached its peak in
terms of the connection activities during the emergency response stage. On the other
hand, the declining tendency that dissolves the registration divide in project collaboration
only came during the long term recovery period. One of the factors that might have
contributed to the lag in the changing group closure dynamic can be the development of
growing familiarity among actors, especially towards the newly emerged grassroots
voluntary groups right after the earthquake. Over time, trust can be built through
increasing opportunities for communication and information exchange. At the level of
making commitments, what can be inferred from this type of structural change is that
actors might be aware of the registration status among each other when designing
collaborative projects together, but such institutional status might not enter as a
determining factor in the actor’s long term behavioral responses to the disaster.

Civil Society Actor Variability
Lastly, I will briefly compare the E-I index measures for some of the civil society
actors across the three time periods. The main purpose for depicting the various levels of

483
group closure at this point is primarily in demonstrating how the individual level E-I
index can be utilized to distinguish actors with various tendencies for in-group and
between-group collaboration ties. When adapted into the context of disaster management,
these measures can be further formulated as policy indicators for purposes to enhance the
social capacity to deal with change and risk in general. The following paragraphs will
illustrate what such indicators could entail when looking at group embeddedness
measures.
Tables (5.2.14, 5.2.15, and 5.2.16) shown below provided summaries of those
civil society actors with higher tendencies for registration status related collaboration
relationships.
Table 5.2.14. Collaboration Network Ranking of Variability across Individual
Actors (Pre-earthquake)
Civil Society Actor E-I index
#100 0.8
#93 0.6
#61 0.556
#51 0.333
#137 -0.778
#70 -0.750
#119 -0.667

484
Table 5.2.15. Collaboration Network Ranking of Variability across Individual
Actors (Emergency Response)
Civil Society Actor

E-I index
#93 0.750
#135 0.714
#100, #61 0.500
#51 0.250
#107, #118 -0.778
#49 -0.750
#3 -0.742
#137 -0.714
#119 -0.684

Table 5.2.16. Collaboration Network Ranking of Variability across Individual
Actors (Recovery)
Civil Society Actor

E-I index
#135 0.846
#93 0.647
#61 0.500
#51 0.385
#19, #64 -0.750
#49 -0.714
#88 -0.667
#123 -0.500
For the period before the earthquake, actors #100, #93, #61, and #51 tended to build
project collaborations with those of different registration status with themselves. Among
these four actors, #51 was the only one that had newly established field office in Sichuan
Province after the earthquake. Right after the earthquake, the actor had slightly higher
tendencies for within-group collaborations in terms of registration status when compared
to before the earthquake. But over the long term, its actions revealed a type of recognition

485
in building up greater cross-group project collaborations with more non-registered social
groups. In order to understand the motivations behind the actor’s cross-group
collaboration initiatives, the following account traced the nature of its collaborative
relationships with both registered and non-registered civil society actors.

The Case of Actor #51(NGO51)
The co-evolutionary process of collaborative initiatives among civil society actors
and their behavioral tendencies for institutionalization towards being a “social
organization” functioning professionally can be further illustrated from an example
showing its interactions within the civil society sector. The following example depicts the
role of actor #51 in such a co-evolutionary dynamic between collaborative relationships
and growth performance inside the civil society domain.

Collaboration development with actor #8
Before the Wenchuan earthquake, as was noted by the program officer, most civil
society actors would simply choose to engage in related activities in the form of informal

486
self-organizing volunteer teams. Not many formal professional social organizations were
in existence around the city of Chengdu back at the time. After the earthquake, however,
one of the many institutional transformations developed from a “volunteer team” to a
formal “social organization” can be exemplified through the interactive experiences of
actor #8 and actor NGO51.
In fact, before the 5.12 earthquake, it was only a volunteer group or team that
relies on individual actions, but not a formal social organization in a strict sense.
This is why all of their activities were conducted in the volunteer format. But after
the 5.12 earthquake, when we started to get to know each other and were chatting
together…because we were all participating in the same field…they would find
out that we were doing this aspect of works. They would then want us to help
their group to become a formal organizational type, a professional one. Since its
formation after the earthquake, we would observe to see if it will possess the
potential to develop further. For example, whether it will have plans for
development in the future and also shows clarity in the requirements of our
programs are important factors. If the only thing that it lacks is the familiarity in
developing itself on a right path, or in other words, through what steps and
processes can they transition towards a professional organization. Also, if they
were not sure how to handle the particular program after developed into a formal
organization, under these circumstances, we will then willingly provide them with
all kinds of support. (NGO51-01-08
106
)
Two key factors came into play in the institutionalization process of actor #8. First, the
awareness of the existence and the works of actor NGO51 after the earthquake event
were critical in actor #8 becoming informed and eventually motivated. From a policy

106
For original Chinese script please refer to Appendix 5.2.3.01.

487
making perspective, this signifies the importance of building various communication and
information sharing platforms for these civil society actors to have the opportunities to
interact with each other on a consistent basis. In this study, I have already shown that the
earthquake did raise the awareness of some emerging civil society actors to take this role
of being the facilitators for different kinds of platforms. What the government can do in
this respect, is to provide an institutional environment that nurtures such interactive
dynamics so that communicative and collaborative relationships can be developed and
enhanced. Building the kind of “resilience” that bridge the institutional status divide for
the society to deal with future disasters or risks in general does not just depend on the
actions of civil society alone. It also requires the conscious efforts on the state side to
contribute to the institutional conditions within which actions inside the civil society
domain can flourish over time. The second factor being illustrated here is related to the
civil society actor’s own determination and awareness while being able to specify its long
term goals and plans. When such an inward recognition of its role inside the civil society
domain is brought about through interaction with its “peers”, the actual collaborative
efforts can be formed based on the areas of specialization among actors. In the current
case, it was actor #8 that first contacted NGO51 and applied to be on its “incubation”
program. Then, based on the expertise of the latter, the support for actor #8 came in the

488
form of introducing it to the “community service platform” ( 社区服务平台). And this
provided the opportunities for #8 to be a collaborative partner and participate in one of
NGO51’s sixteen local service centers in order to develop the needed professional skills
to become a formal social organization ( 社会组织), and at the same time assisting the
local community development process during the recovery period after the earthquake.
The program officer of actor #51 referred such “incubation” mechanism as a “support
platform”. What such platform was established on was not oriented towards general
communication, training, or field-specific coordination like those provided by actor #3
and NGOLF. Rather, it was practice and commitment-oriented and facilitated the hands-
on experiences in developing the skills of the emerging civil society actors
107
.
Two types of collaborations existed within the support platform. One was the long
term support that would last at least for a year. This means that what the “incubated”
actor proposed to do should be intended for long term involvement in the field and at
least one of its staff must be committed to the field activities full time. In the case when
the actor did not have enough full time personnel to allocate for field practices, the kind
of support being available was called “the periodical collaboration”. And this can be
illustrated through the following:

107
For further detailed account, refer to Appendix 5.2.CaseNGO51.4 (1).

489
This means that targeting some of the needs of local communities. Sometime
there might not be a civil society organization providing this kind of service, and
also they could not devote a great amount of human resources since they
themselves are in the initial start-up stage. A shortage of staff is common for them
and sometimes they won’t be able to provide even one full-time employee to
work in the field. For those in Chengdu, some won’t be able to send employees to
Beichuan to work full-time. So, what we can do is to introduce our collaboration
in different phases. The program designs may be for the duration of 2-3 months
short term. We can provide an opportunity for them to design the program,
specifically targeting the 2-3 months period to assist the local communities on
their needs. This way, we will be complementing the local civil society
organizations to execute the program together. Therefore, these organizations
would be called stage-wise groups/organizations. For those inside our incubator
programs, we hope to provide them with more of these opportunities for them to
go into the communities practicing their work there. That would be great practice
on both the operation side and the program execution side of their experiences.
And as a result, they would grow in a quicker pace. This is why we and these civil
society groups/organization did not have previously communications or
collaborations. (NGO51-01-09
108
)
Two key characteristics stood out in this platform design. One is that by bringing in other
local social organizations to work alongside with the “incubated” grassroots actor, the
platform served as an intermediary for building further communication and collaboration
connections among those that work in the same field project. Also note that such
communication and collaboration ties became more of a mutual learning experience for
all civil society actors being involved, particularly for the actor being supported through
the incubation program. This was also the key explanation in that the two actors (#8 and

108
For original Chinese script please refer to Appendix 5.2.3.02.

490
#51) did not engage in any kind of communication and collaboration ties any time before
their partnership establishment through the incubation program. The nature of an
establishment of a collaborative relationship came into being only since the disaster
recovery period was one that based on mutual engagement and commitment aimed for
long term capacity building for civil society actors
109
.

Collaboration Development of Actor #51 with Actor #57
Another type of collaboration tie that came into being during the recovery period
was between NGO51 and actor #57. The latter is a formal NGO performing in the
different areas of social work and was originally established in Guangdong province. As
was recalled by the program officer, actor #57 started its projects in the disaster areas
before actor #51 did. This thus provided opportunities for the latter to observe what had
worked and what had not in the different areas being selected for field practice. Upon
seeking to exchange some experiences, actor #51 initiated the collaboration tie with actor
#57 in the first place.
For example, sometimes we will introduce some smaller local social
organizations, mainly because our energy would not allow us to fully take care of

109
Further detail refers to Appendix 5.2.CaseNGO51.4 (2).

491
all the aspect of the work. In these situations, we would probably invite local
social organizations such as (actor #57) to help us provide guidance and training.
Anyhow, our ultimate goal is to build up the community service organizations at
the entire earthquake impacted area. Then, we will have to coordinate resources
from different sides. Aside from the government, they may come from our
colleagues who already were conducting works in the area. For example, they
might be working there before us, for one year or so, then, they must have
accumulated some valuable experiences that can bring over to share. We will
invite them to share these experiences with the local social organizations that we
introduced in the program, or provide training for them regarding how to carry out
works based on local conditions….so this means that they are part of our
capacity-building framework design. (NGO51-01-10
110
)
Note that the primary motive for the actor to establish this kind of collaboration remained
to be developing local civil society actors that were trained and specialize in community
services. If the collaboration with actor #8 is to be interpreted as representing a direct
effort, on the part of NGO51, to develop the capacity of those emerging professional
“social organizations” in a substantive way, then, the type of collaboration with actor #57
represented an indirect effort in providing an enabling and nurturing environment to
facilitate the functioning of the tasks aimed by civil society actors. The two types of
collaboration designs were performed through an interlocking nature not only to provide
substantive content-wise resources for civil society actors to learn through practice, but

110
For original Chinese script please refer to Appendix 5.2.3.03.

492
also offered valuable relational resources in the form of cross-actor collaboration ties as
part of the capacity-building framework
111
.

Collaboration Development of Actor #51 with Actor #61
Aside from providing social services for “local community recovery” achieved
through the previous kinds of collaboration development, the third type of collaborative
relationship was motivated by the actor’s commitment to local economic recovery after
the disaster. The actions being taken in this regard was in the form of seeking expertise
from other civil society actors followed by establishing collaborative connections for
implementation.
One example of this kind of collaboration can be found in its connection with
actor #61. The specific project in action was called the construction of “Rural economic
cooperatives” ( 农村经济合作组织). The main purpose was to contribute to the local
economic development process as part of the long term recovery. Since the field of
expertise of actor #61 was in assisting the local communities developing their skills in

111
Further detail refers to Appendix 6.2.CaseNGO51.4(3) and also interview accounts of NGO51-01-11 in Appendix
5.2.3.04.

493
growing crops and breeding animal stock, it was chosen to be a collaborative partner for
its expertise in participating in the local economic development process.
Our collaboration experience is like this, they (actor #61) would act like our
consulting team. We carry out the implementation side of the work, including our
program execution and training provision. And they will provide the training
personnel. This is because they’ve been conducting works in the rural areas for
such a long time and gathered a lot of experiences. So they would be part of our
expert group. Every phase of our field visits and training, we will be consulting
with them. They can request demands and provide suggestions. From our
collaboration, we will then design our entire training program and our route for
field visit. So we generally take the lead while they are one of our collaborative
partners. In other words, they are like our consultation team. (NGO51-01-12
112
)
Therefore, the essence of the collaboration was realized through the two actors each
performing its own role. On the one hand, one party played the “supportive role” as the
expertise-provision team, or a knowledge-tank, because of its accumulated field
experiences working with local communities. On the other hand, the second party was the
one that actually implemented the final designs of the project, upon receiving field-
specific trainings from the designated group of experts. As we can see, this kind of
collaboration is characterized by actor NGO51 motivated in developing its own capacity
in a particular area of field practice
113
.

112
For original Chinese script, please see Appendix 5.2.3.05.
113
Further detail refers to Appendix 5.2.CaseNGO51.4(4).

494
What can be learned from the commitment-oriented behaviors of actor #51 as
well as those that it developed collaboration ties with is that the actor became less
restrictive in terms of registration status when it comes to project collaboration. Note that
aside from #51, actors such as #100, #93, and #61 were all registered formal nonprofit
organizations. A tendency for them to have a propensity to collaborate with those outside
their own group meant that projects were more likely to be formed with non-registered
social groups in the field. From the perspective of a policy intervention that aims
promoting the capacity development of the smaller non-registered domestic nonprofit
groups, actors such as these with the increasing propensity towards building commitment
ties across registration groups over the long term are critical to investigate.
In contrast, referring back to table 5.2.14, actors #137, #70, and #119 tended to
form collaborative relationships only with those inside their own group of registration
status. With the exception of #119, the other two actors were Chinese domestic registered
nonprofits. Actor #119 operated as a foreign-based nonprofit and before the earthquake, it
also had a higher tendency to form collaborations with other registered civil society
actors.

495
Shortly after the earthquake, from table 5.2.15, we can see that actors #93, #135,
#61, and #100 remained their position with high level of tendency to build cross-group
collaborations. Another feature for group embeddedness structure at this period of time is
that some of the more active newly emerged grassroots groups started out by having
greater tendency for collaborating with those of similar registration trait. For example,
non-registered actor #3 developed a tendency for collaborating with other non-registered
social groups. On the other hand, registered actor #49 emerged to have high level of
group closure for collaborative relationships. Interpreting from the perspective of civil
society emergence in the context of crisis and change, this kind of behavior signified one
of the primary characteristic for newly formed civil society actors to institutionalize their
relationship with others in a given social structure. In this study, the E-I index measure at
the actor level showed that shortly after the disaster, there was an enthusiastic synergy
that drove the tendency for newly formed civil society actors to seek out each other for
collaboration. Let’s take a closer look at the story of #49 in this context.

The Case of Actor #49 (NGO49)
Just like actor #3 and #51, once entered into the long term recovery stage, several
functions performed by actor #49 gradually developed as enablers for the institutional
development of interactions among actors inside the civil society domain. One of the

496
most important strategies was to support “professional service-oriented organizations”
through a process of what the organizer of #49 (NGO49-01) called “small project bidding”
every year. The reasons why this action could be looked at as an enabler facilitating the
institutionalization of interactions among civil society actors entailed an argument of
three-folds. First of all, it was due to the needs of the local communities. The invitation
for civil society actors to participate in projects that oriented towards providing services
would fulfill the needs of the communities. Secondly, such action would also facilitate
the desires of some civil society actors’ to practice at the local level but lacked sufficient
financial means to implement their causes. In the words of the organizer, this was to
“connect demand with supply”. Thirdly, the enabler originated from the goal of the
organization in general to build a “resource platform” by gathering available outside
resources and then “distributing to the disaster areas for community recovery”, mainly
through the practices of participating civil society groups/organizations. By way of these
three aspects of institutional practices of the organization itself, its interactions with all
other civil society actors can be categorized as a “provider” of resources for the
independent growth of individual actors in the civil society domain. Although similar in
its intention in building a platform for civil society actors to interact with one another, a
key difference between this actor #49 and actor #3 from a relational perspective was that

497
the latter performed as an “intermediary” role in facilitating the connection among other
actor while the former played a role as a “functioning resource provider” for others.
There was an inherent one-directionality for the “provider” type of relationship because
others will tend to reach out more towards the actor.
One other critical factor that performed as the “enabler” of the institutional
development of within sector interactions arose from how civil society actors perceived
to be their own strength of practice as compared to those implemented by the state sector
after the disaster. For actor #49, its organizer pointed out what he saw as the strength of
NGO practices:
In fact, what NGOs are good at is not in the aspect of infrastructure. What we are
doing is related to the so-called “soft services”, such as community service area of
practices. That is why we chose to focus our attention on community services
during the recovery period. It is about engaging in social services when finding
social workers who have a professional background, rather than engaging in
infrastructure recovery. The latter is what the government has been doing.
(NGO49-01-06
114
)
Here, he clearly identified the importance of social aspect of the recovery phase and the
participation of community services as the focus of NGO practices. This recognition also
inferred that the organization as a whole was perceived to have the role of
complementing the practices of the state sector. It was an institutional dynamic

114
For original Chinese script, please see Appendix 5.2.3.06.

498
represented the actor’s perceptions of a joint type of action between the civil society and
state rather than that of a confrontational one
115
.
During the long term disaster recovery period, more civil society actors that came
into being after the disaster started to rank higher in their tendency for group closure. For
example, looking at table 5.2.16, while actor #3 and #49 remained to be more likely to
collaborate with those having the same registration status with themselves, other newly
formed civil society actors include #123 and #64 joined this category at this time period.
At the same time, actors #135, #93 and #61 kept their tendency in collaborating with non-
registered actors. Therefore, what we can conclude for the collaborative structural
characteristic during disaster recovery was that the informal social groups tended to
become more supportive for each other’s growth over the long run after the disaster.
These non-registered civil society actors also tended to be more committed through
collaboration relationships among each other. At the same time, the non-registered
groups also seemed to have sustained outside institutional support from the registered
nonprofit organizations both for the emergency response and the recovery period. Within
the domain of civil society, what can be inferred is that the general institution-building
environment tended to become in favor of the longer term growth of the non-registered

115
For further details in Chinese script, please see Appendix 5.2.CaseNGO49.4.

499
grassroots groups. The availability of choices and opportunities for newly formed civil
society actors to establish collaborative projects existed both inside and outside
registration status groups. What this further revealed was a set of self-grown and self-
recognized embedded capacity for civil society actors in this context to institutionalize
and develop.

Structural Foundations of Sustenance and Support
Top-down Approach
Component
Let’s now look at how the sustenance structure for agency action can be
characterized when represented by the collaboration network environment. Table 5.2.17
below shows the development trends for weak and strong components throughout the
three stages of time before and after the earthquake.
Table 5.2.17. Development Stages of Weak and Strong Components in
Collaboration Networks
Time Period Weak Component
(Action)
Strong Component
(Condition
for Institutionalization)
t1 92 134
t2 38 117
t3 34 108

500
The number of both weak and strong components was the highest during the period
before the disaster. There were a total of 92 weak components and 134 strong
components at this time. For the former, the largest grouping can be found among the
collaboration activities of 44 actors as illustrated in figure 5.2.5.

Figure 5.2.5. Collaboration Network Pre-earthquake Weak Component Structure
The isolated tree structure composed of 4 actors colored in black on the right hand
side of the graph is counted as one weak component by itself. The rest of the weak
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99

501
components are composed of isolated actors that either had not yet come into existence or
were functioning alone without building any collaboration ties with any others at this
period of time. From the results showing the counts of strong components
116
, or when
directions of ties were being accounted for, the structure was separated into even more
pockets of groups. The largest strong component at this time period is composed of only
three actors.
Shortly after the earthquake, we can see that the collaboration structure became
less segregated by groups as the number of weak and strong component decreased
117
. The
weak components went down to 38 and the largest grouping at this time turned out to be
composed of 100 actors, which accounts for 72.5% of those in the network. Graphically,
this group can be observed in figure 5.2.6 as the one colored in red. Correspondingly, the
largest strong component also increased from composing of 4 actors to 19 actors during
the emergency response period.

116
Please see Appendix 5.2.3B1 for UCINET output.
117
Please see Appendix 5.2.3C1 and Appendix 5.2.3C2 for strong and weak component results.

502

Figure 5.2.6. Collaboration Network Emergency Response Weak Component
Structure
What this dynamic revealed was that as the collaboration structure pulls more actors into
the main component due to the active search on the part of civil society actors, they also
became more embracing in developing stronger relationship ties with an increasing
number of others and were more willing to put aside their differences in terms of various
kinds of attribute characteristics. In other words, civil society actors’ capability of
exercising discretionary choices in building collaborative projects across the network was
accompanied by a process of structure integration and expansion.
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117 118
119
12
120
121
122 123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99

503
The similar trends followed during the long term disaster recovery stage. Looking
at table 5.2.7, the action network became more integrated to include increasing number of
actors into the collaboration relationships. This was revealed through the weak
components decreased from 38 to 34. Furthermore, the number of actors in the largest
weak component
118
also increased to 105, which accounted for 76.1% of actors.
Graphically, this is illustrated by the main connected component consisted those actors
colored in red in figure 5.2.7.

Figure 5.2.7. Collaboration Network Recovery Weak Component Structure

118
Please see Appendix 5.2.3D2 for weak component results.
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19 2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99

504
Note that this long term stage also marked the disappearance of isolated tree
collaboration structures. Except those civil society actors that did not participate in the
collaboration activities during this period of time, the rest of the actors were jointly
connected regardless of the direction of the ties. As the main collaborative component
emerged as the result of commitment-oriented agency action, the number of actors in the
largest strong component
119
kept increasing from 19 during the emergency period to 30 at
the recovery stage.
Overall, these structural changes in both the weak and the strong component for
collaboration networks turned out to have similar trend patterns as those in the
communication networks. One can interpret the dynamics working behind these sub-
structures as two types of forces that shaped the general structural frameworks for the
sustainability of agency and further toward institutional change. The formation of the
weak components was driven by a force of agency action on the part of civil society
actors. It worked in such a way that pulled more actors together in expanding the main
networked components not only for information exchange but also project collaboration.
The emergence of strong components was further driven by actors’ desire and capability
to strengthen the connections with a selected group of others. For disaster response and

119
Please see Appendix 5.2.3D1 for strong component results.

505
recovery contexts in this study, this force in driving choice selections among actors
towards further grouping also promoted expansion of stronger ties among an increasing
number of actors in one particular component.

K-core Analysis
I now move on to conduct a k-core analysis for the collaboration network. Recall
that the main difference in examining the communication network and the collaboration
network was that they each revealed a distinct sphere of connection from the civil
society’s perspective. The behavior of collaboration tended to operate at the level of
institutional commitment as actors maintained their agency actions in terms of groups and
organizations. Therefore, finding the core set of actors in the collaboration network
structure can be understood as their willingness to seek after a higher level of resilient
strength particularly through the institutionalization process.
For the period before the earthquake, the core structure was found at the 3k-core
level. Note that this was a relatively low degree level to draw the boundary of the core.
The actors colored in red in figure 5.2.8 were not only few in number but also had rather

506
sparse collaborative connections with others in the connected network. This also shows
signs of less intensity of connectedness among actors in the core at this period of time.
Figure 5.2.8. Collaboration Network Pre-earthquake K-core Structure
Also, when comparing to the number of core actors being found in the
communication network at this same time period, there were less actors inside the core of
the collaboration structure. Despite all this, the state and the market actor (#1 and #2)
both belonged to the 3k-core structure. This means that before the earthquake, there was
already a close bonding among the state, the market and the core members of the civil
society.
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Colors: Identification of K-core structures
Blue Color: isolates

507
Shortly after the disaster, the core structure was being found at a slightly higher
4k-core level. The actors inside the core were then consisted of those that had
collaboration connections towards at least 4 other actors. This means that the project
collaboration initiatives became increasingly intense, thus more cohesive for the core
members. In addition to the increased cohesiveness inside the 4k-core structure, the
number of actors inside the core also experienced an increase. A close examination of the
composition of the core members in the collaboration network indicated that such
increase was mainly the result of the activities of civil society actors. The state and the
market actor remained to be perceived as key collaborating partners at this stage of time.
Therefore, two ongoing structural processes can be depicted as tending to generate a
change in the collaboration pattern behavior. One was the increasing commitment on the
part of civil society actors inside the core. The other was the higher intensity and
cohesiveness among the expanded set of core actors. The two change dynamics together
generated a collaborative structure shown in figure 5.2.9, as the core members were
represented in red color.

508

Figure 5.2.9. Collaboration Network Emergency Response K-core Structure
The collaboration behavior was characterized by highly embedded actors in the
core structure with increased participation of civil society actors. During this period of
time, both actors #49 and #24 emerged as new core members. Here, let me qualitatively
examine the motivational sources of the emergence for both of these actors in order to
trace the formation process that contributed to their core location in the emergency
response collaboration network.

1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122 123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Colors: Identification of K-core structures
Blue Color: isolates

509
The Case of Actor #49 (NGO49-01)
Several factors can also be attributed to the sources of emergency for actor #49
within the civil society domain. First, although the actor’s formal establishment was in
2008 after the earthquake event, its initial actions and functioning as a civil society group
since the year 2005 meant that the organization as whole was not a completely newly
emerged entity in terms of the timing being considered. However, when the factor of
action was being considered, the self-organizing action enacted by the “coming-
togetherness” of groups and NGOs across China functioning as a “joint office”
immediately after the earthquake event signified a transformed level of emergence inside
the civil society domain. This type of emergence was first characterized by an initial joint
identification process through which different types of actors functioning inside the civil
society sought out each other performing similar tasks related to disaster response. This
transformation dynamic was being further stabilized as the joint entity formally
established itself in the disaster-impacted area. Therefore, the “action” point of view
presented a quite different nature of this civil society actor emergence through its
transformation process after the disaster event.

510
Self-initiated “solidarity” and “cohesiveness” could be described as some of the
natures of emerging sources from which this particular organization arose. This could be
illustrated by first comparing the formation and self-organizing process of actor #49
(NGO49) and actor #3 (SG3). Recall that when the latter was first formed, the dynamics
took place in a pre-determined social environment among acquaintances. On the contrary,
when the actor #49 was first initiated, participants in the newly established “joint office”
were “all complete strangers” (NGO49-01) from the accounts of the organizer. This case
reflected a different aspect of the source of civil society emergence in the context of a
crisis situation. Within such a short period of time immediately after the earthquake, the
connection bond that formed among civil society actors originated from what the
organizer (NGO49-01) called a sense of “common mission and value”, rather than from a
sense of trust that usually develops over longer period of time.
LU: If none of you know each other beforehand, how has the level of trust been
built?
NGO49-01: I think this is an issue of mission and value-system. We are working
toward the same mission, in fact, I think building trust is not an issue for the short
term. This is because the time span was very short back then (emergency response
period), almost like for a month or so. We all had a very strong sense of mission,
and that is what I think was important, we were the so called “like-minded”
people”. (NGO49-01-03
120
)

120
For Chinese script, please see Appendix 5.2.3.07.

511

As we can see, within the one month mark after the earthquake, which was
commonly defined as the emergency response period, the strength of the relationship
bonding process was perceived to be primarily originated from a common sense of
mission towards similar goals. The shared desire to take actions to provide voluntary
assistances for the disaster-impacted areas overcame the concerns for trust-building
during the emergency response period. Such was the source foundation from which the
“cohesiveness” was formed when looking at the actions being taken among all the civil
society groups/organizations participated in the joint office of actor #49.
The transitioning period from emergency response to short term and long term
recovery also witnessed signs of institutional structural change of the actor #49. The
initial “joint office” established at the very first moment of response stage was eventually
replaced by a formal field office intended to function over the long term. From the
account of the organizer (NGO49-01), the institutional structure of the “joint office” was
constructed as a “temporary collective entity”, while the field office established during
the beginning stage of the recovery phase was a “permanent agency”. The latter’s
“governing structure works in the same way as those of other formal NGOs”.

512
As the actor emerged from a temporary office towards one that focused its
permanent field practices in the longer term, the organization’s activities related to the
recovery phase can be categorized into the following three kinds of projects. The first
type of practice that the organization focused on was the establishment of the “public
spaces” inside local communities that were impacted by the earthquake. These “public
spaces” were built through community centers that provided services for various groups
of local people. For example, services for taking care of the children, elders,
entrepreneurship and job training, as well as microcredit supports. The second type of
activities was related to “social enterprise ventures” by supporting local hand-crafted
jewelry-making as a way for local people learning to establish their own livelihood in the
longer term. At the time of the interview, the organization explored such projects in three
local communities that were significantly impacted by the disaster, and they were in:
Mianzhu, Shifang, and Dujiangyan. The third type of projects that the actor as an
organization focused on was of similar kind to those engaged by actor #3 (SG3), which
was the formation of an “NGO support platform” (NGO49-01) particularly providing
assistances for smaller domestic grass-roots. While the practices of actor #3 was
perceived as supporting platforms in information and communication, the platform
initiated by actor #49 was more of a financial support for other smaller civil society

513
actors particularly interested in initiating practices related to “community services”. To
put it in the words of the organizer:
We have provided support for seven to eight this kind of small grassroots groups
every year. Because it’s a small amount of money, for about one million, we
could support seven to eight such small groups/organizations to engage their
works in community services. The experiences from these two years have been
quite successful. Many of the small grassroots with only one person on staff have
developed tremendously, some grew faster than the speed of our growth back then.
So it’s been great, I think this kind of support is a much more meaningful and
interesting one. (NGO49-01-04
121
)
From here, we can see that there was a key difference in the types of institutional
emergence between actor #3 and actor #49. Despite of their similar core engagement in
building civil society support platforms during the disaster recovery phase, they had
diverging focuses in the types of services being provided towards the development of
civil society actors. The former aimed at providing a bonding environment that facilitates
the interactions among actors in the civil society domain. In other words, the long term
“cohesiveness” among actors emerged as part of the agency source in its actions. On the
other hand, actor #49 focused its actions on “particularity”. Its practices aimed at the
individual growth of those actors who just emerged as grass-roots NGO. Despite of these
differences in action, it was the service practiced towards the assisting the institutional

121
For Chinese script, please see Appendix 5.2.3.08.

514
development of others inside the civil society domain and seeing them grow in “strength”
over time that instilled a sense of satisfaction and fulfillment for the participants from
both entities. These can be thought as a type of “indirect” source of “solidarity”
formation revealed through the network mapping investigated in the earlier chapters
122
.

The Case of Actor #24 (NGO24)
When touched upon the topic of institutional action during the emergence and the
recovery stages, the director of actor #24 enthusiastically related back to his own
experiences in this regard. Throughout my own conversation with him on this subject at
the time, I noticed that his tone of voice was filled with passion and care towards the long
term development of grass-roots NGOs. His personal dedication to the works of building
up a Chinese civil society was also further reflected through his awareness of the need for
future capability growth for civil society groups and organizations. The following
paragraph illustrated a summary of the account:
For grass-roots groups and NGOs emerged and took actions after the earthquake
event, finding different ways in keeping a balance of the “interest demands” from
the government side and the general mass side is critically important. After the
earthquake, the immediate needs were concentrated towards housing
reconstruction which was mainly led by the government. These activities were
regarded as the “hardware” side of reconstruction. The NGOs were in advantage

122
For detailed interview account, refer to Appendix 5.2.CaseNGO49.2

515
in social reconstruction, which can be counted towards the “software” side of the
recovery process. These social recovery related activities were considered to be
more compatible with what the NGOs were good at practicing in the long term
stage. On the one hand, from the perspective of the government, the earthquake
recovery was already (at the time of the interview) towards completion since its
attention for the recovery stage was primarily focusing on reconstructing the
“hardware” part. On the other hand, from the perspective of NGOs, the recovery
process might have just begun. However, there were all kinds of difficulties
surrounding the long term functioning of grassroots NGOs, such as the exhaustion
of sources of financial support and human resources. At the same time, the
activities that civil society actors most dedicated into were related to community
development and this area of practice would have a demand in the long term and
can be sustained over time. (Summary account of NGO24-01
123
)
Recall that such an awareness of the long term role of civil society and a refined
definition of “disaster recovery” from a social development perspective were also in
accordance with the visions shared by participants from actors illustrated in earlier
sections. This showed the long term orientation of civil society actors in terms of an
awareness of their roles in making the connection between disaster recovery and long
term social development of the country.
Continuing examining the k-core results over the long term recovery stage, actors
in the core structure developed further cohesiveness by being inside the 5k-core. Each of
the core members had at least 5 collaboration ties with others, which was one connection
more than those core actors during the emergency response stage. Figure 5.2.10 shows

123
For original Chinese script, please see Appendix 5.2.CaseNGO24.2.

516
the graphic representation of the 5k-core structure consisted of actors colored in red. Note
that the recovery stage also marked the integration of the previously existing isolated tree
structure into the main connected network.

Figure 5.2.10. Collaboration Network Recovery K-core Structure
Therefore, what we can say regarding the development of project collaboration
network structure is that it tended to increase in cohesiveness and intensity at the core
level. The collaboration behavior for the core actors mainly focused on building more
connections over time. It is another piece of evidence showing not only the high level of
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19 2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Colors: Identification of K-core structures
Green Color: isolates

517
commitment in terms of the increasing ties being built among core actors but also an
increasing willingness for civil society actors to be engaged in the longer term.
Furthermore, the state and the market actors were consistently perceived to be critical
collaboration partners by civil society participants both before and after the earthquake.
The policy implications from a disaster mitigation perspective based on this outcome
would be the possibility of mapping out the differentiating role of state and market actors
from the perspective of civil society actors. With further information regarding the types
of project collaboration practices for actors inside the three domains, a trend detailing the
nuances in collaboration according to activity types and actor traits can be made clear. In
the case of China, the collaboration among the three domains experienced a process
towards high cohesiveness and intensity in terms of tie connections. This means that after
the earthquake, from the short term to long term recovery, civil society actors learned to
put a high value on their collaboration relationships with the government agencies and
private businesses. In general, a recovery planning policy that aims to engage the civil
society in this kind of structural environment can generate a higher level of participation
and much more welcomed by grassroots organizations.
From figure 5.2.10, also can be noted is that during the long term recovery stage,
for the first time, actor #97 emerged as a 5k-core member alongside with other newly

518
established civil society actors such as #123 and #49. This type of behavior deserves a
further qualitative examination on the factors that contributed to its emergence at the
group level.

The Case of Actor #97 (SG97)
In the case of actor #97, the motivational factor for civil society emergence can be
traced back to the ways organizer interpreted several disaster related concepts as well as
how they were connected to the organizer’s (NGO97-01) own personal experiences after
the earthquake event. Along with actors #3 and #49, the type of agency actions that were
being valued among the participants across all three civil society actors was the desired
contribution made towards the development of a social capacity for local communities to
cope with “unforeseen uncertainties” or risks in general. These uncertainties or risks were
perceived either as results from a natural disaster or from other types of crisis situations.
Participating in the long term recovery period of the 2008 earthquake aroused a type of
inherent motivation for building a society that is “resilient” in aspects beyond the needs
to recover from one specific incident. The aspect of “sustainability” for a society to
function in a “healthy” condition at the community level had been a consistent theme

519
throughout the development of three actors, while each of them defined “healthy” in
different ways. In order to perform these functions to contribute to the making of a
“resilient” society that is sustainable in terms of its ability to cope with risks, one
characteristic that was perceived to be essential was stated as the “indigenous-ness” or
“home-grown” civil society actors performing in this area of practices. The organizer of
#97 particularly referred such “bottom-up” nature as those rooted in the efforts of
Chinese people themselves.
The following were several reasons that the organizer of actor #97 gave to support
her position as such. First of all, when the participation focus was in initiating long term
development–oriented projects, reliance on “foreign funding agencies” often would not
serve as a sustainable source when civil society activities were transforming from
emergency response to long term social recovery and development. While “foreign funds”
were relatively easy to obtain especially immediately after a catastrophic disaster, they
are mostly “temporary” supports and rarely turn into long term assistance. If “real change”
inside the society is the goal, “the power and strength developed from the society itself”
would be critical. And these “strength” would not only include the appropriate domestic
funding sources but also the people initiating and running of domestic NGOs. After the
emergency short term response period, the attention coming from foreign agencies was

520
perceived to be “limited”, especially under the condition when life transitioning at the
ground level takes more than a few moments of emergency supports. When disaster
recovery is perceived as a longer term endeavor incorporating the intentions for social
capacity-building in dealing with risks, it is that “indigenous” source of “strength” that
will be eventually “reliable” for supporting and sustaining long term-oriented actions.
Secondly, the carrying out of this type of sustainable long term practice on the
part of civil society actors in general could not be realized without the role of the state.
If NGOs in China want to survive and function long term, the role of government
is absolutely critical. This is because of the coordination factor. You cannot avoid
dealing with the political system when organizing your activities. For example, it
is impossible for you to implement the activities without the permission of the
government, this is the pre-condition. At the same time, the coordinating power of
the government is strong too. You have to admit it objectively. The only thing is
that its ways of functioning is different from us, but can generate significant
impacts. (SG97-01-06
124
)
The appropriate coordination with the government through all kinds of NGO activities
was treated as a pre-condition for long term functioning of Chinese NGOs. The
consideration of being informed on the part of the government is regarded as an active
sign of its cooperation with the NGO activities in the field. Again, the complementary
role of civil society in this particular political and cultural context was well recognized.

124
For original Chinese script, please see Appendix 5.2.3.09.

521
Note that such “complementarity” was perceived to be revealed in NGOs functioning
through different task channels aside from those traditionally performed by the state. For
the case of this particular actor #97, such difference arose from the diverging ways of
making fundamental changes not only to how people live their lives but also in how their
ways of thinking and perception in developing a “resilient” society that would be well
equipped for crisis situations. Government, on the one hand, was well-fit for conducting
direct informing-oriented activities for the target population. Civil society actors, on the
other hand, was advantaged in exerting their strength through a process of “gradual
influence” that requires “a very long time for trail-and-error and adjustments” with close
proximity and emersion towards the local communities
125
.
Therefore, the emerging 5k-core membership of actor #97 during the long term
recovery stage generally arose from two factors related to its institutional establishment.
One was the focus on developing indigenous sources and strength for disaster recovery
and long term social development. The other was the recognition of building a resilient
society through a very gradual process of influence from the practices of civil society
actors.

125
For detailed interview account, refer to Appendix 5.2.CaseSG97.2

522
Community Structuring in Project Collaboration

The Girvan-Newman Detection Method
In Part I of this Chapter, I have shown that communication actions inside the civil
society domain after the earthquake actually worked in favor of generating cohesiveness
response and recovery efforts. The community structures developed before the
earthquake quickly dissolved after the disaster and this happened at the same time when
new actors were emerging in the field. This kind of timeliness in terms of information
exchange dynamics was a key indicator in perceiving the level of willingness for civil
society actors to take action after a catastrophic disaster and their desire to make an
impact for response and recovery.
In this section, I compare and contrast the community structure development of
collaboration networks over the three time periods. First of all, the color representation in
figure 5.2.11 below shows the different communities being detected by the Girvan-
Neman method
126
in various color representations. It is apparent that the largest
connected community being identified was composed of the nodes colored in red.

126
For Gephi graphs using the Louvain method, please see Appendix 4.2.3G1.

523

Figure 5.2.11. Pre-earthquake Collaboration Network Community Structure
(Partition=8, Highest Q Modularity Value=0.391)
Aside from the isolated actors (colored in blue), the rest of the network can be divided
into 8 community groupings. Note that four actors colored in black on the right hand side
of the graph formed an “isolated tree” structure and had singled themselves out from the
main network structure. The main connected part of the network was further divided into
7 communities. Also note that the state and the market domains were found to be in the
community with the largest number of actors. In terms of building collaboration projects
with the government agencies and the private enterprises, the role of several actors
became critical by observing the connections among the communities. For example,
actors #70, #20, #88, #24, and #51performed as “bridges” for the members in their own
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Size: Out-degree
Color: communities
Circle: registered
Square: non-registered

524
community to build collaborative ties with actors either in the state or in the market
domain. Others from their communities had to connect through these five actors to realize
the opportunities for collaboration with actor #1 and actor #2.
This leads us to another structural characteristic regarding the main component
for this network. To think of it in another way, actors #1 and #2 in turn also became
“bridges” for members in the smaller communities to make collaborative connections
with those resided inside the largest community. Note that with the exception of #70, the
collaborative ties that the government sector and the business sector had with actors #88,
#24, and #20 turned out to be the only channel through which their community members
could access or possibly build future collaborative ties with the ones in the larger group
colored in red. Recall that by research design, the state and the market aggregate were
embedded into the civil society network without any outgoing ties. All the tie connections
that they had were being named by those actors inside the civil society domain. Their role
in making critical connections between communities at this time stage revealed their
perceived significance in being project collaboration partners from the perspective of
civil society actors before the earthquake. On the one hand, nonprofit groups and
organizations did build up their own community within which they developed closer

525
collaboration ties among each other. On the other hand, they realized the importance of
making long term attachment to the state and the business sectors.
The project collaboration community structure was being significantly altered
shortly after the earthquake event. Figure 5.2.12 shows the community composition for
this period. Visually observing the graph, we can see that the main component of the
network became denser as more actors were engaged in the largest connected community
for collaborative projects.

Figure 5.2.12. Emergency Response Collaboration Network Community Structure
(Partition=10, Highest Q Modularity Value=0.247)
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Size: Out-degree
Color: communities
Circle: registered
Square: non-registered

526
Based on the Girvan-Neman detection method, there were 10 communities. The
community colored in red was the one with the most members and was certainly
composed of more actors than the largest community found in the previous period. One
“isolated tree” structure was still disjointed from the main connected part of the network,
and it was found between actor #103 and #42. The rest of the smaller communities were
the ones with significantly less collaborating members inside them. There were still a
number of actors performing the function of “bridges” that connected those others whom
would otherwise become isolates. For example, actors #101, #107, #62, and #106 were
all found to play the role of mediating the “hangers
127
” that were attached to them on the
periphery to the main component of the connected network. The appearances of these
kinds of structures showed that the collaborative actions still had certain degree of
disintegrated-ness.
However, one noticeable structural change during this period of time was the
decreasing number of civil society actors who performed as “bridges” linking possible
collaboration ties with the state and the market sectors. Note that the two aggregate
domains became further embedded into the largest community. Actors from other smaller

127
“Hangers” are formally defined as “points that are connected to a member of a cyclic component, but which do not
themselves lie on a cycle. Hangers simply ‘hang’ on to a cyclic component” (Scott, 2001, p106).

527
communities that could reach the government agencies or the private enterprises for
collaboration ties also developed various alternative “routes” with those that are in the
largest community. What this meant from a structural change point of view was a pattern
of dynamics that civil society actors created for further attachment both “horizontally”
and “vertically” to others in the network. By “horizontally”, I mean more collaborative
initiatives were being made among civil society actors themselves. The impact of this on
the structure was an increase in interconnectedness in terms of the availability of
channels for collaborative projects developments through indirect ties. By “vertically”, I
mean that the actions were being further concentrated towards higher out-degrees, or the
emergence of those who were increasingly active in reaching out to others for
collaboration at this period of time. Furthermore, the state and the market domains were
no longer perceived as the only “bridges” for the smaller community members to connect
to the rest of the others. Civil society actors started to actively look to others who took
actions in similar activity areas for collaboration. While actor #1 and #2 continued to play
a significant part in the life of civil society actors when examining the intensity of two
aggregates’ in-degree, since the earthquake event, their existence were recognized as only
part of the collaborative functioning for civil society actors. The roles played by others in
the civil society domain were being recognized and activated at this stage of time.

528
This type of structuration process continued during the long term recovery stage.
One of the marked differences when comparing this period to the previous emergency
response stage was the disappearance of the disjointed isolated tree structure formed
between actor #103 and #42. It is apparent from figure 5.2.13, that both actors were
drawn to become attached to the main component of the collaboration network. Actor
#103 is a member of a smaller sized community during the long term recovery, while
actor #42 initiated a collaboration tie with actor #51 and both of whom belonged to the
largest community being detected.

Figure 5.2.13. Long-term Recovery Collaboration Network Community Structure
(Partition=9, Highest Q Modularity Value=0.130)
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19 2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
Size: Out-degree
Color: communities
Circle: registered
Square: non-registered

529
Looking closely at the attachment structure of all other smaller communities to
the largest community grouping, notwithstanding that there were still some “hangers” on
the periphery of the structure, one can find that there were significantly less of those who
performed bridge roles when compared to those existed before the earthquake and the
shortly after the disaster. What this means is that actors were being further drawn into the
main component of the network and their collaboration ties also became diversified
across a variety of pathways.
From looking at the changes in component structures and the community
groupings inside the collaboration networks over time, a few points can be summarized
regarding the tendency towards institutionalization process after the disaster. First of all,
the general collaboration structure went from actors being loosely attached with a sense
of disintegration to one that attracted attachment with tendencies to integration as a
whole. Secondly, actors were being increasingly drawn to have collaborations with those
inside the larger community. This process left the boundaries among the different
communities less distinctive over time. What this signifies is that after the earthquake, the
pattern of the collaboration structuration was altered from one that showed more
segregation among community groupings to one that promoted cohesiveness and
conviviality among actors. Such a tendency also built itself up in intensity so as to sustain

530
the similar patterns of change in the longer term. Thirdly, the attitude and behavior of
civil society actors went from one that was more focused on building particular
partnerships with a specific set of actors to one that valued diversity, especially in the
context of cross-sector collaborations. In other words, the pursuit of commitment was no
longer characterized by “single-mindedness” and “boundary-protection”. Rather, with a
sense of “open-mindedness”, civil society actors were able to approach project
collaboration in such a way that enhanced their ability to connect further with more
others across the entire network. What this revealed about the character of civil society
over the long term was that its action incorporated a “vision” of integration instead of
“near-sightedness” in its way of perceiving and acting.
This kind of embrasiveness and open-mindedness toward integration can be
further represented by the community structural changes detected by the alternative
detection method (Blondel et al., 2008). Note that the initial integration for the
collaborative structure can be illustrated by a decrease in the number of communities but
with an expansion of members inside each community found in the connected network
immediately after the earthquake (see figure 5.2.15 and figure 5.2.16). With the
emergence of new civil society actors, the collaboration behavior adjusted in such a way
that actors were being further drawn into the different kinds of commitment-oriented

531
activities over the long term. Therefore, over time, one can observe a gradual expansion
in the diversification of the number of collaborative communities in the connected
network (see figure 5.2.16). The “vision” for integration on the part of civil society actors
was activated by their search for a variety of attachment styles.

532

Figure 5.2.14. Collaboration Network Pre-earthquake
128
Community Structure
(Blondel et al., 2008)

128
Please see Figure 5.2.2A for the original network graph in Appendix 5.2.2.
Size: Out-degree; Color: Communities

533

Figure 5.2.15. Collaboration Network Emergency Response Community Structure
(Blondel et al., 2008)

Size: Out-degree; Color: Communities

534

Figure 5.2.16. Collaboration Network Recovery
129
Community Structure (Blondel et
al., 2008)

129
Please see Figure 5.2.2C for the original network graph in Appendix 5.2.2.

Size: Out-degree; Color: Communities

535
Bottom-up Approach: Structural Transformation of Civil Society
Overview
The process of identifying “maximal and complete” sub-structures or “cliques”
for the collaboration networks turned out to be more complicated than that for the
communication network structure. Recall that for a directed network taking into
consideration of who sends and receives a tie, the concept of clique requires the actors
inside such an environment to be not only directly connected to all others but also each of
the connection to be reciprocated. The cliques that were found in the communication and
collaboration networks were therefore “strong cliques”. In the previous chapter, I found
that the clique analysis revealed a strong tendency for civil society actors to build such
“tightly-knit-together” sub-groups for communication and information exchange
purposes. However, such strong directly reciprocated connections were rare for project
collaboration. The initial analysis on the collaboration networks did not identify any
strong cliques for the period before the earthquake. In addition, only one clique
composed of three members was being found for the two time periods after the
earthquake. This could have two characteristic implications for this structural
environment at the most stringent definition level of cliques. First, although actors could

536
be engaged in close communication relationships with a set of others, but when it came to
building project collaborations, the relationships were not only much more loose but also
with less tendency for “exclusivity” behavior such as one determined inside strong
cliques. Secondly, the nomination of one collaboration tie from an actor towards the other
was less likely being reciprocated. On the one hand, the inclusion of actors who were
non-responsive to the survey and the generalization of aggregate state and market actors
had contributed to the inherent non-reciprocation of ties. On the other hand, the definition
of “collaboration” could vary based on actors’ interpretations of the circumstances that
could be counted towards such activities. Regarding the former circumstance, the
network analysis offers an alternative method to take the non-responsiveness and
aggregation features into consideration. And it is called “n-clan” analysis. For the
following sections, I will utilize this concept to discuss the sub-structural characteristics
of the collaboration networks. Regarding the second circumstance, it can be interpreted as
a sign that the concept of “collaboration” and its functions might be derived from
different contexts, particularly when taking into consideration individual perceptions.
And this aspect of consideration is a limitation of this research and is worthy of further
exploration in future research studies.

537
In order to identify the development of n-clan structures in the collaboration
networks, it is necessary to first provide a conceptual clarification of the origin of “n-
clan” in network analysis. Recall that use of the concept of cliques as maximal complete
sub-structures put a rather stringent requirement on finding closely “attached-together”
groups for collaboration structure environment. The finding from the clique analysis
revealed that it is uncommon for actors to engage in such tightly-knit-together
collaborative projects both before and after the earthquake. However, this does not mean
that actors were not involved in closer relationships if the restrictions can be slightly
relaxed to a certain degree. The detection of “n-clique” is one way to do so. When n=2,
the method will identify sub-groups by including those “friends of friends” indirect
contacts through a common neighbor rather than strictly focusing on the direct
connections. However, one drawback for such n-clique analysis was the possibility of
including an actor that played a role in connecting n-clique members but might not have
ties to all clique members, and thus not part of the clique itself (Scott, 2001). The result
of using this method was that one might end up including too many distant “friends” of
clique members into each n-clique. One way to focus entirely on the clique members
alone was to restrict the diameter of the cliques found to be no greater than n. For
example, if I identify 2-clan sub-groups, this means I am only examining members no

538
further than 2 path steps away from another and all connections would have to be reached
by way of another member inside the n-clique. The detection of n-clan, therefore, would
yield tighter bottom-up sub-groups by paying closer attention to those who were
members of the n-clique. Another point worth noticing was that the n-clan detection
method ignored the directions of the collaboration ties. In terms of interpreting the results
given this condition, as long as one actor made a nomination of project collaboration
towards another actor, the tie between the pair of actors would be counted in n-clan
analysis. In other words, the variation in understanding and defining what it means to
participate in collaboration among pairs of actors are being ignored in this study context.

Two-clan Analysis
To implement the n-clan analysis in this study, I adopted a detection method by
restricting n=2. This means that for each tightly knit together sub-group being identified,
all the connections occurred among members inside the clan would be no more than at
distance 2 paths. Compared to the maximal complete sub-structure identified as cliques
for the communication network, this 2-clan approach allowed those who were indirectly
connected to the focal actor by way of an intermediary actor to be part of the sub-group.

539
In other words, it is possible for actors to hear about possible project collaboration
opportunities through a mutual acquaintance who already has engaged in collaboration
activities with the focal actor. The further away or the more steps of pathways that
collaboration ties were being built from some members, the more difficult it was to
interpret the formation process of collaboration ties for those part of a closely attached-
together group. Therefore, this study restricts the number of intermediary connections to
be at the level one so as to observe the behavior of actors within a more tightly defined
diameter boundary.

Pre-Earthquake Stage
For the period before the earthquake, a total of 18 2-clans were found in the
collaboration network
130
. The minimum number of members in a 2-clan was 3 and the
maximum was 11. Observing the member composition across the clans, the state actor #1
appears to be an important collaboration partners from clans 1 to 7. Among these seven
2-clans, three were represented by both the state and the market actors, in addition to the
participation of other civil society actors in the same sub-group. This could be understood

130
Please see Appendix 5.2.3E1 for UCINET output.

540
as one primary piece of evidence showing signs of inter-sector collaboration involving a
limited number of civil society actors. This also demonstrated that the participation of
civil society actors in project collaborations was not completely absent from even before
the disaster in the Chinese case.
Examining through the “hierarchical clustering of overlap matrix
131
” at the actor
level during the pre-earthquake stage, one can see that at the highest clustering level,
actor #110 and #37 joined together as particularly close to each other in terms of sharing
common membership together. As the stringency of clustering becomes more relaxed at
the next level, actor #1 and #2 were actually attached together in terms of incidences of
sharing clan membership. There are a couple of implications that can be said regarding
this particular outcome. Recall that the grouping status of the state and market aggregates
were derived from the collaboration linkages each of them had with civil society actors.
In other words, by way of the nominations from civil society actors on their perceived
conception of collaboration activities, the government agencies and the private
enterprises could indeed be categorized together in terms of being close in sharing
collaborative projects together. Not only so, from the perspective of civil society actors,
the activities of those inside the two domains also showed signs of exclusivity. At the

131
Please see Appendix 5.2.3E2 for UCINET output.

541
clustering level of 3.0, one can see that the joining sequence between actor #1 and #2 was
separated from the attachment among other civil society actors. This kind of grouping
behavior was not an ideal one for developing a type of social structure that could promote
commitment-oriented relationships such as inter-sector project collaborations. In this
study, such a tendency for exclusive attachment between the state and the market
domains was being captured from the perspective of actors inside the civil society
domain. This type of perceived attachment for the two aggregate actors turned out to be
persistent when the clustering stringency was being relaxed all the way until level 1.360.
Before then, the structure can be characterized by the separation of a number of different
attachment groups in terms of clan membership sharing. In summary, I use figure 5.2.17
to illustrate the state of the sustenance structure in terms of the dynamics among civil
society, the state, and the market domains for the pre-earthquake period.

542

Stage 1

Figure 5.2.17. Pre-earthquake Cross-sector Structure (Collaboration Network
Dynamics)

Emergency Response
Shortly after the disaster, the number of 2-clans went up dramatically to a total of
86. Just as the trend in clique development for communication networks, actors at this
period of time were also inclined to expand collaboration ties not only in terms of the
variety of others across the network but also tended to develop close bonding
relationships with more actors inside their own clan-circle. Like the structural change
inside the communication network shortly after the crisis, the change inside the
State
Market
Civil Society

543
collaboration structure was characterized by both breadth and depth of sub-group
relationship-building. The maximum number of actors in a 2-clan increased to 29
members. One can also observe that the state sector was being heavily represented in
most of the clans. The actor-by-actor membership sharing matrix identified that the state
was a member of 64 out of a total number of 86 2-clans
132
. From the “hierarchical
clustering over-lapping matrix
133
” at the actor level, the findings showed a major
structural change in how the civil society actors, the state, and the market were being
intertwined together during the emergency response period. At the most stringent
clustering level (49.000), the newly emerged social group actor #3 and the state actor #1
were first joined together as sharing the most clan memberships. This piece of evidence
represented a significant structural break-through in terms of how the civil society and
the state domain formulated collaborative bonding with each other. The two actors were
closely attached by participating in 49 2-clans together, which was the most among all
other pairs of actors in the network. The newly established informal status of actor #3 and
its attachment to the state sector signified the emergence of a bonding process within
which the state domain started to become a critical collaborator penetrating the inner-

132
Please see Appendix 5.2.3F1 for sample UCINET output.
133
Please see Appendix 5.2.3F2 for sample UCINET output.

544
most circles at the grassroots level of the civil society domain. As the clustering level
relaxes, it can be found that the state actor was gradually being closely attached to a
variety of other actors who resided inside the civil society domain. One key difference of
the collaboration structure at this period of time was the separation of the attachment
between the state and the market domain in terms of their close ties with the civil society.
The results showed that while the state actor was tightly knit together with a larger group
of civil society actors as the clustering stringency level was being relaxed, the market
actor seemed to be tightly collaborating with a few other civil society actors that
altogether, their closeness in grouping was separated from the state-civil society grouping
until the clustering stringency level was lowered to 4.071. Therefore, one can summarize
that the dynamics of collaboration structural change went from a perceived bondage
between the state and the market domain towards a process that activated the bond of
civil society actors towards both the state and market separately (shown in figure 5.2.18
below).

545

Stage 2

Figure 5.2.18. Emergency Response Cross-sector Structural Transformation
(Collaboration Network Dynamics)

Long-term Recovery
Moving on to examine the collaboration structure of long term recovery stage, the
first characteristic that stood out was the continuing expansion of the number of 2-clans
groupings, showing the tendency for civil society actors to make further commitment for
institutionalizing collaboration connections among themselves as well as towards the
state and market actors. The total number of 2-clans went up from 86 during the
emergency response period to 109 during the recovery stage. In addition, the sheer
State
Market
Civil Society

546
number of actors inside each one of the 2-clans increased dramatically as compared to the
previous period. The number of membership participation of the state actor increased
from 64 clans to 93 clans. This demonstrated that from the civil society actors’
perspective, the government agencies maintained its sub-group membership participation
commitment and became increasingly engaged in establishing long-term collaborative
relationships with a wide variety of civil society actors. The joint membership for the
state and the market sector almost doubled from 23 2-clans during emergency response to
49 2-clans during recovery. The two actors only shared membership in 3 2-clans before
the earthquake. Whether the perceived actions of the state and the market were joined
together in terms of project collaboration activities among civil society actors at the
recovery stage was identified from the “hierarchical clustering over-lapping matrix”
134
.
The outcome showed that civil society actor #24 and #135 first joined together as sharing
the most membership together. As the clustering level was being relaxed, the state actor
first joined the others as being close in terms of sharing membership together in
collaboration projects. The attachment of market sector towards the grouping behavior of
the state and the civil society actors came later when the clustering level is being further
relaxed. The main difference for the long term recovery period was that the joining

134
Please see 5.2.3G2 for sample UCINET output.

547
sequence of attachment finally brought the state, the market, and the actors inside the
civil society domain together in terms of being closely tied together when it comes to
common clan membership-sharing activities. The clustering divide between the state and
the market actor was not as sharp as the period immediately after the disaster.
There are two points that can be implied from these pieces of evidences. One is
that the collaboration structural change resulted from the emergency response after a
crisis had indeed been transformed into one that focuses on in-depth relationship-building
among the civil society actors and the state as well as the market sectors. The other point
that is worth making inferring from the 2-clan results is that the recovery stage marked an
initial integration of close attachments among the three domains of action, graphically
shown in figure 5.2.19 below.

548
Stage 3

Figure 5.2.19. Recovery Cross-sector Structural Transformation (Collaboration
Network Dynamics)
One unique characteristic regarding the structural change process depicted from
figures 5.2.17, 5.2.18, and 5.2.19 was the active participation of civil society actors in
collaborative projects towards the state and the market sectors shortly after the disaster.
In other words, the role of civil society was an active one in promoting a process of
structural transformation among the three domains.
Lastly, I would like to emphasize the bottom-up sub-group formation examined in
this section. From the clique analysis conducted on the collaboration network earlier,
recall that no cliques were found for the pre-earthquake period. However, for the short-
term and long term recovery periods after the disaster, three civil society actors were
found to be consistently engaged in close collaboration relationships such as one inside a
Civil Society
State Market
State
Market
Civil Society

549
clique. They were actors #3, #24, and #119. Their collaboration ties not only met the
most stringent requirements as a maximal and complete, the connections were also being
mutually reciprocated when taking directions of ties into account. Therefore, the three of
them emerged as a “strong clique” together during the emergency stage after the disaster
and were able to carry the collaboration ties forward into the longer term recovery stage.
Briefly examining the attribute characteristics of the three actors, #3 and #24 were locally
grown grassroots working inside the Sichuan Province. Although actor #119 originally
entered into mainland China as a Hong Kong based nonprofit organization, it had
significant amount of experience in participating in local community affairs since before
the earthquake.

550
Chapter 6
The Autopoietic Civil Society: Rules of Network

Longitudinal Modeling for Network Dynamics
Motivation
In the previous chapters, the examination focused on understanding the macro and
micro structural characteristics of how actors build their communication and
collaboration networks over time. Each of the six structural environments (three
communication networks and three collaboration networks) was investigated separately
to look at how actors were connected and embedded within their local and global network
environment. Descriptively speaking, the analysis demonstrated some of the primary
evidences in understanding the formation, persistence, and the sustenance of the action
structures for both communication and collaboration network environments before and
after the 2008 Wenchuan earthquake. To put it in another way, it was an exploration of
how the resiliency of an independent civil society structure was built after a catastrophic
crisis in China.
However, when facing the unique nature exemplified by each one of the network
structures and trying to synthesize their change patterns over time, the descriptive tools

551
essentially deal with the structuration processes in a static state of being. The examination
of the dynamics of structural changes is conducted for the following reasons
135
. One is to
find out if there were indeed “rules” that would govern the network behavior over the
specified three periods of time. In the context of current research for example, these
governing “rules” would include whether the institutional status in terms of registration
would have an effect on the evolution of communication and collaboration behavior,
whether there were structural tendencies that would affect the specific formation patterns
of the communication and collaboration network development, and most importantly,
whether there was a tendency for the communication structures to co-evolve with the
collaboration structures. Discovering and specifying these rules is an essential step in
making both theoretical and policy contributions for societies to deal with catastrophic
changes that alter the social relationship structures within which human decisions are
being made on a daily basis.
Another reason for conducting this modeling analysis was to specifically
incorporate the time factor into a setting that investigates an actor-oriented social change
process. Theories of civil society and institutions have all paid attention to how social
changes came into being from an actor-attribute perspective, but rarely do investigators

135
Also see the results from Homophily Tests in Appendix 6.1.

552
look at the picture from a relational point of view, which is essentially to develop a
coherent story that allows the interaction of agency decision-making and social structural
change being revealed over time. I will further develop these lines of arguments
throughout the discussions in the various sections of this chapter. Let’s start by
illustrating some of the basic characteristics of the model that was being used in this
research.
In this section, I propose a longitudinal network modeling framework to look at
social group and NGO activities over time in the context of Chinese disaster recovery.
Two types of stochastic actor-oriented models were being proposed. One was for
communication exchange network and the other was for collaboration network.

Research Setting and Design
The Stochastic Actor-Based Models (SAB Models)
Dynamic social networks consist of ties that change over time. If I want to
investigate the rules governing the network evolution, there are a set of basic assumptions
when modeling such dynamics. The following two factors are the foundational
assumptions made in this respect: a) Network ties are not treated as events, but as states
with a tendency to endure over time. Social relations such as the ones studied here

553
(communication and collaboration) satisfy this requirement of endurance once first
established. b) It is also further assumed that changing networks can be interpreted as an
approximate outcome of a Markov process. This means that “for any point in time, the
current state of the network determines probabilistically its further evolution” (Snijders et
al. 2010). Snijders, Van de Bunt, and Steglich (2010) depicted six general assumptions
and I will discuss each of them in accordance with the theoretical and methodological
approach chosen in my research study.

Assumption #1:
“The underlying time parameter t is continuous. The parameter estimation procedure,
however, assumes that the network is observed only at two or more discrete points in
time. The observations can also be referred to as ‘network panel waves’.” (5)

This study had three ‘waves’ each representing the time stage: before the
earthquake, emergency response, and recovery. One advantage of the continuous-time
assumption is that it allows for complex relationships to be formed from micro-steps
rather than “out-of-nothing”. In other words, it “represents dependencies between
network ties as the result of processes where one tie is formed as a reaction to the
existence of other ties” (5). While appealing in the sense that the micro-formation
processes were allowed to be traced and represented, it also posed some challenges when

554
applied to the current research context. One issue was its lack of capacity in explaining
the factors that prompted the formation of a tie between two actors in the first place. In
other words, with the significant patterns of change in network structures immediately
after the earthquake, the structural dependencies between the network before the disaster
and the emergency response period would be challenging to model precisely. This factor
was further discussed when taking into account of the “time heterogeneity” issue into
consideration to enhance the prediction power of the model in this respect. The point that
I wanted to emphasis here is that a qualitative aspect of investigation would make
contribution to discover the motivational side of the factors that drove the initial
formation of connections among civil society actors as the result of the earthquake event.

Assumption #2:
“The changing network is the outcome of a Markov process, i.e., that for any point in
time, the current state of the network determines probabilistically its further evolution,
and there are no additional effects of the earlier past. The total network structure is the
social context that influences the probabilities of its own change.” (6)

On the one hand, taking the network structure at a particular point of time as a
social context that influences the decision-making of actors is in line with the network
approach of this study. The relational context of the social structure that actions were

555
made was an important lens through which I looked at my data. On the other hand, this
would again generate issues in understanding the potential for the model development in
understanding a social process with drastic changes between two waves. Similar to the
first assumption, the challenge here again lies in the particular characteristic of this
disaster recovery dataset with abrupt changes in between time wave 1 and wave 2.
However, for dataset with more than two panel waves, it also possible to propose
improved models with less restrictive assumptions regarding this time dependence
(Snijder et al., 2010). This study thus built up a preliminary foundation in venturing into
the discovery of more sophisticated models specifically designed for civil society actions
in disaster recovery contexts.

Assumption #3:
“The actors control their outgoing ties. This means not that actors can change their
outgoing ties at will, but that changes in ties are made by the actors who send the tie, on
the basis of their and others’ attributes, their position in the network, and their
perceptions about the rest of the network.” (6)

This assumption directly speaks to the theoretical foundation that I chose to guide
this research study. The “agency freedom” (Sen, 1999) being enacted inside the civil
society domain rests upon the proactivity of actors, and it also takes into consideration of

556
the relational environments into their decision-making processes. The duality of purpose
and constraints is being implemented explicitly, and this made an appealing feature of
SAB modeling to be applicable to research taking hold of similar theoretical orientations.

Assumption #4:
“At a given moment one probabilistically selected actor—‘ego’—may get the opportunity
to change one outgoing tie. No more than one tie can change at any moment. This implies
that tie changes are not coordinated, and depend on each other only sequentially, via the
changing configuration of the whole network.” (7)

Within the context of this research, especially the time period between wave 1 and
wave 2 was characterized by communication ties being initiated by actors voluntarily
came together for response and recovery efforts in the disaster-impacted area. The
abruptness of the crisis event, however, poses a challenge for the assumption regarding
the sequential order of tie changes. I will come back to this point when going through
data treatment procedures in the later sections. For now, it is important to point out that
each focal actor, as an “ego”, has an opportunity to make a change to its immediate
network environment and also by way of the entire structure of the whole network. The
emphasis is given to investigating the nature of self-organized relationships emerging
from “ground up”, thus the willingness factor of civil society actors.

557
The network change processes are further composed of two sub-stochastic-
processes (Snijders et al. 2010):
“The change opportunity process, modeling the frequency of tie changes by actors. The
change rates may depend on the network positions of the actors (e.g., centrality) and on
actor covariates (e.g., age and sex).”(7)

“The change determination process, modeling the precise tie changes made when an
actor has the opportunity to make a change. The probabilities of tie changes may
depend on the network positions, as well as covariates, of ego and the other actors
(‘alters’) in the network.” (7)

These two modeling processes highlighted the active agency on the part of actors
being capable of making stand-alone decisions, and “are not regarded as subservient to
others’ interests in any way” (7). This is precisely in accordance with the theoretical
outlook of this chapter in that the Chinese civil society as a “standing-up” domain in its
interactions with the forces of the state and the market.
I adopted the longitudinal SAB model perspective because it takes into account
the dynamic nature of the networks under investigation
136
. From other parts of my
qualitative studies regarding social group and NGO activities after the earthquake, the
catastrophic event not only brought a set of latent altruistic and self-organizing values to
the surface, the understanding of the network changes from these civil society actors was

136
See Appendix 6.2A for further considerations.

558
nowhere near a static account. Making an implicit assumption that the networks at hand
were in equilibrium would be going against the meaning how actors understood civil
society in this research context. In this respect, longitudinal modeling became important
to discover the rules and principles that govern social change.

The SIENA Methods in R Statistical System
“SIENA” (‘Simulation Investigation for Empirical Network Analysis’) is a
software that specifically designed to estimate longitudinal network models. It is also
implemented in the R statistical package and together they are often regarded as the
“RSiena” statistical system. The following discussions are the analysis procedures
through this application. The estimation processes mainly followed the program and
documentation guidelines depicted through the Siena Manual (Snijders et al. 2008) and
the R-scripts available from the Siena web-page (http://www.stats.ox.ac.uk/siena/).
Adjustments were made according to the specificity of the current dataset.

559
Data Treatment Procedures
Missing data treatment
Originally, the survey was distributed to 136 social groups and NGOs with the
original response rate was 46%. Missing actors, particularly in social network analysis,
will have a large impact in analyzing longitudinal models. There were two types of
concerns with this level of missing-ness. One was the theoretical concerns of sound
model-building. The loss of information in terms of how groups and organizations
connect to each other will create problems in fitting models to the data, and this can lead
to difficulties in reaching good convergence when running the specified models. The
parameter estimates as a result, might be biased (Huisman and Steglich, 2008). The other
concern was related to technical implementations. The SIENA methods do allow for
some amount of missing data on network variables, covariates, and dependent action
variables. The missing data is treated as non-informative. When the percentage of
missing data reaches more than 20% on any variable, getting good estimates may be
doubtful (Ripley, Snijders, and Preciad 2012). In their simulation studies examining
possible treatments of non-response in longitudinal networks, Huisman and Steglich
(2008) concluded that a model-based approach within the actor-driven models is the best

560
method to adopt. In the SIENA framework, the authors recommended utilizing the built-
in missing data treatment from the software itself, as compared to reducing the data into
complete cases or imputing the networks into an “alien” structure.
Considering the nature of my data and the current available tools in treating the
incompleteness of responses of longitudinal network analysis, I developed a three-stage
data manipulation process to reduce the missing data effect so that the high level of non-
responsiveness of the original survey will have less effect on further analysis.

QAP Correlation Analysis
The first stage involved examining whether there was correlation between
communication and collaboration networks over the three periods of time. I used the
QAP (Quadratic Assignment Procedures) correlation analysis
137
on the original data
composed of 138 actors (136 civil society actors, one state aggregate actor and one
market aggregate actor) to determine whether the probability of communication ties was
related to the probability of collaboration ties. If the two types of networks were indeed
correlated with each other, I would use the type of network that showed more inclusivity
in relationship-building agency activities as the basis for narrowing down the dataset.

137
The QAP correlation is implemented in UCINET.

561
The discussion revealed the findings by using the QAP correlation analysis. I
examined the question: if there was a communication tie between two actors, was there
likely to be a tie between them in project collaboration? From the descriptive analysis in
the earlier chapters, I also found signs of co-evolution between the two types of
structures. So another question to look into would be: if there was a collaboration tie
between two actors, was there likely to be a tie between them in communication?
In addition to looking at the relationships between the two types of network structures, I
also incorporated two of the actor attributes into the correlation investigation to examine
their relationship with the communication and the collaboration networks. Thus, I
developed three general hypotheses in relation to this part of the analysis:

Hypothesis #1: Information exchange and communication relations were positively
correlated with project collaboration relations, and vice versa.
This hypothesis can be understood in two ways. On the one hand, when groups
and organizations were engaged in communications and information sharing activities,
they were more likely to get to know each other’s work better and develop further trust.
This kind of shared understanding would more likely lead to commitment-oriented
project collaborations. On the other hand, if two actors had been collaborating on

562
projects, the experiences of having to make joint decisions over the duration of the
project would provide unique opportunities for the engaged parties to develop closer ties
with each other, and thus making it easier to communicate as well.
Hypothesis #2: Institutional status similarity was positively correlated with
communication/collaboration relationships.
Here, the institutional similarity was examined through actors’ registration status.
If an organization was formally registered either with the Ministry of civil affairs or
under the business category, would this institutional type in terms of formality make it
easier and lead them in building more communication/collaboration partners?
Hypothesis #3: Geographic proximity of actors was positively correlated with
communication/collaboration relationships.
In other words, if two actors both have field offices operating in Sichuan
Province, their close proximity and understanding of the local culture would facilitate
them building communication/collaboration ties.

563
Pre-earthquake stage (t1)
The following table 6.1.1 and table 6.1.2 showed the QAP correlation results for
communication and collaboration networks, as well as their correlations with the
registration status attribute during the period before the earthquake.
Table 6.1.1. Pre-earthquake Stage QAP Correlations/P-values (with Registration
Attribute)
Communication collaboration Registration Status
Communication 1.000/0.000 0.504/0.000 0.028/0.134
Collaboration 0.504/0.000 1.000/0.000 0.010/0.244
Registration Status 0.028/0.134 0.010/0.244 1.000/0.000

Table 6.1.2. Pre-earthquake Stage QAP correlations/P-values (with Geographic
Location Attribute)
Communication collaboration Location
Communication 1.000/0.000 0.504/0.000 0.011/0.295
Collaboration 0.504/0.000 1.000/0.000 -0.003/0.468
Location 0.011/0.295 -0.003/0.468 1.000/0.000
First of all, the tables demonstrated that there was certain degree of correlation
between the communication ties and collaboration ties at the 0.504 level, and the p value

564
for this correlation is 0.000. At a typical 0.05 level, this correlation is significant
(0.000<0.05). This means that before the earthquake, if there was an existence of a
communication relationship between two actors, it was indeed more likely that it would
facilitate collaboration connections. On the other side, if the two actors were already
engaged in collaboration ties, it was more likely that the two would be open to
communication and information exchange as well. Examining the correlation results
related to registration status and location, the findings showed that none of the correlation
measures were significant at the 0.05 level. This implied that whether an actor was a
registered organization or an unregistered social group would not be making the
formation of either communication and collaboration ties more likely. This was also the
case for the location factor. Whether the two actors were operating in geographic
proximity was not correlated with the possibility that the two building up more
communication or collaboration ties.

565
Emergency response stage (t2)
Immediately after the earthquake, the level of correlation between communication
and collaboration experienced a slight decrease from 0.504 to 0.416 (see table 6.2.1 and
6.2.2).
Table 6.2.1. Emergency Response Stage QAP correlations/P-values (with
Registration Attribute)
Communication collaboration Registration Status
Communication 1.000/0.000 0.416/0.000 -0.000/0.536
Collaboration 0.416/0.000 1.000/0.000 -0.006/0.416
Registration Status -0.000/0.536 -0.006/0.416 1.000/0.000

Table 6.2.2. Emergency Response QAP correlations/P-values (with Geographic
Location Attribute)
Communication collaboration Location
Communication 1.000/0.000 0.416/0.000 0.002/0.447
Collaboration 0.416/0.000 1.000/0.000 -0.015/0.243
Location 0.002/0.447 -0.015/0.243 1.000/0.000
This measure was significant at the 0.05 level. This piece of evidence implied that the
happening of the disaster did not alter the existence of a positive correlation between
communication and collaboration connections among actors. When two actors were

566
actively engaged in communication activities, it was more likely that they would be
participating in joint project collaborations, and vice versa. The consistency of the
correlation between the two types of the networks before and after the earthquake was
one thing, there was also a sense of uniqueness regarding this situation. The fact that the
correlation between communication and collaboration networks remained to be strongly
significant, especially after such a catastrophic disaster, can be developed as one of the
primary indicators that revealed the emerging resilient nature of the Chinese civil society
domain. After such a tragic earthquake, not only the communication and collaboration
connections did not break down, they were being maintained through the active agency of
civil society actors and the existence of one also showed signs of kept enhancing the
existence of the other. The capability of a society in developing “resilience” did not just
depend on its ability to restore to the original state of being before the disaster incidence.
The results of this analysis demonstrated that such capability to be “resilient” also
depended on the social actors’ ability to relate to each other at the time of a crisis as well
as in enhancing the different types of the relationships being built. In the Chinese context,
such a nature of “resilience” was being clearly evidenced by the persistence of a strong
willingness of civil society actors to maintain communication and engage in project
collaboration while having to face the tragic consequences caused by the earthquake.

567
Therefore, “resilience” can also be understood as the social actors’ capability to
strengthen and renew the social structural environments which in turn influence their
decision-making and behaviors. The key driving factor for realizing this kind of
resilience, I would argue, is the awareness of agency freedom being carried out by the
actors themselves. It is essentially an actor-oriented approach to looking at the conceptual
formation of “resilience”.
The other part of this section of the results also demonstrated that neither the
registration nor the location factors were correlated with the likelihood of communication
and collaboration ties. This was another corroborating piece of evidence that showed the
structural transition towards integrated cohesiveness in communication and collaboration
networks. First of all, whether an actor was registered as a formal organization might not
make a communication and collaboration relationship more likely. This can be
interpreted as sign of civil society actors being less inclined to intentionally separating
themselves based on their institutional similarity or location proximity with others. For
the emergency response period, this can be understood as a positive “virtue” of how civil
society reacted to disasters. It is because a cohesive and coordinated response would not
be possible if actors cared about grouping each other based on these two attributes and
became exclusive in terms of “picking out” the ones of some pre-determined preferences.

568
The communication and information sharing ties would experience more “disruptions”
through pathways. The evidences at hand, however, showed a completely different
picture from this scenario. There were encouraging signs demonstrating that the
communication and collaboration network connections were less likely to be interrupted
by actor attribute factors such as registration status or location proximity.

Recovery stage (t3)
There was a slight increase in the correlation measure between communication
and collaboration network during the long term recovery stage (see tables 6.3.1 and
6.3.2).
Table 6.3.1. Recovery Stage QAP correlations/P-values (with Registration Attribute)
Communication collaboration Registration Status
Communication 1.000/0.000 0.435/0.000 0.009/0.383
Collaboration 0.435/0.000 1.000/0.000 0.010/0.310
Registration
Status
0.009/0.383 0.010/0.310 1.000/0.000

569
Table 6.3.2. Recovery Stage QAP correlations/P-values (with Geographic Location
Attribute)
Communication Collaboration Location
Communication 1.000/0.000 0.416/0.000 0.006/0.410
Collaboration 0.416/0.000 1.000/0.000 -0.014/0.309
Location 0.006/0.410 -0.014/0.309 1.000/0.000
Again, the p value of the measure was significant at the 0.05 level. This means that over
time, the “bond” being built between communication and collaboration became stronger.
Two actors having either type of connections would be even more likely to be engaged in
the other type of activity. In other words, the strength of “mutual enhancement”
characteristic of the civil society resilient structures was being stabilized over the long
term. The implication of this piece of finding is that the time factor did not present as an
interference that causes the elapse of the close correlation between communication and
collaboration networks. Rather, the strength of the correlation withstood the test of time
and was made even stronger than the emergency response period. This further
demonstrated the willingness to commit on the civil society actors’ side.
The institutional similarity and the location proximity factors remained to be
insignificant in their correlations with the communication and collaboration networks. Up
to this point, we can safely say that for the development of the two types of the networks
before and after the earthquake event, it was less likely for them to be correlated with an

570
actor’s registration status and geographic location. This means that the structural
transformations of the two networks over the three periods of time were less likely to be
related to the actor attributes. On the one hand, this implies that civil society actors are
not likely to perceive institutional similarity and location proximity as determining
factors when they make communication and collaboration decisions towards others. On
the other hand, some other structural factors could be at work in “stimulating” the
transformations of the two networks over time. I will delve into this part of the analysis in
the modeling sections.
In summary, the QAP analysis showed that the communication and collaboration
networks were indeed positively correlated with each other both before and after the
earthquake. Going back to the original purpose in this investigation, which was to
minimize the missing-ness of the original network data, these results allowed me to
further look at the core-periphery structures in order to select the one with the most
inclusive network connections.

571
Core-periphery Analysis
138

The second step of the missing data treatment procedure involved conducting a
core-periphery analysis to the actor-by actor data matrices of the communication network
in UCINET. This analysis identified two sets of actors. One set was composed of actors
(core) with high density of ties among themselves while the other set was the set of actors
(periphery) with low density of ties among themselves. The denser the ties among the
actors meant that the members in the core interact with many more others in the same
category as compared to those in the periphery. Eventually, actors in the core were able
to “coordinate” their actions while those in the periphery were not. Since one of the main
research questions this study intended to answer was how groups and organizations
communicate and collaborate over time after a significant event, the density of ties that
represented both naming and being named by other actors in the networks was an
important factor to consider. Collapsing the original number of actors to only those in the
core category thus allows this study to further focus on those with relatively high level of
coordinated activities.

138
Please refer to Appendix 6.6 (A-C) for samples of UCINET communication network core-periphery output results
for three time periods.

572
First of all, I conducted core-periphery analysis for all three periods of the
communication network. Then, actors that appeared as core members in each one of the
period were combined together into a single group of actors. Observing the core class
membership Table 6.4 for the communication networks over the three periods of time,
there were a total of 70 actors appeared in the core of communication networks
139
. Then,
I conducted a core-periphery analysis for the collaboration network. The core members
for all the three time periods are shown in table 6.5.
Table 6.4. Communication Network-Core Class Membership
Waves Actor Identifier
t1 1 100 104 106 108 110 115 118 119 134 137 14 19 20 24 25 27 37 38
50 51 57 6 61 65 70 88 93 94 95
t2 1 100 107 109 118 119 123 134 135 137 2 24 27 3 32 34 37 4 49 50
51 57 6 61 7 74 93 96
t3 1 100 107 109 111 118 119 12 123 134 135 137 19 24 27 3 32 37 49
51 57 6 61 62 64 7 74 76 85 93 97

Table 6.5. Collaboration Network-Core Class Membership
Waves Actor Identifier
t1 1 100 104 110 119 137 2 25 37 61 65 70 93
t2 1 100 119 135 137 2 24 25 3 37 6 61 93
t3 1 118 119 12 123 134 135 2 24 3 32 34 37 51 61 76 93
Here I would like to make further observations in terms of comparing the core
class membership results for communication and collaboration networks. First of all,
looking at the core membership composition for the two types of structures across time,
we find that the actors that made up of the cores in the communication networks turned

139
See Appendix 6.2B for further explanations.

573
out to be a more expansive set of that composed of the collaboration networks. In this
research, when adopting the SIENA models to determining the rules governing network
dynamics, especially in the incidences of modeling multiplex dynamics of both
communication and collaboration networks, a consistent composition of network actors is
a necessary requirement for proper execution of the testing procedures. In this
circumstance, one needs to decide whom to include in the modeling analysis when
minimizing the effect of missing data of the original 138 actors. Within the context of this
research, I would like to present the following two reasoning-steps to determine the sub-
composition of the actors whose networking activities would be analyzed in the SIENA
models. First, the core-periphery analysis showed a more inclusiveness of core actors for
the communication network. The counting of those actors in the core of communication
activities incorporated those who were in the core of collaboration activities. Second,
from the QAP correlation analysis conducted in the earlier section, the results confirmed
that the inclusion of actors inside the communication core would also enable the study to
look at the collaboration activities. Thus, I decided to use the 70 actors that appeared to
be in the core of the communication network over three time periods to be the main
narrowed-down group of actors to enter further modeling analysis for the study.

574
From the three sets of density matrices following each of the core-class membership
tables (see table 6.6 and 6.7), the following findings can be documented.
Table 6.6. Core-periphery Density Table (Communication Network)
t1 t2 t3
Core Periphery Core Periphery Core Periphery
Core 0.163 0.023 0.386 0.188 0.391 0.198
Periphery 0.003 0.000 0.030 0.005 0.024 0.008

Table 6.7. Core-periphery Density Table (Collaboration Network)
t1 t2 t3
Core Periphery Core Periphery Core Periphery
Core 0.199 0.011 0.288 0.048 0.290 0.058
Periphery 0.006 0.001 0.025 0.005 0.024 0.004
First, the density of communication interactions among the core members
experienced a jump from 0.163 to 0.386 immediately after the earthquake. And it was
followed by a more gradual change when the level of information exchange among the
core members increased to 0.391 throughout the long term recovery stage. A similar
change pattern can be observed for the core development of the collaboration network.
The density of interaction across the core collaboration partners jumped from 0.199
before the earthquake to 0.288 shortly afterwards. The intensity of collaborative
interaction kept increasing to 0.290 during the recovery stage.

575
Composition Change
At stage three, I utilized the composition change function in SIENA to further
reduce the impact of non-responsiveness in the data. At time wave 1, which was the
period before the earthquake, some actors were non-existent. At time wave 2, which
includes emergency response period and up to one year after the earthquake, these actors
came into existence and started joining the network. At time wave 3, they continued to be
part of the network and no one left. Such change of joiners and leavers is a qualitatively
different type of missing data (Huisman and Steglich 2008) and can be specified with a
composition change file being modeled as exogenous events. For the Chinese disaster
recovery context, I implemented a specific decision-making procedure for coding to take
into consideration the different types of actors involved during the post-earthquake
period. Issues still remain in how actors would define their first date of establishment (see
Appendix 5.3) and future qualitative investigations could bring further clarity in the data
collection process for disaster recovery research.
As a result of this non-responsiveness treatment process, the final percentage of
missing data is 15.7%, which was between the relatively safe range between 10% and
20% to reach stable estimations as specified in the RSiena Manual (Ripley et al. 2013).

576
Some Descriptive Measures
After narrowing down the size of the actors to 70, I first provide descriptive
statistics for this dataset with each of the networks considered on its own. Then, I will
briefly discuss the association between the communication and collaboration networks.
Table 6.8 shows that the average degree of both types of networks experienced a rather
dramatic change particularly immediately after the earthquake.
Table 6.8. Three-period Network Measures Comparison (Communication and
collaboration Networks)

Com.t1 Com.t2 Com.t3 Col.t1 Col.t2 Col.t3
Av.degree 3.169 11.593 13.119 1.051 2.797 3.678
s.d. in/out 2.941/5.18
2
6.785/13.489 6.930/13.966 1.608/1.894 3.252/3.427 3.807/4.796
Reciprocity 0.113 0.146 0.171 0.069 0.090 0.142
Clustering 0.168 0.237 0.275 0.074 0.107 0.135
Note: results based on a total of 70 actors. Av. degree: Average degree. s.d. in/out:
standard deviation for in-degree and out-degree measures.

The average degree for information exchange increased from 3.169 to 11.593,
representing increased level of agency activities among civil society actors. The graphs in
figure 6.1 confirmed this point. Note that not only the isolated actors appeared in the
period before the earthquake were drawn into the main connected network immediately
after the disaster, the network also became more compact and dense over time, as could
be demonstrated through increased reciprocity and clustering activities.

577

Figure 6.1. Communication Network Evolution (70 actors)

578
The tendency continued through long term recovery stage with a slight increase of
average degree to 13.119. The collaboration networks also showed increasing trend in
average degree over time but less dramatic than the communication networks. The level
of reciprocity for both networks increased over time. However, neither showed a strong
tendency towards reciprocity with measures well below 0.5. And collaboration networks
showed weaker reciprocity than the communication networks in general. As for
transitivity, communication networks generally showed stronger tendencies for clustering
than collaborative networks over time. From the changes in the standard deviation
measures, we can say that after the earthquake, both the in-degrees and the out-degrees of
communication networks became increasingly variable with the latter higher than the
former throughout the three time stages. This means that on the one hand, the agency
motivations of civil society actors were being activated as prompted by the earthquake.
On the other hand, the response of the activation may be differentiated. For the
collaboration networks, the in-degrees were approximately as variable as the out-degrees
with the latter slightly higher than the former.
The QAP correlations (70 Actors) between the communication and the
collaboration networks across three observational moments showed positive signs and
were relatively high in intensity.

579
Table 6.9. QAP Correlations between Communication and Collaboration Networks
(70 Actors)
t1 t2 t3
Col. Col. Col.
Com. 0.501 0.404 0.426
Note: Com.: communication; Col.: collaboration
At the tie level (for communication and collaboration), this indicated that the two types of
networks were already rather highly correlated before the earthquake. The measure
dropped from 0.501 to 0.404 during the emergency response stage after the earthquake. A
cause might be the mass emergence of the new grassroots groups after the disaster event
and their participation shook the existing social system in such a way that all actors had
to re-orient themselves to get to know one another, especially the emerging ones in the
field. Over time, the system seemed to re-establish itself. But this time, the increased
correlational state would represent a qualitatively different dynamic equilibrium than
before the earthquake. This is because of the new actors participating and staying in the
system. These trends in cross-network associations can be primary indicators in
measuring social change in a post-disaster setting.

580
Testing and Model Specification
In this section, I first state the goal of this empirical study and the related research
questions. Then, I discuss the model testing and specification processes. The first general
aim of the longitudinal network modeling using RSIENA was to understand the dynamics
of two structural types of relationships. More specifically, one was to look at the factors
that contributed to the formation and sustainability of each of the communication and
collaboration networks. Secondly, I investigated the cross-network dependencies at the
actor level between communication and collaboration networks. In other words, I looked
at the co-evolution of the two types of networks.

Basic Model
In this study, I focused on the most basic longitudinal model specification with
the objective function depicted as follows:
1
( , ) ( ),
L
i k ik
k
f x S x 




The symbol
i
represents the ‘ego’ or the focal actor in consideration. The weights
k

are
statistical parameters indicating strength of effect
()
ik
sx
(linear predictor).
( , )
i
fx 

represents the value of the objective function for actor
i
depending on the state
x
of the

581
network, which is a state being perceived from the focal actor’s point of view. Such state
can be in terms of relationships and in actor covariates. The network effects are all
included in the function
()
ki
Sx
.
The objective function thus determines the probabilities of change in the network,
given that an actor has the opportunity to make a change, as in accordance with the
modeling assumptions stated earlier. It can be depicted as “the rules of network behavior”
as actors make their decisions to make or terminate a tie based on their overall evaluation
of how they view the current state of the network and the effects of covariates.
In this study, the dependent variables were the observable communication and
collaboration networks over the three time waves. The network evolutions of these two
types of networks were functions of three general categories of independent variables: 1)
Structural effects (network endogenous effects, ex. reciprocity, transitive triplets, etc.); 2)
Explanatory actor variables (exogenous effects, actor-dependent, ex. actor registration
status); 3) Explanatory dyadic variables (exogenous effects, dyad-dependent, ex. actor
participation in disaster recovery activities). I particularly focused on the social selection
(Steglich, Snijders, and Pearson 2010) process in this basic model. This process
postulates that actors make their choice of ties based on attributes and the network
embeddedness of the actor as well as those others in the network. Another type of

582
investigation in network evolution is called social influence (Steglich, Snijders, and
Pearson 2010), which is a way of seeing how actor behaviors, for example, the changes in
the types of recovery activities over time, would be influenced not only by their own
attributes and network positions, but also on the behavior and attributes of other actors
directly or indirectly tied to the focal actor. The social influence models can be explored
in future research endeavors as more data on the changing behavioral data are available.
This is especially promising when civil society actors develop changing fields of
practices before, during, and long term after the disaster event. Future models related to
disaster recovery can be built based on the co-evolution of social selection and social
influence processes.
The basic model can also be advanced in two particular directions. One is to allow
actors to change their ties at differential frequencies rather than treating them as constant,
depending on the actor attributes or on the positional characteristics. This is called
differential rates of change in relations to the rate function. The other type of variance of
the basic model is to consider the endowment function (Snijders 2010), which operates
only for the termination of ties. This can be examined together with the evaluation
function, which is the component taking into consideration of the creation of ties.
Considering the general trend of increasing tie formation for the two periods after the

583
earthquake event, I temporarily ignored the differential rate of change and the
endowment function in this study in order to focus primarily on the network growth
aspect of the change dynamics. But in research projects beyond this basic model
investigation, especially looking into the longer terms into disaster mitigation, the
consideration of both types of functions should be considered for a better representation
of empirical networks.

Uniplex One-mode Specification
Uniplex analysis looks at the evolution of the communication and collaboration
networks each on its own. The following illustrated the particular questions that the study
intended to answer:
A) Structural effects: basic network effects
a. Out-degree of actor i: Was there a basic tendency for the network to have
ties at all? (Do actors tend to reach out to others at all?)

b. Reciprocity: Was there a tendency for ties to be reciprocated?

B) Structural effects: transitivity and other triadic effects

584

a. Was there a tendency in communication/collaboration networks towards
transitivity? (“friends of my friends are my friends?”)
b. Was there a tendency in communication/collaboration networks towards
hierarchy?
c. Was there a tendency in communication/collaboration networks towards
generalized reciprocity?
d. Was there a tendency for actors in communication/collaboration networks
to position themselves between not directly connected others? (tendency
towards “brokerage”?)
C) Structural effects: degree-related effects
a. Was there a tendency for actors with high in-degrees to attract extra
incoming ties ‘because’ of their high current in-degrees? Do high in-
degrees reinforce themselves?

585
b. Was there a tendency for actors with high out-degrees to attract extra
incoming ties ‘because’ of their high current out-degree?
c. Was there a tendency for actors with high out-degrees to send out extra
outgoing ties ‘because’ of their high current out-degrees?
D) Covariates effects: exogenous effects
a. Did actors’ registration background have an effect on
communication/collaboration, controlling for reciprocity and transitivity?
(Actor independent variables)
i. Ego effect: whether groups and organizations that were registered
tended to nominate more others and hence have a higher out-
degree?
ii. Alter effect: whether actors that were registered tended to be
nominated by more others and hence have higher in-degrees?
iii. Ego-alter interaction effect: whether actors that were registered had
a greater preference for other actors who likewise are also
registered?

586
iv. Interaction effect of registration similarity with reciprocity: was
there a tendency to reciprocation in organizations with similar
registration status?
b. Do the types of activities that the actors engaged in after the earthquake
had an effect on communication/collaboration? (Dyadic independent
variables)
i. Did the activity choices in general have an effect on
communication/collaboration networks?
ii. Were there tendencies for the actors to focus on any specific types
of activities over time?

Cross-dependencies between one-mode networks
E) Did collaborations follow earlier communication among actors in the network?
Vice versa?

Testing
The exact mathematical representation of each of these effects can be found in Appendix
5.4. The null hypothesis of a single element of the parameter vector is zero,

587
0
: 0,
k
H  
which can be tested by the t-statistic
ˆ
. .( )
ˆ
k
k
k
t
se 



in the standard normal distribution (Snijders 2004).

Issues with Data Requirement in SAB Models
Time Heterogeneity
Referring to tables 6.10.1, 6.10.2, 6.11.1, and 6.11.2, both the communication and
the collaboration networks experienced significant amount of change in tie formation
from period 1 to period 2 (t1 to t2). From wave 1 to wave 2, there were 544 connections
being formed or communication contacts being initiated, with the density measures
jumped up from 0.046 to 0.168. At the same time, a total of 124 collaboration ties
emerged, along with an increase of density measures from 0.015 to 0.041.

588
Table 6.10.1. Communication Network Density Indicators
140

observation time 1 2 3
Density 0.046 0.168 0.190
Average Degree 3.169 11.593 13.119
Number of Ties 187 684 774
Missing Fraction 0.157 0.157 0.157
Average Degree 9.294
Note: 70 actors
141

Table 6.10.2. Communication Network Tie Changes between Subsequent
Observational Periods
Periods 0 => 0 0 => 1 1 => 0 1 => 1 Distance Jaccard Missing
1 ==>
2
3340 544 47 140 591 0.192 759
(16%)
2 ==>
3
3095 292 202 482 494 0.494 759
(16%)
Note: 70 actors

140
Results from R version 2.15.1
141
For dyad counts measures, please see Appendix 6.5.

589
Table 6.11.1. Collaboration Network Density Indicators
observation time 1 2 3
Density 0.015 0.041 0.053
Average Degree 1.051 2.797 3.678
Number of Ties 62 165 217
Missing Fraction 0.157 0.157 0.157
Average Degree 2.508
Note: 70 actors
142

Table 6.11.2. Collaboration Network Tie Changes between Subsequent
Observational Periods
Periods 0 => 0 0 => 1 1 => 0 1 => 1 Distance Jaccard Missing
1 ==>
2
3885 124 21 41 145 0.220 759
(16%)
2 ==>
3
3808 98 46 119 144 0.452 759
(16%)
Note: 70 actors
The average degree in the evolution of communication network increased from
3.169 to 11.593. This means that on average, approximately 8 more ties in the
communication network were being formed when considering the earthquake as the
critical time point of dividing wave 1 and wave 2. Compared to the period from t2 to t3,
the change in average degree was rather smooth from 11.593 to 13.119. The changes

142
For dyad count measures for collaboration network, see Appendix 6.5.

590
from t1 to t2 and from t2 to t3 for the collaboration network followed similar patterns.
There were 124 ties being formed from wave 1 to wave 2 as compared to 98 ties from
wave 2 to wave 3.
From these measures, we can conclude that both the communication and
collaboration networks were in a period of growth with the second network wave consists
many more ties than the first wave. Such changes in the growth of both networks
provided primary descriptive indication that the catastrophic event such as the Wenchuan
earthquake did trigger a change process in the growth of civil society, especially in terms
of the activities of social groups and NGOs. Over time, the connections formed in the
communication and collaboration networks were also being sustained.
The Jaccard index is normally used when one observes significant changes in the
network ties in order to detect whether the data collection time points are not too far
away. This measure calculates the amount of change between two waves by:

where
11
N
is the number of ties present at both waves,
01
N
is the number of ties newly
created, and
10
N
is the number of ties terminated (Snijders, Bunt and Steglich 2010). If

591
this value is preferably greater than .3, and values lower than .2 would lead to doubts
about the assumption that the change process is gradual, compared to observation
frequency. The Jaccard Index from t1 to t2 for the communication network is .192 and
for collaboration network is .220. However, as Snijders, Bunt and Steglich (2010) pointed
out, if the first wave has a much lower density than the second and the network is indeed
going through a period of growth, one may look at the proportion:

Measures higher than .6 are preferable and between .3 and .6 are still acceptable. By
calculation, the proportions for both communication and collaboration networks are
higher than .6 in this study.
An important issue that arises from this type of changes in tie growth in
longitudinal network data is time heterogeneity. If the model specification from t1 to t3
does not take into account such tie changes over these multiple time periods, the
statistical inferences regarding the rules governing the network evolution will lead to
erroneous conclusions. Previous network simulation research (Lospinoso et al. 2011)
considering time heterogeneity indicates that estimation results will average over the

592
heterogeneity if the models are specified homogeneously. In order to specify the rules for
network dynamics accurately while taking into account the need to assess time
heterogeneity for communication and collaboration networks, my model specification
includes: 1) Estimating model #1 from t1 to t2; 2) Estimating model #2 from t2 to t3; 3)
Estimating the complete model #3 from t1 to t3. The results will be composed of a total
of 6 separate models.

Conditional and Unconditional Model Estimation
There are two Methods of Moments estimation methods in SIENA: conditional
and unconditional. The difference between them is the stopping rule for the simulations
of the network evolution (Ripley et al., 2012). In the unconditional estimation, the
simulation of the network evolution in each time period carries on until the
predetermined time length has elapsed. In the conditional estimation, the simulations for
each period run until a stopping criterion calculated from observed data is reached
(Ripley et al. 2012)
Models estimating period from t1 to t2 and period from t1 to t3 utilized the
unconditional method for the following reason. As part of the research endeavor in this
study, one characteristic of the network data collected is there are actors that were non-

593
existent before the earthquake and being self-organized into informal social groups after
the event, in this case, the event is distinguished by the dividing time point of t2. In the
model specification process, such change is called “network composition change” and a
separate file needs to be created for SIENA to recognize this characteristic. For networks
with such composition change due to actors joining the network at t2, only the
unconditional estimation procedure is available.
Models estimating period from t2 to t3 used the conventional conditional method. During
this period, no actors left the network so the number of actors stayed the same. In general,
the conditional estimation is considered slightly more stable and efficient.

Uniplex Results
143

Communication Network Evolution
Initially, dynamics of period 1 (from wave 1 to wave 2) was estimated separately
from period 2 (from wave 2 to wave 3) for communication and collaboration networks
respectively. Doing so allowed me to highlight the substantial time heterogeneity in
parameters such as out-degrees as a result of the impact of the earthquake event. This can
also distinguish the exact structural effects (within-network effects) that govern the

143
Uniplex results refer to the findings on the evolution of each type of network by themselves.

594
changes from before the disaster to shortly afterwards and for the period into the long
term recovery stage.
Since the Wenchuan earthquake, which happened between wave 1 and wave 2,
was such a dramatic event that induced significant changes (in important parameters of
interest) in the Chinese civil society, a separate analysis enabled me to take a closer look
to understand how rules governing network structural changed comparing immediately
after the crisis event and during the long term recovery period. Practically, knowing
which structural effects were consistent in making statistically significant contributions to
network evolution for these two time periods also allowed me to make better judgment in
specifying comprehensive models for period from t1 to t3. This sequence would also
result in faster convergence in the final model-building.

Properties of Communication Network
Emergency Response (t1-t2)
Due to the existence of both one-mode and two-mode networks in the model
estimations, both the in-degree popularity and out-degree popularity parameter estimates
were transformed by a square root for decreased variability across the two types of
networks (Snijders et al., 2012). It is also worth noting that the constant covariate such as

595
actor registration status is dichotomous with values of 1 and 2. The value 1 means
registered and value 2 means unregistered. This is the case when interpreting results for
each network separately.
For emergency response period, the second and third columns in table 6.12
illustrate the findings.
Table 6.12. Rules Governing Emergency Response and Recovery Periods
Communication Network Evolution
Effect t1-t2
(Emergency
Response)
t2-t3
(Recovery)

Within-network par. (s.e.) par. (s.e.)
Out-degree -3.2974 (.1848) -4.1786 (.6945)
Reciprocity
.4064
(.2316) .9131 (.1729)
Transitive triplets .1127 (.0391) .0782 (.0354)
Three-cycles -.1190 (.0746) .0127 (.0451)
Transitive ties -.8942 (.2094) 1.0070 (.3500)
Balance -.0896 (.0111) - .0092 (.0103)
In-degree popularity (Sqrt) .9301 (.0724) .4276 (.0626)
Out-degree popularity (Sqrt) -.1655 (.0760) - .2902 (.1184)
Out-degree activity (Sqrt) - - .1463 (.1310)
Int. Registration similarity x
reciprocity
-1.3940 (.4035)
- .4819
(.2692)
Activity - .0734 (.1141)
Registration alter - .0648 (.1314)
   
†
 
  
   
 
   
 
 
†

596
Registration ego .4123 (.1225)
Registration ego x registration
alter
.1150 (.2488)
Housing alter .0338 (.1196)
Housing ego - .2772 (.1332)
Housing ego x housing alter .8441 (.3068)
Eld_dis alter .0022 (.1116)
Eld_dis ego
- .2187
(.1114)
Eld_dis ego x Eld_dis alter .1376 (.2154)
Wom_chil alter .0628 (.1461)
Wom_chil ego .7393 (.1536)
Wom_chil ego x wom_chil alter .1786 (.2206)
Env alter - .1015 (.1089)
Env ego - .4972 (.1451)
Env ego x env alter .3074 (.2344)
Psy alter - .1589 (.1291)
Psy ego .3977 (.1191)
Psy ego x psy alter .3261 (.2258)
Liv alter .1657 (.1193)
Liv ego - .0224 (.1362)
Liv ego x liv alter .3210 (.2171)
†
P<0.10,

P< 0.05,
 
P<0.01 (two-sided).

 

 
†
 
 
 

597
The most basic effect was the out-degree of actor
i
and it was strongly
significant. This represented a basic tendency for the network to have ties at all. This
might be driven by the altruistic intentions of the social actors to get involved in the
emergency response period right after the earthquake. However, according to Snijders,
van de Bunt, and Steglich (2010), most of the networks are sparse with densities below
0.5, which means that the costs of initiating a tie to an arbitrary actor “with no
characteristics or tie pattern making him/her especially attractive to
i
, the cost will
usually outweigh the benefits (10). This will generally yield a negative parameter in out-
degree effect.
Regardless, the statistically significant parameter estimate of out-degree effect
indeed demonstrated that the communication network during this period did have a
tendency to build ties in order to counteract with the costs of an arbitrary tie
144
. Actors
started out by contacting others that were also involved in the emergency response
process to seek out further information immediately after the earthquake. Communication
network at this period of time showed tendencies toward reciprocity. Reciprocated
responses were being valued positively and this revealed an initial evidence of the
emergence of a Chinese civil society triggered by the Wenchuan earthquake.

144
Please refer to Appendix 6.4 for further mathematical representations of the effects.

598
Aside from reciprocity, the emergency response communication network also
showed tendencies toward various types of clustering, or network closure effects. One
was the transitive triplet effect representing closures of the type
  ; i j h i h   
as
well as
  ; i h j i j   
. This effect postulated that more intermediaries will add
proportionally to the tendency to transitive closure. From the results table, we can see that
two measures for transitivity, transitive triplet effect and transitive ties effect, were
positively significant. This means that the “friends of friends tend to be friends”. The
communication relationship triangle had a tendency to be closed. The transitive triplet
effect is often discussed together with the three cycle effect, which can be interpreted as
generalized reciprocity as the opposite of hierarchy. In the output table (see table 6.12),
the findings showed that the emergency response communication network did not have a
tendency to develop three-cycles that would work against the hierarchical ordering
exemplified by transitive triplet effect. In the context of emergency response after the
earthquake, this can be interpreted as a sense of “eagerness for sociability” for civil
society actors to build up ties with others. Actors on the initiation end of a triadic
relationship also tended to close-up the transitive circle on their own action, rather than
waiting for the actor on the other end to reach out.

599
Another network closure effect being tested was the balance effect, which
measured the tendency for the network to have and create ties to other actors who make
the same choices as the ego (actor i). This effect was significant but negative. This meant
that there were few balanced triadic closures that would reveal a network tendency to
have and create ties to other actors who made the same choices as ego. In other words,
the role formation of Chinese civil society actors during the response period might be at a
stage of diversification when considering the ways how actors reach out to others. When
looking at the number of outgoing choices and non-choices that actor had in common,
few actors can replace the role of the others, or having structural equivalence with respect
to out-ties (Snijders et al. 2010).
In-degree popularity effect was positive, indicating actors with higher in-degree
were more attractive for other actors to send further incoming ties. In other words, high
in-degrees reinforce themselves and there was also a tendency for differentiated actor in-
degrees in the communication network. This might be due to the fact that some actors
emerged to become more trusted than others immediately after the disaster. The out-
degree popularity effect was negative, indicating that those actors who nominated many
others in the communication network were actually less popular when considered by
others as potential information exchange partners.

600
In this model, I also included an interaction effect of registration status similarity
with reciprocity. The result found negative interaction between actors’ reciprocity and
them having the same registration status. At the first glance, this result was counter-
intuitive because we would expect actors would be more likely to reciprocate those who
have the similar traits with themselves. For this study, the negative parameter can be
interpreted as follows. Tie reciprocation might be easier to form if two actors had the
similar registration status and this won’t bring much benefit or “satisfaction” to the
actors. But for a tie to be reciprocated for two actors with different registration status, the
other actor would develop more appreciation and also a sense of accomplishment in
terms of reaching out for those that were of different traits.
Overall, the civil society communication network immediately after the
Wenchuan earthquake showed the following evolutionary characteristics: there was a
relatively strong tendency for reciprocity as demonstrated by the significant reciprocity
parameter; there was a strong evidence for transitive closure, as seen in the significant
effects of transitive triplets and transitive ties. The positive parameter for the former and
the negative sign for the latter showed that there was a strong tendency for civil society
actors to develop their proactivity immediately after the crisis event; the network closures
also showed few structural equivalence with respect to outgoing ties, as demonstrated by

601
negative balance effect; The tendencies toward closure were not completely egalitarian
and showed some evidence for local hierarchical formation, as seen in significant positive
in-degree popularity effect and negative out-degree popularity effect; the emergency
period did not show evidence of tendency to reciprocation being segregated by
registration status.

Disaster Recovery (t2-t3)
The findings for this period of time are shown in the third and fourth columns of
table 6.12. When compared to the communication network in the emergency response
stage, the recovery period showed the following properties. First of all, the network
continued to have strong tendencies for reciprocation, indicating that civil society actors’
initiation of communication and information exchange ties to others during the
emergency response stage were not merely impulsive, they showed signs of commitment
and sustainability into the longer term. Secondly, the network remained to have strong
tendencies for transitive closure, demonstrated by significant transitive triplets as well as
transitive ties effects. However, the balance effect turned out to be insignificant during
the recovery period. This showed that the communication network might start to stabilize
over time. Thirdly, the tendency for local hierarchization continued to be evidenced by

602
positive in-degree popularity effect with negative out-degree popularity effect. This
meant that over the long term, active civil society actors with higher out-degrees were
still less likely to be chosen as communication partners, demonstrating some kind of
status effect in the network dynamic. Fourthly, the stronger tendency to reciprocation in
cross-registration status than same-registered status relationships was being sustained
from emergency response into the recovery stage.
In the earthquake recovery case, since the non-registered actors tended to be those
that were most grass-root and domestically originated, the stronger tendency to
reciprocation in actors with differentiated registration status than same-status
communication connections can be interpreted as a tendency for civil society actors to
develop a general nurturing environment to help the development of non-registered social
groups. Despite the fact that China is known for its harsh institutional environment for
newly formed social groups to establish and sustain over time, particularly after the
government established three-year time limit for earthquake recovery, civil society actors
showed consistent tendency for cross registration status ties. The network is less likely to
be segregated by registration status and there could be a sense of reaching towards a
common goal of disaster recovery.

603
For the recovery stage, a constant dyadic covariate was added to the model. In the
original survey, the respondents were asked to name the type(s) of disaster recovery
activity their groups or organizations had been engaged in after the earthquake into the
recovery period. I then tested the significance of activity as an aggregate constant dyadic
covariate to examine its overall effect on communication network evolution. I also
singled out each one of the activity types as constant covariates in order to see their
respective effects on the dynamics of the communication and collaboration networks.
These constant covariates are dichotomous with value of 0 meaning not participating in
one type of activity and a value of 1 meaning participated in one.
The results showed that activity as an aggregate independent variable was not
significant. However, when testing for the significance of a variety of types of activities
during the long term recovery period, the following results can be concluded. Civil
society actors who participated in the activities of housing recovery, caring for the elders
and disabled population, and environmental protection tended to communicate less with
others in the network. On the other hand, those who participated in the social work areas
of activities such as caring for women and children and psychological counseling tended
to engage in communication and information sharing long term after the earthquake.
Furthermore, actors who engaged in housing recovery activities also tended to have

604
greater communication preference for other actors who also participated in the field of
housing. This homophily effect only existed for the area of housing and not with any
other types of activities. Lastly, note that the registration ego effect turned out to be
significant during this period, indicating homophily with respect to registration was
strong. Since higher values of registration status represented non-registered actors, the
result meant that during disaster recovery stage, informal social groups tended to engage
in building and initiating more communication and information sharing relationships.
In general, the main differences between the emergency response and the long
term recovery communication dynamics were that ties were more strongly being
reciprocated, less tendencies for local hierarchy, for recovery as compared to emergency
response period. The communication network evolution immediately after the disaster
was also more strongly dependent on tendencies to reciprocation in cross-registration
status relationships.

605
Collaboration Network Evolution
Properties of Collaboration Networks
Emergency Response (t1-t2)
The findings for the emergency response properties are illustrated in the second
and third columns of table 6.13 below.
Table 6.13. Rules Governing Emergency Response and Recovery Periods
Collaboration Network Evolution
Effect t1-t2
(Emergency
Response)
t2-t3
(Recovery)

Within-network par. (s.e.) par. (s.e.)
Out-degree - 3.8537 (.4141) -3.7640 (.4236)
Reciprocity 2.0212 (.3194) 1.6925 (.4885)
Transitive triplets .0223 (.1356) .5256 (.1630)
Three-cycles .0776 (.3556) - .2431 (.3105)
Transitive ties - - - -
Balance - - - .2360 (.0926)
In-degree popularity (Sqrt) .8369 (.1174) .6352 (.1165)
Out-degree popularity (Sqrt) - - - .7571 (.3011)
Out-degree activity (Sqrt) .3505 (.1200) - -
Int. Registration similarity x
reciprocity
- 1.3953 (.6009) - .7686 (.6725)

Registration alter - .5694 (.4224)
   
   
 

   
 
 


606
Registration ego - .2072 (.4999)
Registration ego x registration
alter
.1603 (.5943)
Activity -. 3967 (.3250)
Housing alter -. 4900 (.3257)
Housing ego - 1.8031 (.8568)
Housing ego x housing alter 2.3617 (.8604)
Eld_dis alter .2598 (.3179)
Eld_dis ego - .4378 (.3010)
Eld_dis ego x Eld_dis alter .5907 (.5950)
Wom_chil alter - .0693 (.3207)
Wom_chil ego .2287 (.3957)
Wom_chil ego x wom_chil alter - .0352 (.5166)
Env alter - .1723 (.2782)
Env ego - .3602 (.4528)
Env ego x env alter .6036 (.5503)
Psy alter .8445 (.3447)
Psy ego .4556 (.4736)
Psy ego x psy alter
.8709
(.4998)
Liv alter .4918 (.3198)
Liv ego .5026 (.3636)
Liv ego x liv alter .7183 (.5244)

†
P< 0.10,

P< 0.05,
 
P<0.01 (two-sided).

 

†

607
The collaboration networks during the emergency response period showed much
stronger tendency toward reciprocity than their communication network counterparts.
This is reasonable because project collaboration normally infers greater level of
commitment on both engaged parties than information sharing activities. There was no
evidence showing tendencies for transitive closure. This can be interpreted as actors
tended to be more cautious in building collaboration ties. The three cycle effects were not
significant. For example, there was no tendency for the following two circumstances: 1)
two-path
i j h 
closed by the tie
ih 
, 2) two-path
i j h 
closed by the tie
hi 
. This could be due to the nature of collaborative ties that were inherently stronger
and harder to build than communication ties. This also tells us that even though actors
were connected by a common collaborative partner, the condition was not sufficient to
develop a trend for actors on the two ends (
i
and
h
) to close the relationship loop by
collaborating. This could be due to two reasons. One is that there could be certain
institutional barriers, such as going through the government required establishment
procedures which usually takes time and resources from both engaged actors. The other
reason could be the inherent nature of the catastrophic event. For emergency response
period from t1 to t2, the attention of the actors were very much focusing on providing
immediate support to alleviate the impact of the earthquake. At this period of time, a

608
huge amount of efforts could be diverted towards the group or organization acting alone
while it was still possible that they could always communicate with others.
Similar to communication networks at this period, in-degree popularity was again
positive, indicating high in-degrees reinforced themselves, which would lead to high
dispersion of in-degrees across the network. Also being found significant was the positive
out-degree activity effect reflecting the tendency for actors who nominated many
collaboration partners sending out extra outgoing nominations. There was a high
dispersion in naming collaboration partners among the civil society actors. Essentially,
the emergency response period showed that the collaboration networks had tendencies to
differentiated in-degrees and out-degrees of the civil society actors. This meant that the
earthquake event not only triggered actors to search for communication partners for
information, but also mobilized their actions in seeking the more commitment-oriented
project collaboration relationships.
There was also a tendency towards reciprocation homophily with respect to
registration status. Similar to the communication networks at this period of time,
extending collaborative relationships with actors of different registration status was
perceived to be more remarkable and would be more appreciated.

609
Disaster Recovery (t2-t3)
The picture was quite different when it came to the long term recovery period
(results shown in the second and third columns of table 6.13). First of all, the tendency
towards reciprocity was sustained, but less strongly as compared to emergency response
period. Secondly, the collaboration networks showed evidence for transitive closure, as
seen in the significant effects of transitive triplets. This meant that actors became
proactive in initiating collaboration ties toward others. Such could be a sign indicating
civil society actors started to think about different organizational survival strategies to
become sustainable over time. They could start getting invested in their particular area of
expertise. At the same time, they were getting more familiarized with others who were
also involved throughout the emergency response period. Actors started to look each
other with a long term perspective as more were willing to focus on collaboration in a
particular project. However, a positive transitive triplet effect here also means that there
was a strong tendency towards hierarchical order in these transitive collaboration ties.
Certain actors would be more central and being perceived as more proactivity than others
in initiating a closure on triadic collaboration relationships. With a non-significant three
cycle effect, there was no tendency to ameliorate such an ordering.

610
Thirdly, the balance effect was negative, indicating a tendency for differentiating
roles among civil society actors with few that were structural equivalent in terms of
outgoing ties. In other words, actors were less likely to have or create collaboration ties to
other actors who make the same choices as themselves. This was an encouraging result in
that collaborative partners tended to diversify the tie choices rather than targeting the
similar set of outgoing choices and non-choices. Compared to the communication
counterpart, this effect came in the long term recovery stage in the collaboration network
rather than during the emergency response stage.
Fourthly, the in-degree popularity effect was positively significant, indicating the
tendency to differentiated in-degrees being sustained from emergency response to
recovery stage. Just like the communication network, high in-degrees reinforce
themselves. Actors that were being reached for collaboration projects also tended to
attract extra incoming ties. A negative effect on the out-degree-related popularity means
that those who nominated many collaboration partners were less popular when
considered by others as potential collaborators. Together, the network evolution started to
show some evidence for local heriarchization with some actors being more “well-
recognized” or “well-known” than others.

611
For the activity-related constant dyadic covariate effects, actors who participated
in housing reconstruction turned out to be less initiative for forming communication and
collaboration ties during the recovery stage. On the contrary, those who engaged in
providing psychological counseling services tended to create more collaborative
relationships. For those who participated in these two types of activities, actors also
tended to have greater preference to develop collaboration ties with others who engaged
in the same type of activities as themselves, thus demonstrating homophily with respect
to activity type (stronger in housing activity).
Over the long term, the collaboration networks developed tendencies to have
network closures, which demonstrated a process of institutional development with actors’
willingness to sustain their practice in the field beyond the emergency response. The
perceived “popularity” and “social status” remained to be an important factor
contributing to the tendency for differentiated in-degrees in the collaboration network.
The tendency to appreciate and value cross-registration status ties towards reciprocation
started out to be strong immediately after the earthquake, but no longer significant during
the recovery stage. The activity constant dyadic covariate remained insignificant for both
communication and collaboration networks in the recovery stage. However, there seemed
to be increased activation and mobilization among non-registered civil society actors to

612
seek out others building psychological counseling collaborative ties particularly in the
long term.
Comparing between the emergency response and the recovery periods, for a short
period of time, the earthquake did trigger a propensity for actors who already reached out
to many others to keep sending out more project collaboration initiatives. This “energy”
receded during the long term recovery period. One possible explanation for this kind of
popularity and activity effects related to out-degree could be that after the initial surge in
reaching out to others for collaborative projects, actors gradually developed stable
partnering relationships with certain other actors that either possessed similar traits with
themselves or the existence of connections to them were strategically significant for the
survival of these actors. By “strategic significance”, I mean over time, it was possible that
actors realized that was necessary for them to get connected to particular actors in the
network. As we have seen the results for the clique formation in the collaboration
network from the last Chapter, the state actor and the market actor were regarded as the
top targets from the perspective of civil society actors. This was also being manifested as
a way to gain access to the target population in the earthquake impacted area, to sustain
their legitimacy over time, or a channel to show their commitment to the field of social
development of the disaster impacted communities. Here, I particularly illustrate the

613
example of actor NGOLF in terms of its efforts in seeking legitimacy so as to facilitate its
long term functioning commitment in its own field of specialization.

The Case of Actor NGOLF
The uniqueness of the actor’s institutional formality lies in it having successfully
gaining a special type of legal registration status without having a sponsor unit from
inside the government. Being a grassroots group obtaining such a registration status to
become a formal NGO marked the actor being the first in the city of Chengdu. According
to the participant’s recall, there were no such pilot examples in other cities such
Shenzhen, Guangzhou, and Beijing back at the time of the interview in 2011. As
participant was communicating with the Ministry of Civil Affairs through the process, it
became clear that although the tasks being performed by the actor are education-related,
the functions were more related to “general social education”, rather than formal
education provided by schools. Therefore, the activities being performed ruled out the
possibility of having Ministry of Education as its sponsor. “On the other hand, neither

614
does the Bureau of Civil Affairs would want to be our sponsor, as a result, we became a
stand-alone formal registered work unit without a sponsor”
145
.
Tracing back the motivations for the actor to pursue such a status in the first place,
the participant recalled that it was not their primary intention in targeting this kind of
unique registration status
146
. What he thought more about was how to receive a legal
status for the group. This is because without having a legal status by being registered, the
actor would face two types of functioning obstacles. First of all, the channels available
for obtaining funding will be significantly reduced. When applying for funding as a
formal legal entity, it not only shows the actor’s “formal status” but also signifies that the
actor is operating as a “work unit” with “its own governing board and supervisors”. The
second factor is related to the actual functioning of the actor. When making connections
to the local schools in Chengdu and suggest assisting them in building a set of youth-
development activities for their students, “it will no longer be us showing up with an
individual request, but in the form of a ‘work unit’”, as described by the participant of
NGOLF. For example, “when contacting the youth palace (with a legal status), it will be
discussions between two work units rather than at the individual level. (The other party

145
For Chinese, refer to NGOLF-01-05 in Appendix 6.6.1
146
For detailed account, refer to Appendix 6.6.CaseNGOLF.3.

615
will perceive us as) an entity that is nationally accredited with a registration certificate.
Compared to showing up without such formally established institutionalized status, there
will be a sense of basic trust at the beginning. It might work out if there were pre-existing
personal acquaintance connections, but if no one knows who we are, having a certificate
is critical”. (NGOLF-01)
For example, if we make contacts with places like Youth Palace, (a registered
status would mean that) we will be discussing matters with them as a formal work
unit, rather than as representing individuals. This entity is formally approved by
the government, and we also have certification. Comparatively (as when you
don’t have such status), (others) will have some initial basic trust for you. It
wouldn’t matter much if they know you in person beforehand, but if not, (having
such a formal status) would be better. (NGOLF-01-06
147
)
Along with the discussion on the topic of the importance of institutional
informality, the participant went further to explain one other alternative kind of
registration status that was available for it to choose from back at the time. As commonly
understood among civil society actors at the time of the interview, there are two general
sub-categories in distinguishing an already obtained registration status. One represents
those registered with the Ministry of Civil Affairs as nonprofits, also called ＂民政注册＂
in Chinese. The other one represents those actors registered as a private enterprise, named

147
For Chinese, refer to Appendix 6-6-2.

616
as ＂工商注册＂．As an alternative “route” to gaining a formal legal status, NGOLF
could have applied for a status representing itself with a business identity even though it
can still perform tasks as a nonprofit entity. For the participant, however, this was not
such an appealing alternative due to the following two reasons. On the one hand, once
formally registered as a business entity, any kind of outside donation will still be subject
to paying taxes, thus being treated as a for-profit organization. On the other hand, it
makes it difficult to gain the initial trust from the people that the programs are intended to
serve. The participant explained this point further:
Many were registered as private enterprises, and when signing agreements with
others, they will always be seen as a business…including the stamps they use
would represent them as companies. If the other party doesn’t know you well,
they would think that you were probably a fake or something, or maybe wanted to
take advantage of the warm-heartedness of others to run a business. We’ve
experienced all of these kinds of incidences. But over the long term, they would
get a chance to see that you are truly into it to do real works, then, things will be
fine. The thing is that there will always be people who don’t know you well, and
if you show them that you are a formal social entity, not a private firm, many of
their worries will be cancelled out immediately. (NGOLF-01-07
148
)
This account illustrates the main drawback of registering a civil society actor as a private
for-profit enterprise. For actor NGOLF, whose communication platform-building
activities centered on reaching out to potential social and state actors that might not know

148
For Chinese, refer to Appendix 6.6.3.

617
well about the works of NGOLF beforehand, not gaining the initial sense of trust from its
potential partners would greatly hinder the initial “facilitating” role that the actor
intended to play in the field of youth education and development. So far, from the
accounts of different key participants among the civil society actors thus examined, each
of the two kinds of the most commonly named registration status has its own drawbacks
that were perceived to hinder the independent growth of the civil society actor. Recall
that from the examination of actor SG4, although registering with the Ministry of Civil
Affairs would help the actor gain a legal institutional status, there was still reluctance in
gaining such official recognition mainly because the existence of a government sponsor
would be considered submitting to government oversight and control. The desire to make
its own independent decisions free from government attachment is a synergy at work for
such grassroots groups. For the case of the actor in this section, NGOLF, another concern
was raised regarding the alternative way of gaining legal status, which is by registering as
a business entity. Again, it is the independent functioning of the tasks intended to be
performed by the civil society actor that would be impeded upon receiving this kind of
formal status. It is amidst of these perceived institutional hurdles that a third type of
alternative “route” emerged through the registration experiences of actor NGOLF. Figure
6.2 below shows a summary of the different ways of gaining institutional formality and

618
the box marked in red signifies the alternative identity gained by this particular actor
during the disaster recovery stage after the earthquake event.

Figure 6.2. Civil Society Domain Institutional Formality Structures (Post-
earthquake China)

Registration with Ministry
of Civil Affairs
民政注册
Civil Society
Groups and
Organizations
Registered
NPO
Private Enterprise
Registration
工商注册
GONGO
(Government-organized
NGOs)
Post-
earthquake
Dynamic

619
General Model for the Role of Civil Society
149

Since the Chinese disaster recovery dataset contains substantial time
heterogeneity in some critical parameters of interest as noted earlier, I first estimated
period 1-> 2 separately from period 2 -> 3 so as to mitigate the time heterogeneity. By
doing so in the previous two sections, I essentially treated all of the effect parameters as
dummy variables. This way of dealing with a dataset containing dramatic changes in
actor behaviors due to a significant event happening in between time waves will be useful
in developing separate models for emergency response and disaster recovery, thus
characterizing the differences in the specific features shown in these two periods. I will
now explore a more comprehensive approach in understanding of the rules governing the
dynamic processes by taking into account the changes of all three time waves.
With the intention of developing a general model to understand the role of
Chinese civil society in both emergency response and disaster recovery, I developed a
preliminary model for each of the communication and collaboration networks while
combining the three time waves altogether. Considering the time heterogeneity of
parameters of interest, I incorporated a network data analysis procedure called time test

149
The revised evaluation function when taking into account of time heterogeneity:
( ) ( ) ( )
m
k k k
k
f A s A  

, time dummy terms
m
k

are being estimated through forward-selection procedure.
(Lospinoso 2010)

620
throughout my model-building process. It is an iterative approach to utilizing a strategy
for forward model selection
150
(Lospinoso, 2010; Lospinoso et al., 2011) implemented in
stochastic actor oriented models. Incorporating and testing for time heterogeneity is
especially important for research in disaster response and recovery settings because there
are often disruptions in actor behaviors and drawing inferences on this type of
heterogeneity will help with understanding the periods of disrupted behavior such as in
“cooperation networks for natural disasters” (Lospinoso, et al. 2010, 149).
I generally followed the iterative decision-making procedure for conducting the
time test implemented by Lospinoso, Schweinberger, Snijders, and Ripley (2010) in their
simulation study. I estimated my updated models with time dummies interacted with
those effects with score type tests yielding p values of 0.05 or less for each of the
communication and collaboration networks. Table 6.14 below shows the results. The
term “Dummy2” indicates the corresponding values for a dummy term interacted with the
preceding effect. Because the significant changes in the descriptive measures of
parameters happened at time wave 2 (t2), the dummy terms were being included for this
particular period to take into account of the heterogeneity issue.

150
The test procedures are implemented by the Siena Time-Test function explained in RSiena Manual (Ripley et al.
2012, 107). Specific forward-selection procedure for time-test followed the implementation tutorial in RSiena by
Lospinoso (2010).

621
Table 6.14. General Rules Governing Communication and Collaboration Network
Evolution
Effect communication collaboration
Objective function par. (s.e.) par. (s.e.)
Out-degree
-3.0270
 

(.1785)
-3.1413
 

(.1898)
Reciprocity
.6821
 

(.1265)
1.9836
 

(.2784)
Transitive triplets
.1045
 

(.0176)
.4435
 

(.1252)
Three-cycles -.0482 (.0328) -.2322 (.1992)
Transitive ties -.1037 (.1652) - -
Balance
-.0506
 

(.0073)
-.1413
 

(.0484)
In-degree popularity (Sqrt)
.6405
 

(.0423)
.7213
 

(.0732)
Out-degree popularity (Sqrt)
-.2439
 

(.0494)
-.5907
 

(.2071)
Out-degree activity (Sqrt) - - - -
Registration alter
-.2545
 

(.0914) -.1728 (.1682)
Registration ego
1.0272
 

(.1632)
1.0409
 

(.3058)
Registration ego x registration
alter
.2643 (.1987) .3560 (.3403)
Registration similarity x
reciprocity
-.8706
 

(.2267)
-1.1860


(.5052)

622
Dummy2: ego (out-degree)
-1.1919
 

(.3928)
-.8425
 

(.2957)
Dummy2: ego x reciprocity - - -.6133 (.5496)
Dummy2: ego x transTrip -.0088 (.0327) - -
Dummy2: ego x transTies
1.7816
 

(.3452) - -
Dummy2: ego x balance
.0753
 

(.0133) -.1174 (.0879)
Dummy2: ego x inDegree
Pop(Sqrt)
-.4234
 

(.0848) - -
Dummy2: ego x outDegree
Pop(Sqrt)
- - -.4215 (.2914)
Rate function
Rate period 1 26.9117 (2.0641) 7.5469 (1.0093)
Rate period 2 16.7529 (1.0638) 4.3298 ( .4698)

P<0.10, P< 0.05, P<0.01 (two-sided)

Communication Network Evolution
The score tests confirmed the heterogeneity in the following structural effects:
out-degree, reciprocity, transitive triplets, transitive ties, balance, in-degree popularity,
out-degree popularity. Overall, it was interesting to see the dummies completely negate
†
  

623
two of the effects in communication network evolution during wave 2. One was the
balance effect. It appeared that when modeling the three time stages together, there was a
tendency for actors to develop structural equivalence with respect to communication
outgoing ties. This means that civil society actors actually tended to communicate and
exchange information with those other actors who had a similar set of outreach ties as
themselves. When looking at the overall trend of communication network evolution,
there was indeed a propensity for role stabilization in terms of how actors reached out to
others. The other finding was regarding the in-degree popularity effect. With the dummy
interaction term at wave 2, it appeared those who were already being reached out a lot for
communication purposes were less popular when considered by others as potential
information exchange partners. This means that over time, civil society actors no longer
tended to attract to the already more “well-known” or “popular” ones for information
exchange. The communication network was less likely to be differentiated in in-degree.
The transitive triplet time dummy interaction effect was not statistically
significant anymore, indicating that after considering time heterogeneity at time 2, the
communication network did not have a tendency for transitive triplet type of network
closure. However, the significant dummy interaction transitive ties effect still showed
evidence for network closure in the long term evolution of information exchange

624
network. Civil society actors maintained its ability to sustain their proactivity in building
communication ties.
The communication network in general still showed evidence of strong
reciprocity among civil society actors. Out-degree popularity is negative, indicating that
those who initiated many communication ties were less popular when considered by
others as potential information exchange partners. Together with the negative effect of
the in-degree popularity dummy interaction term, this means that the communication
network neither had a tendency for high dispersion in actor in-degrees nor a tendency for
higher correlation between in-degrees and out-degrees.
Three of the covariate effects turned out to be significant for the general model
estimation. First, there was a tendency towards homophily especially with respect to
reciprocity. The negative effect of the interaction of registration status with reciprocity
indicated that cross- registration status communication ties were indeed being valued and
appreciated among civil society actors. There was a strong propensity to reciprocation
when interaction happened between one non-registered actor and a registered actor. The
network was less likely to be segregated over time in this respect. The registration ego
effect was positive, demonstrating that non-registered actors tended to communicate
more and thus being more active than their registered counterparts. However, with a

625
negative registration alter effect, these non-registered social groups tended to be
nominated by less others and hence had lower in-degrees. These two effects together
showed that the informal grassroots actors were indeed in their emergence stage after the
earthquake. On the one hand, they were not well-recognized by others in the
communication network. Indeed, if a social group or organization was just formed after
the earthquake and in the process of establishing itself in the field, it would indeed take
some time for others to recognize its name and its works. But the encouraging finding
was that these non-registered grassroots social groups exercised their agency freedom
nonetheless. Regardless of such difficulties, they remained their synergy for being active
and kept reaching out towards others.

Collaboration Network Evolution
In the general model estimation of the collaboration networks (see table 6.14),
four dummy interaction effects were detected through the time test procedure and
included in the result table: out-degree, reciprocity, balance, and out-degree popularity.
It appeared that when taking into consideration of time heterogeneity, none of these
effects that previously were prominent features of actor behavior turned out to be
statistically significant. This means that the collaboration networks did not have general

626
tendencies for reciprocation, structural equivalence with respect to nominating
collaboration partners. Neither was there a tendency for higher correlation between in-
degrees and out-degrees, indicating that it was less likely that actors with high level of
nomination for collaboration partners to attract extra nominations.
In terms of other structural effects without dummy interactions, the results
showed similar patterns to communication network evolution. Both types of networks
had strong tendencies for transitive triplets and in-degree popularity. For the former, the
effect was positive, indicating evidence for network closure due to the sustaining of
proactive synergy in actors building collaboration ties over time. For the latter in-degree
popularity, the effect was positive, showing that both networks had tendencies for
differentiated in-degrees. Actors attracting higher incoming nominations for
communication and collaboration ties tended to be even more attractive for others who
perceived them to be “prominent” and thus reached out to them even more.
For the overall evolution patterns of communication as well as collaboration,
there was a tendency towards homophily especially with respect to cross-registration
status reciprocity. Neither of the networks tended to be segregated in this respect. It
showed that there was a general “welcoming” environment particularly for the
nonregistered grassroots social groups emerged after the earthquake. These groups also

627
had a strong tendency to reach out to others for information and establishing
collaboration ties. Non-registered civil society actors were therefore characterized by
high level of activity and thus agency long term after the earthquake. The collaboration
network evolution, however, did not show signs of difficulty being recognized by others
among non-registered actors like in the communication network dynamics. This means
that when it comes to commitment and institutional development, formality status of civil
society actors did not enter the decision-making process. In general, the collaboration
network evolution showed less tendency for status effects contributing to less
hierarchization as compared to communication network. But there were signs of local
hierarchy in collaboration networks with respect to transitive triplets effects.
The parameters for the rate functions were included at the bottom of the table 6.14. For
communication network evolution, each actor in the network had about 27 opportunities
to change immediately after the earthquake. The opportunities decreased to an average of
17. For collaboration network evolution, each actor had 7 opportunities to make a change
during the emergency response period and it further decreased to 4 over the longer term.
These patterns suggested that for both types of networks, the amount of change peaked in
wave 2, shortly after the earthquake, and slowed down towards the long term recovery.
The communication networks experienced drastically higher changes than the

628
collaboration networks. If a social group or organization has just formed after the
earthquake and in the process of establishing itself in the field, it would indeed take some
time for others to recognize its name and its works.

Institutional Formality and Registration Similarity in Reciprocity
In this section, I use qualitative data to understand why institutional formality in
terms of registration status did not turn out to be a significant factor in governing the
overall evolution of both communication and collaboration networks. I define
“institutional formality” in terms of a state of existence when civil society actors gaining
legal registration status under the current legal system in China. How such institutional
status was being perceived and approached by emerging civil society actors constituted as
the characteristics of pre-conditions for sustainable group/organizational actions during
the long term disaster recovery. The interview accounts illustrated in the following
paragraphs were from informants of those actors that emerged to have high level of in-
degree and out-degree since the emergency response stage after the earthquake.

629
The Case of Social Group Actor #3 (SG3)
Since the group SG3 was formed as a non-registered nonprofit entity after the
earthquake event, I named these not-yet-registered newly emerged entities as “social
groups” in order to make note of their grass-root nature. Those registered nonprofits were
named as “organizations” to emphasize on their formality. These two types can also be
categorized as “formal” and “informal” according to institutional literatures. But such a
formulation of analysis would not capture the bottom-up nature of the groups emerged in
the Chinese context. The name of “social groups” will takes into account of the emerging
nature of civil society being discussed in this study.
To the civil society actors in China, whether an entity is registered or not was perceived
as being closely related to having a legal identification. And one of the major obstacles
when not having such an identity for actor such as SG3 was the capability to plan and
develop its own organizational structure. The young participant further explained these
difficulties as the following:
Although we can still do the kind of things that we need to do as of now, but there
are still some negative effects…the most obvious one is…because we don’t have
the financial resources to support ourselves, and thus not being able to do more of
what we would like to do, so we still have to rely on foundations to provide us
with the needed financial resources. If you’d like to receive the other party’s
support, from their perspective, by us having a legal status would greatly facilitate

630
them managing their finances and program in general. If we don’t have such a
status, it would raise certain concerns on their part. If our financial resources run
out, we won’t be able to continue to function. So this is the one thing that has
given us such a headache. Our biggest hurdle right now is not being able to find a
sponsor, an appropriate one that fits our functions, in order to register. This is a
big problem facing us. The Ministry of Civil Affairs in Sichuan Province also got
in touch with us in 2008 and told us that they were aware of our activities. And
they said that it would be better for us to have a legal status and hope us be
registered soon. That was their position on this matter. Therefore, it would be a
big issue if we don’t find a “mother-in-law. (SG3-01-10
151
)
Clearly, the legal identity issue arose from the group’s non-registration status is
closely tied up to the entity’s future financial stability and this is critical for the
sustainable functioning of the group. For SG3 in particular, one of the major obstacles in
obtaining such a formal identity from the Ministry of Civil Affairs was in finding an
appropriate sponsor inside the government. The term “mother-in-law” used in her
account has been used as a common term for Chinese nonprofits organizations to refer to
their sponsors. However, what is unique about the case of SG3 in its journey seeking an
appropriate “mother-in-law” arose out of its own emergence activities and motivations to
provide information services towards other nonprofit groups and organizations, not vice
versa.
For example, those NGOs that were being approved to formally register were the
ones that it (the government) would want you to go into the local communities

151
For Chinese, refer to Appendix 6.6.04.

631
providing services. For us, on the other hand, we are not an organization that
provides direct services to the disaster-impacted population. We only service
other NGOs. This is why we cannot find a work sponsor. No one would, ministry
of civil affairs wouldn’t, science and technology association wouldn’t, they would
say that we are not a research entity, and neither are we technology-related. For
the Red Cross, since they assist in emergency response efforts, they would say
that we are doing information-related tasks, so no. This is what happened to us so
far. But we are still trying. Chengdu is now a designated as a pilot city that allows
the possibility for nonprofit groups to register without a sponsor unit. But after
our consultation on this matter, we found out that although this is the case, only
those focusing on providing services for disadvantaged populations can be
eligible for applying. Since we do not provide direct services, and neither do we
face directly towards disadvantaged population, therefore, we are not eligible.
Regardless, we have clearly positioned ourselves, we are providing services
directly for NGOs. (SG3-03-06
152
)

Therefore, the essential difficulty in the group’s seeking its own registration status
could be attributed to the type of service that it provided. Different from other service-
oriented nonprofits operating to provide support directly towards people inside
communities or other disadvantaged groups in the society, SG3 provided services to an
entirely different group. The nature of its agency actions and field practices was one that
promoted the development and growth of those other nonprofit groups and organizations.
From the institutional perspective, it was a motivation for the group to promote a
“standing up” process of actors inside the civil society domain. From the account of the
senior participant (SG3-03), the group had also been persistent in pursuing its practices in

152
For Chinese, refer to Appendix 6.6.05.

632
institutional-building services regardless of the registration obstacles. Here, it is essential
to make a case regarding the “institutional climate” surrounding the group from two types
of lenses. One was in terms of the issue of formality of the group. Then, the main
obstacle was in it obtaining a legal identification by being registered from the Ministry of
Civil Affairs due to its unique service partnerships. On the other hand, when such a
“climate” is understood in terms of an issue of structural development that the group
actively chose to embedded itself in, a persistent focus and desire to serve the growth of
those other civil society actors showed preliminary signs of an emerging “enabling
environment” that tended to counter the obstacles arose from the issue of formality. For
the case of actor #3, such enabling environment for the development of Chinese civil
society was manifested in the persistent self-identification process to define its own
functioning role despite of the difficulties in formality-seeking. I thus call it the tension
between formality and structurality (“enabling environment”) exemplified itself through
the experiences of SG3 participants after the earthquake
153
.

153
For further details regarding the institutional obstacles for actor #3, refer to Appendix 6.6.CaseSG3.3.

633
The Case of Actor #49 (NGO49)
From the exploration of the emergence process of the actor #49 depicted in the
earlier chapter, I found that one of the key factors that facilitated the actor’s execution of
field practices was its ability to obtain appropriate financial resources to make them
available for those it provided service towards. Thus, it became important to understand
whether and how the institutional formality exerted its impact on the general emergence
and development of the actor from the perspectives of its key participants.
From the organizer’s (NGO49-01) point of view, the institutional formality in terms of its
registration status had been only a “very small factor” in the actor’s journey obtaining its
resources for functioning. What mattered in the end, are the following:
I think for this field…it is an issue of credibility. In other words, it is what
Chinese called the term “word of mouth”, this is very important. Our team was
first established in 2005, and it’s been almost five to six years now. We have
never created any flier or anything to publicize ourselves. We have not created
our own professional webpage even. It’s always been the case that other partners
have sought after us to establish collaborative relationships. It’s rare for us to
promote and publicize ourselves. And also, we have been quite selective in
accepting projects. It’s not the case that we will for it as long as they have the
financial resources. We have our own standards and criteria. So I think our
credibility is always very important. (NGO49-01-05
154
)

154
For Chinese, refer to Appendix 6.6.06.

634
Note that the factor of “credibility” was perceived as a critical one in sustaining the field
practices of this actor. The organizer’s interpretation of the “credibility” concept,
particularly when practicing in the field of civil society, had a relational origin. What the
traditional Chinese phrase “words of mouth” became essentially a way how information
got spread through networks of inter-organizational relationships, when understood
within the context of this study. From a network perspective, the actor as a whole had an
inherent intention to refrain from reaching out towards others to promote and publicize to
actively establish its own influence. Only in selected circumstances that it will reach out
to initiate a collaborative relationship. In other words, perceptions of how others regard
the actor’s practices from its previous performances and their willingly spreading the
“words” throughout the network played a greater role in how the actor gained its status
over time.

The Case of Actor #4 (SG4)
When it comes to the difficulties faced by the group during its transitioning stage
from performing emergence response tasks to taking actions related to the long term
recovery phase after the earthquake, the “dilemma” for the group to make changes to its

635
institutional structure centered on the issue of obtaining funding. This was a similar issue
faced by actor #3 through its transitions to the long term recovery stage. On the one hand,
the founder and the participant himself both held a view that in order to maintain the
group’s grassroots nature formed by ordinary Chinese citizens, relationship with the
government was at all avoided. According the participant (SG4-01), the founder rejected
the interview invitations from the media so as not to intentionally publicize the action of
the group itself. Essentially, to the founder and the participant, even obtaining a
registration status through a process of having to search for a government-related sponsor,
would be a restraint for the functioning of the group because the registration linkage is
perceived to represent an “NGO inside the government” rather than performing as an
independent social group existing separately from the functioning of the state. On the
other hand, the participant was being well aware of the importance of gaining a formal
institutional status through registration towards the group’s securing outside funding
sources. However, the decision to maintain its informal institutional status as a “social
group” was reached through the following claim: “Rather than being incorporated and
used by the government, we’d prefer the freedom in focusing on the tasks that we are
able to perform and do them well”
155
(SG4-01).

155
Original Chinese script: 与其那样的为政府所用，还不如自由一些做自己团队能做的事并且作好.

636
This dilemma faced by actor #4 revealed a particular kind of decision-making
dynamic regarding the emergence of civil society after the crisis event. This actor
consciously oriented its actions in a rather “distant” networked environment with the state
actor. But it did not hold back in developing the initiative to reach out to others inside the
civil society domain, especially right after the earthquake. From the descriptive results
found in the previous chapters, the actor reached a rather high level of out-degree in
terms of information exchange. From the actor’s ego network analysis results
156
, the
“brokerage” role that the actor engaged most in since its establishment (during
emergency response period) was as a “coordinator” in between other non-registered civil
society actors. Brokerage measurements indeed demonstrated its efforts in acting as a
medium among others brokering communication relations between registered and non-
registered actors.
From the perspective of understanding the meaning and emergence of civil
society in the Chinese context, the case pointed towards the following directions. First,
the initial action towards an emerging civil society was taken in the form of establishing
voluntary groups through self-organization. Second, obtaining institutional formality
through registering with the state sector could be perceived as a form of control and

156
Please see Appendix 6.6D1 to 6.6D3 for actor’s ego-network UCINET output over three time periods.

637
attachment thus restraining the original purpose of the civil society actor in being a “free
agent”. Third, a “standing up” civil society was indeed being understood in relation to the
level of “agency freedom” in terms of activities performed in separation from the state.
Institutional transformation from informal to formal status could create “survival”
dilemmas for civil society actors who held a rather strict view of “freedom” in the
Chinese context. Fourth, the role of a civil society, from the participant’s point of view,
was one that “complement and correct” the state.
As a result of the actor’s position in the issue of institutional formality through
registration, it had been a rather challenging task for the group to maintain its long term
functioning in the field. For example, the group once planned to build a nursing home for
those elders who lost their relatives by using funds donated through various channels.
However, the project eventually came to a halt due to a lack of enough funds and labor
that can be devoted to finishing it up. Since then, the group was divided with group
member holding two types of views on the kinds of tasks that it should perform. One side
preferred the continuation of bigger and grander projects while the other preferred
switching to smaller projects with narrowed-down boundaries. Up until the time of the
interview with participant SG4-01, the group was still in search of an appropriate
development path so as to engage its team members in the longer term. The struggle for

638
the actor to further define its role and identification over the longer term resulted in a
significant drop in its communication outreach activities for the recovery period.

The Case of Actor #24 (NGO24)
First of all, several internal difficulties arose within grassroots groups and NGOs
themselves contributed to the additional burdens on the pathways for these civil society
actors to grow over time. One was the lack of the funding sources and the shortage of it
could further hinder the “professional capacity-building”. The other internal factor was
regarding the types of activities particularly chosen by the voluntary groups after the
earthquake event. To the director of actor #24 (NGO24-01), most of these voluntary
groups rushed into the disaster hit areas and focused much of their attention on
emergency response activities. And less attention was being paid to the long term
recovery aspect of activities being established at that period of time.
Secondly, although there was still a lack of sufficient attention, or support,
coming directly from the government side (at the time of the interview), the director held
a rather positive outlook to the general institutional environment for the future
development of grassroots groups and NGOs. For example, he perceived that the

639
government actually was “slowly starting to realize the importance of NGOs and was
gradually re-directing its attitude towards us”. According to the recall of the director, it
had stated clearly in year 2011 twelfth fifth year plan of the important role of social
development in disaster mitigation and preparedness. As he was describing these
observed progresses in government policy orientations, the director finally stated that “I
believe that the development of existing (institutional) environment will be in more and
more favor of the survival of NGOs”
157
.
LU ：Then, how should one raise the government awareness of the importance of
NGOs?
NGO24-01: “In fact, there is a gradual change in government’s attitudes towards
NGOs and they started to recognize our importance. For example, in the twelfth
five-year plan, it raised the importance of social development and its role in the
process of disaster mitigation and preparedness. In the plan, one of the critical
tasks raised for the government is for it to manage well-functioning collaborations
with NGOs and nurturing the development of ‘big society’. So I believe the
surviving institutional environment for NGOs will keep getting better.” (NGO24-
01-02)
Note that from the perspective of the director of NGO24, the “surviving institutional
environment” of this envisioned “big society” incorporated civil society actors of all
kinds, regardless of the registration status. This can be demonstrated by the actor’s own
agency actions right after the disaster event. It had some of the most intense

157
Detail refer to Appendix 6.6.CaseNGO24.3.

640
communication and collaboration outreach activities toward actors across the entire
network during the emergency response stage.

The Case of Actor 51 # (NGO51)
For actor NGO51, “institutional formality” was more than just having a public
recognizable formal registration status. In essence, it was less of an outward nature in
possessing certain title that was meant to generate certain impressions from others, but
more of an inward-looking lens through which actors could self-examine the level of
devotion to the chosen activities related to civil society. From the program officer’s
perspective, the concept of “social group/organization” was essentially based on how an
actor looked at its own field activities and self-reflected on the nature of its actions.
This will mainly depend on their own judgments. Take *** (actor #8) for example,
they will define themselves still as a volunteer group (team). Their founder used
to be a full-time worker elsewhere. He would participate in the volunteer
activities in his spare time. But his team had done a good job in organizing and
conducting the volunteer works, and this might be related to his own work-
orientation. His full-time job was at the blood center and this would allow him to
meet all kinds of volunteers. He would often organize them together to conduct
different kinds of activities. This is why in his mind, all of their activities belong
to a volunteer act, not as an organization. He has never wanted to transition it into
a formal organization neither. And he would not devote full time into it. One of
our criteria in our decision-making is to see if they are doing it full-time. Only if

641
you are full-time will you be devoting all of your efforts in this work. (NGO51-
01-06
158
)
Note the recognized importance of the participants of social groups/organizations in
making a “full–time” commitment in practicing in the field of civil society. In this case,
the registration status was not a determining factor for NGO51 to make the decision on
whether to support a civil society actor through its “incubation” program. If an actor was
already registered but none of its participants operated at a full time level, based on the
account of the program officer, actor NGO51 still would not accept it as one of its
incubation development partners.
The particular role of actor NGO51 was motivated by its long term vision in
providing an enabling supportive environment for other emerging civil society actors to
develop a specific type of “institutional formality”
159
. This type of institution was
characterized not only by a determination to develop full-time commitment in the field
but also the capacity to sustain independent functioning. Compared to the interpretation
that “institutional formality” simply as a way of representing an outcome of having a
registration status, the action of NGO51showed that the phrase could also symbolize a
process through which the within sector collaborative efforts co-evolving with the

158
For Chinese, refer to Appendix 6.6.07.
159
For detailed interview account, refer to NGO51-01-07 in Appendix 6.6.08.

642
institutionalization of civil society actors, regardless of the types of achieved status such
as being registered entities.

Interpretive Summary of Institutional Formality
In general, informants representing different civil society actors perceived a
strong linkage between obtaining formal legal registration status with seeking financial
means for survival. However, there was a perceived disadvantage in gaining registered
status by having sponsorships through the central and local governments, and it was the
concern for the possibility of the loss of independency in making stand-alone decisions.
Furthermore, the actors valued the quality of the functioning as part of the process of how
they defined their identity as their roles were being institutionalized. This latter factor
was perceived to be detached from the institutional formality reflected in the actors’
registration status. In fact, the actors cared more about providing quality supports for the
survival and growth of each other, thus generating a particularly welcoming enabling
environment for the newly established grassroots emerged after the earthquake. This
further explained the strong tendency for reciprocation between registered and non-

643
registered actors in both the communication and collaboration network general evolution
patterns.

Multiplex Results
Co-evolution of Communication and Collaboration Networks

In the earlier sections, I examined each of the two types of networks in separate
models where mutual dependencies between the networks were ignored. The description
of the dynamics of the networks also served as a point of reference when considering the
co-evolution of the two networks. Here, I tested three cross-network effects in order to
determine the rules governing the dynamics of the focal network (X). When the
dependent network is communication, the effects are illustrated as follows
160
(adapted
from RSiena Manual, Ripley et al., 2013):
If both networks have the same number of columns, then the basic effect is the
entrainment of X (communication) by W (collaboration), i.e., the extent to which the
existence of a tie
W
ij   
promotes the creation or maintenance of a tie
X
ij   
.

160
For dependent networks being collaboration, X represents the collaboration and W represents communication.

644
The reciprocity effect with W (collaboration) on X (communication), representing
the extent to which the existence of a tie
W
ji   
promotes the creation or maintenance
of a tie, in the reverse direction,
X
ij   
.
Mutuality effect with W (collaboration) on X (communication), representing the
extent to which the existence of a mutual tie
W
ij   
promotes the creation or
maintenance of a tie
X
ij   
.

Properties of Emergency Response
The dyad level effects between the communication and collaboration networks
showed different intensity of strength across time periods. Table 6.15 showed the results
for emergency response.
Table 6.15. Co-evolution of Communication and Collaboration during Emergency
Response Period (t1-t2)
Effect Communication Collaboration
Within-network par. (s.e.) par. (s.e.)
Out-degree
(density)
- 2.6858
 

(.1752)
- 3.0841
 

(.4086)
Reciprocity
.4080
(.2164)
1.8484
 

(.5166)
Transitive triplets
.1159
 

(.0381) .2510 (.2407)
†

645
Three-cycles
- .1772
 

(.0686) .1353 (.5831)
Balance
- .0630
 

(.0113) - .0908 (.0595)
In-degree
popularity (Sqrt)
.5502
 

(.0725)
.7532
 

(.2255)
Out-degree
popularity (Sqrt)
- .1326 (.0689) .0097 (.2041)
Between-network:
dyadic

Communication - - .8897 (.6207)
Reciprocity with
communication
- - .4946 (.8179)
From
communication
agreement
- - - .4566 (.2742)
Collaboration .4534 (.3640) - -
Reciprocity with
collaboration
.6866 (.6963) - -
From collaboration
agreement
- .5661
 

(.2111) - -
Rate function
Rate Period 1 27.4873 (2.4715) 7.3177 (1.3183)
P<0.10, P< 0.05, P<0.01 (two-sided)
During this period of time, none of the direct and reciprocal effects were
statistically significant. Only the mutuality expressing the effect from agreement on
collaboration to communication turned out to be strong, but negative. The existence of a
mutual collaboration tie actually made it less likely for the creation or maintenance of a

†
  

646
communication tie. From the first sight, this result was somewhat counter-intuitive
because normally a collaborative relationship between a pair of actors would promote the
less binding communication relationship. But bringing in the context of the disaster
response in China, this could be interpreted as evidences of significant social
reconstruction among civil society actors and the development of a possible condition for
the institutional emergence process to occur. Previous rules and bindings were being
broken due to the earthquake event thus creating opportunities for new kinds of social
settings to emerge. It also appeared that the within-networks effects for the collaboration
network were not mediated by the existence of any communication ties. This could be
understood as the creation and maintenance of emergency response collaboration
networks were actually the result of self-emergence and has a dynamic of its own.

Properties of Disaster Recovery
Compared to the emergency response stage, the dyad-level effects between
communication and collaboration networks showed increasing strength over the long
term disaster recovery period. The “between-network” effects listed in table 6.16 below
illustrate the findings.

647
Table 6.16. Co-evolution of Communication and Collaboration during Recovery
Period (t2-t3)
Effect Communication Collaboration
Within-network par. (s.e.) par. (s.e.)
Out-degree (density)
-1.8860
 

(.1990)
-3.1639
 

(.2772)
Reciprocity
.9052
 

(.2071)
1.2428
 

(.3929)
Transitive triplets
.0966
 

(.0144)
.4612
 

(.1299)
Three-cycles .0659 (.0681) -.3245 (.2796)
Balance
-.0172
 

(.0029)
-.1402
 

(.0544)
In-degree popularity
(Sqrt)
.3521
 

(.0565)
.4333
 

(.1134)
Out-degree popularity
(Sqrt)
-.4548
 

(.1234)
-.5320


(.2429)
Activity -.0121 (.0434) .1383 (.0895)
Between-network:
dyadic

Communication - -
.9713
 

(.3851)
Reciprocal
communication
- - .5913 (.4678)
From communication
agreement
- - -.0518 (.0336)
Collaboration
2.0719
 

(.6736) - -
Reciprocal -.0836 (.4409) - -

648
collaboration
From collaboration
agreement
-.0708 (.0866) - -
Rate function
Rate Period 2 19.7853 (2.0148) 5.2267 (.7013)
P<0.10, P< 0.05, P<0.01 (two-sided)
Direct effect from collaboration in the formation and maintenance of
communication network, as well as the effect from communication for the establishment
and maintenance of collaboration networks were both strong, with estimated parameter
values of 2.0719 and 0.9713 respectively. This means that the existence of a collaboration
tie strongly promoted the creation or maintenance of a communication tie, and vice versa.
The effect was even stronger for collaboration on mediating the evolution of the
communication relationships. Therefore, after the initial “turbulence” of the social
system, the new structures emerged short-term after the disaster event was clearly being
maintained through the cross-mediation of collaboration and communication networks.
The binding force was particularly strong for the long term recovery period. The
structural effects of the uniplex networks could be interpreted as the by-products of the
multiplexity of communication with collaboration.

†
  

649
General Model of Network Co-evolution
When examining the three time waves together to develop a general model, the
effects of co-evolution between collaboration and communication networks turned out to
be more intense than when considering the emergency response and recovery stage
separately. The corresponding results can be found in table 6.17 below.
Table 6.17. Co-evolution of Communication and Collaboration (t1-t3)
Effect Communication Collaboration
Within-network par. (s.e.) par. (s.e.)
Out-degree (density)
- 1.7302
 

(.1125)
- 3.2158
 

(.2288)
Reciprocity
.8464
 

(.1984)
1.6446
 

(.3298)
Transitive triplets
.0915
 

(.0147)
.4330
 

(.0988)
Three-cycles
- .0721
 

(.0238) - .1684 (.2283)
Balance
- .0232
 

(.0031)
- .0862
 

(.0229)
In-degree popularity
(Sqrt)
.2318
 

(.0360)
.5042
 

(.0788)
Out-degree
popularity(Sqrt)
- .2215
 

(.0415)
- .3694
 

(.1069)
Between-network:
dyadic

Communication - -
1.1285
 

(.3299)
Reciprocal - - .2137 (.4141)

650
communication
From communication
agreement
- -
- .0615


(.0263)
Collaboration
2.3954
 

(.6272) - -
Reciprocal
collaboration
.0695 (.4643) - -
From collaboration
agreement
-.2981


(.1238) - -
Rate function
Rate Period 1 125.1282 (30.3747) 15.2952 (3.9505)
Rate Period 2 15.5145 (1.0192) 4.2783 (.5025)
P<0.10, P< 0.05, P<0.01 (two-sided)
First of all, the direct effects for overall network evolution trend were stronger
than considering the recovery period alone, with estimated parameter values of 2.3954
and 1.1285 for communication and collaboration network respectively. This indicated
direct cross-enhancement of institutional development for both types of network
structures. The mutuality effects were marginally strong, with parameter values of -
0.2981 and -0.0615. The negative signs can be interpreted as follows. On the one hand,
the existence of a mutual agreement on collaboration tie made it less likely for creating or
maintaining a communication tie. On the other hand, the agreement of communication
ties also made it less likely for creating or maintaining collaboration ties. This means that
looking at the three time stages altogether, the significant structural effects appeared in
†
  

651
each of the two types of networks were not mediated by the mutuality agreement of the
other type of network. The structural evolution of communication and collaboration
networks was better explained by the “embeddedness” of each kind of tie in a multiplex
tie through direct effects. In other words, the network dynamics of communication and
collaboration were the consequences of multiplexity of only the direct influences between
communication and collaboration. The creation and maintenance of the after the
earthquake shock was less mediated by the more binding mutuality ties, instead, the mere
existence of one type of tie was sufficient. This also inferred that the force undergirding
the institutional change— creation and maintenance of a tie in each network
environment—was certainly strong enough to support structural endurance and
transformation. For all three types of examinations, the reciprocity cross-network effects
were not statistically significant at all, thus indicating the existence of a reciprocating
communication or collaboration tie did not have any effect on the creation and
maintenance of the other type of ties.
The direct cross-enhancement of these two types of networks can be further
illustrated by the following example of the within sector (Civil Society) action of
NGOLF.

652
The Case of Actor NGOLF
Throughout the recovery period, the actor NGOLF developed a collaborative
relationship with one of the network actors (NGO33), which performed as a volunteer
coordination service provider in the field of youth development in the city of Chengdu.
Actor #33 was actually functioned as a collaborative program operated by a joint effort
among the Communist Party Youth League, China Youth Palace Association, and the
Hong Kong Youth Association. Therefore, the nature of formation for this particular
actor did involve certain conduct of the state through local government. For example, the
initial physical operational space of actor #33 was built through the approval of the local
government in terms of the provision of designated building spaces and the supporting
staff. However, the program had to reach out to other nonprofit volunteer social groups
and organizations to perform and execute the part of service provision for the targeted
local communities. As one of its many social service partners, NGOLF later established
its collaborative relationship with actor #33 by providing the program with related
educational classes. Other collaborative initiatives also included groups and organizations
performing environmentally related activities. “It’s like they built the house and we fill in
the content”, as described by the participant upon reflecting the characteristics of such a
collaboration. On the other hand, participating in the program also provided valuable

653
networking opportunities for NGOLF to expand its connections with local schools and
teachers. As a newly emerged civil society, part of the advantage of working with actor
#33 was because its long established government-related background easily gained
recognition among the public. Therefore, this kind of collaboration can be thought of as a
mutually beneficial one
161
. The existence of their collaboration tie forged through joint
operation of a program benefiting the targeted communities enabled actor NGOLF to
develop its own expertise in youth development through building its own immediate
communication and collaboration networks, such as with schools, volunteers, and other
social service providers (shown in the red parenthesis in Figure 5.1.23). It also provided
further opportunities for NGOLF to connect with others, such as actor #124 and #4.
Figure 5.1.23 also provides a graphic illustration of the type of institutional arrangements
that mapped out the possibilities of cross-enhancement of collaboration and
communication network dynamics surrounding actor NGOLF.

161
For detailed account, refer to Appendix 6.6.CaseNGOLF.4

654
Chapter 7
Civil Society, State, and Market System

Formation and Expansion of Available Capability Set
As both of the communication and collaboration networks experienced structural
expansion in terms of size and intensity in connections, one capacity that the civil society
actors learned to develop was their ties with the state and the market actors. The changes
in the dynamics among these three domains, I would argue, can be thought as primary
indicators that marked the independent functioning capabilities on the part of the Chinese
civil society after the earthquake event. In this chapter, I explore this aspect of the picture
in detail, particularly paying closer attention to actors’ experiences qualitatively.
I start out with a discussion on the strength formation in terms of identifying the
maximum flow measures among actors within the domains of civil society, state, and the
market. Recall that actor #1 represented the state and was designed as an aggregate for all
the government agencies that participated in the disaster response and recovery process.
And actor #2 represented the market and designed as an aggregate for all the private
enterprises inside the domain. Therefore, the results from identifying the maximum flow

655
for the state and market actors can only be interpreted as the number of different flow
pathways that were available for all other civil society actors to reach out and pass
information towards them, but not vice versa. This was because the two actors were set as
not having any outgoing actions through the design of the original survey. This way, the
flow measures presented here would function as an evaluation of the differing channels
the civil society actors could communicate with these two actors. I argue that the
dynamics of the changes developed over time can have significant policy implications for
managing inter-sector relationships especially under extreme uncertainties like those after
catastrophic disasters.

Pre-earthquake Stage
First of all, for the period before the earthquake, the ways that the state actor
could get information from the existing civil society actors varied and there seemed to be
many cases that the state could not obtain any information from or establish
communication ties with those in the civil society domain. The highest number of flow
paths from which civil society could reach to the state was generated from actors #137
and #51. Both actors had 8 intermediaries in flows from them towards the actor #1. In

656
turn, the state had 8 possible alternative channels from which to obtain information from
actor #137 and actor #51.
After actor #51, the ranking is followed by the communication actions of actor
#119 and #6. Both actors had 7 alternative routes to reach out to the state actor. From the
action initiators’ point of view, such a variety of opportunities for these civil society
actors to communicate and pass information to the government agencies would put these
actors in a relatively powerful position in the formation of structure resilience after a
catastrophic disaster, upon considering the following conditions. For one, the exchanges
of information and communication were in the nature of assisting the development of
expertise on the part of the civil society actor. For another, the communication ties could
opening up further opportunities for the actor to interact with other branches of the
government agencies. From a policy-making perspective, the availability of alternative
routes across a broader range of the remaining actors in the network would be more
promising in understanding the institutional structure of civil society and thus designing
relevant policies, particularly in preparing for the times of disaster response and recovery.
However, for the period before the earthquake, the flow pathways of the network showed
significant disconnectedness and there were large pockets of information disruption
where numerous actors did not have any sources to obtain information from the rest of

657
the network. The vulnerability of such a network in terms of lacking agency activity on
the part of civil society actors generated a kind of social structure with only a few key
players who occupied the dominating positions in passing information and
communicating with the state and the market actor. The rest of the actors remained
“dormant” in action and were, to some extent, less aware of the possible opportunities
that would have held when multiple neighborhood pathways could get them connected to
the government agencies and the private sector. At the same time, the flow connections
towards the business enterprises showed similar trend as that of the state. Only a few of
the civil society actors established alternative routes to communicate with the private
sector while the larger part of the remaining network remained in a state of lack of
communication.

Emergency Response
This state of existence in connecting to the state and the market was completely
changed shortly after the earthquake. Looking through the number of intermediaries that
civil society actors had during the response period, not only the alternative routes for a
particular set of actors grew by size, but also more actors across the remaining of the

658
network started establishing a variety of communication channels towards the state and
the market actors. In turn, they (actor #1 and #2) were also able to obtain information
from multiple pathways of the particular sources originated in the domain of civil society.
During this period, civil society actors were actively building communication ties with
each other and as a result of such proactive efforts, the government entities in the state
actor were having increasing diversity of in-flow of information from a broader
“spectrum” of civil society actors.
Actors #24, #3, and #34 had distinguished themselves by having significantly
higher number of intermediaries when communicating with the government than any
other civil society actor in the communication network. The alternative routes in flows
from these three actors to the state went up to thirty-two connections. What this means is
that upon building up connections with the state, these actors had a wide variety of
opportunities to communicate with or making their voice heard through connecting with
the government agencies. The failure of some information channels did not mean that the
entire communication line between a particular civil society actors to the state actor was
broken. There were abundant supplies of alternatives in the neighborhood of the focal
actors that had the potential to pass the information along towards the target. In others
words, the bond between the civil society actors and the state became more resilient to

659
shocks that tended to break the connections. The higher the number of mediums among
pairs of actors, the more resilient the communication ties were to hold their place at times
of social change or crisis. On the other hand, as civil society actors became more
embedded within the network, the diversity of communication pathways from
increasingly wide variety of sources also became more informative for the government to
find policy measures to enhance the strength of bonds for purposes of disaster
preparedness and mitigation. As actors in the civil society domain built each other up
with the increasing number of connections among themselves, they themselves might not
recognize the roles or positions that they occupy from the perspective of the whole
network structure. It could be an opportunity for the government to step in to raise the
awareness of civil society actors for their roles in strengthening the resiliency of the
structure against future crisis. I now look at the interactions of civil society actors with
the state and market from a number of civil society actors’ perspective.

The case of Social Group Actor #3 (SG3)
For actor #3 in particular, although indirect communication channels were found
to be abundant since its establishment after the earthquake, direct interaction and

660
cooperation with the Chinese government had been rare and one reason was due to the
group’s self-identified function as an “information platform”
162
.
First of all, from the perspective of looking at the self-defined tasks positioned by the
group as a whole, the nature of its functioning in servicing civil society
groups/organizations in general did not provide many task intersections with those
performed by the state.
Since the very beginning of our work, the government must know about us,
doesn’t matter through which channel. They must have known what we were
doing. But for the more direct contacts, such as those activities that would lead to
working connections are very rare. And neither did we intentionally focus on this
aspect of the work. As we have claimed to ourselves, our direct service is
provided towards these NGOs and our work directly complements the
government tasks. On a different level, we have not made direct contacts with any
government branches. (SG3-01-16
163
)
What is important to note is that the participant viewed the group’s role as one that was
complementing the tasks performed by the state after the event of the disaster. This
implied a desire for this civil society actor to make a contribution to disaster relief by
accompanying the efforts of those actions performed by the government. In other words,
the role of the state was being respected, while at the same time, civil society actors were
also keen on observing other aspects of needs for recovery in the society after the

162
Refer to interview account SG3-01-15 in Appendix 7.1.01.
163
For original Chinese script please refer to Appendix 7.1.02.

661
earthquake. Therefore, the interaction between SG3 and the state could be summarized as
a “complementary” type and the origin of this dynamic was based on the definition of
tasks being performed at each stage of time. Secondly, note that although direct
collaboration with the state sector was absent, the information exchange communication
tie with the government branches did exist at the emergency response stage. This means
that there was a common awareness of the state recognizing the initial emergence of this
civil society actor, and at the same time, the latter also functioned in such a way to
facilitate and assist the duties performed by the state sector.
Aside from the types of direct and indirect contacts between the civil society actor
and the state, another aspect of institutional development that could be thought of as
“cross-sector” in orientation was the registration process through which the actors had to
find their own “sponsors” inside the government. This aspect of relationship was peculiar
to the Chinese context, especially during the process when informal social groups were
trying to gain formal registration status. One of the biggest hurdles for group SG3 in its
institutional development was not being able to obtain registration status, as was recalled
by the young participant. This aspect of informality was perceived to be the key in the
future organizational growth of the actor. The following was a direct account from the

662
young participant pointing out the key factor that prevented its group to establish such a
registration relationship with the government.
…we were right stuck at finding a proper sponsor. Looking at the type of work
that we are engaged in, which directly servicing NGOs alone, according to the
current NGO law in China, our work does not belong to the usual categories
servicing disadvantaged groups such as elders, disabilities, children, women, or
the common field that NGOs regularly engage in, such as environmental
protection. So, we could not find a sponsoring unit. Or more precisely, no one
would be willing to. This is a big problem for us. We even asked the Red Cross,
and they had never sponsored anyone like us, nor did they have any conceptual
clarity of sponsoring any kind of civil society organizations. Back then, it was me
and G (SG3-03) went to see them, and in order to help us, they did even prepared
a report asking about our case, but none of us ever got an answer until now and it
became a pending issue now. They (government) won’t provide us with a definite
answer, nor did they tell you yes or no, rather, they just hang us waiting
indefinitely. At last, we came to the Academy of Social Sciences here in Chengdu,
the experience became even more bizarre. They have never heard such a concept
before and would like us to help them draft a set of administrative measures in
managing civil society organizations. In other words, we had to draft some
guidelines to manage ourselves. If they don’t even have these kind of things, there
is no need to even mention the work function matters. If you ask me and G how
shall this issue be resolved, so far, we don’t have answer yet. (SG3-01-17
164
)
The problem of finding an appropriate “sponsoring unit” within government branches
partially arose from the group’s self-defined nature in servicing other civil society
groups/organizations. From the experience of SG3, there were three possible sources that
tend to act as institutional obstacles. One was from the legal point of view. The targets

164
For original Chinese script please refer to Appendix 7.1.03.

663
that the group was trying to provide services towards could not be appropriately
categorized as any one of the common service types among “NGOs” functioning up until
the disaster event. The second factor was related to those organizations that actually did
qualify for being a sponsor, but with a lack of understanding in both the meaning of “civil
society groups/organizations” as well as in the practicing experience of actually being a
sponsor. The fact that the Academy of Social Science was asking the participants of
group SG3 to draft a set of rules in administering civil society organizations exemplified
a primary need for formal rule-based institutions to be aware of and recognize the
functioning of the various forms of institutionalization dynamics among the
group/organizations inside the civil society domain. On the other hand, this also
represented the challenges faced by the Chinese institutional system in the aspect of
governing the emerging processes of civil society, both in terms of formal organizations
and informal social groups, in order to learn and adapt in cases of social change after
catastrophic disasters.
“According to the traditional survival and funding modes of Chinese civil society
groups/organizations, or more precisely one could also say those based on historical
cultural practices, we (SG3) are unique in a sense that nothing with similar nature and
functions of this group ever existed and thus were not supported (by the state). Thus in

664
other words, our work compensations are not being supported”
165
(SG3-02). Notice that
the participant referred to the functioning and funding practices for Chinese civil society
organizations as a type of tradition that can be passed down by culturally accepted habits.
From an institutional perspective, such was a conceptual identification of institutions
through which Chinese civil society groups or organizations’ made their existence as one
belonged to cultural practices rather than institutions as formal rules. The senior
participant provided his interpretation regarding the institutional source of the group’s
registration obstacle as follows:
Therefore, many issues related to our work still remain. The field of Chinese civil
society has just begun its journey. The related institutions are almost non-existent,
or there are no laws or legislations in this regard. Neither the operational
mechanism has been properly established so far. How could there be
implementation mechanisms if there are no institutional rules. Therefore, this
created the problem of non-standardization. Such is not referring to the non-
standardized actions of any one specific organization, but I’m referring to a
phenomenon that is happening to the entire field. We are functioning and
developing the field under a condition where there are no rules or standards, and
neither does any evaluation criteria exists. Besides the hurdles that I mentioned to
you now, there must be many more. If you go to another place, they probably will
tell you another set of problems. This would mean that the field is not
standardized, or there are no rules to follow. If there are rules of conducts for
certain things, their institutions would be complete, and the implementation
mechanisms would be complete, the functioning would be effective. And if all of
these are in place, the rest will be simple…I will do whatever I will have to do

165
For Chinese version, refer to Appendix 7-1-04. Also see Appendix 7-1-CaseSG3-5.

665
and you do whatever you have to do. Then I can execute my accountability. But
for now, we don’t have anything in place. This is the current situation now. (SG3-
02-04)
As we can see, one of the fundamental issues facing civil society actors,
particularly one like SG3 whose functioning areas were not yet clearly being recognized
and incorporated into the rule-based institutional operating mechanisms of the country,
created a certain degree of confusion in terms of the group’s informality status. At the
same time, what could be reflected in terms of the growth of emerging Chinese civil
society actors was one that characterized by actors being increasingly aware of their
institutional environment and a desire to be recognized by the state through their rule-
based responsibilities and duties. Despite of difficulties in functioning in an institutional
climate in relation to the state actor, the participant’s agency drive to make a contribution
for the “betterment” of the society provided a backbone source for continued action in the
civil society domain. As he put it as part of the group’s struggle to survive,
…so in fact, hardships at work still amount and will be to a great extent, but
regardless, we rely on our own ability, on our confidence in being able to
servicing the society, and on all these kinds of dedications… (SG3-02)

666
The Case of Actor #24 (NGO24)
From the findings in the communication network flow measures, actor #24 was
one of the civil society actors having established high level of communication
relationships with both the state and the market sectors immediately after the earthquake
event. The director of the NGO further revealed the specific types of these connections.
For the connections with the state sector, he (NGO24-01) described that two kinds of the
state involvement could be distinguished. One was at the different levels of government
branches, such as at the levels of provincial, regional, city, county, and township. The
other was separated by various kinds of functions performed inside government branches,
such as ministry of forestry, environment etc. For the connections with those inside the
market domain, one type of relationship was with foreign companies operating inside
China and the other one was with domestic private enterprises. From here, we can see
that the actor had established a rather broad cross-sector base to develop its
communication and collaboration connections with the various levels of actors inside the
state and the market sectors. This piece of information was significant in designing future
network studies when investigating detailed connections between civil society actors and
the different kinds of state and market actors. Although the current study treated both
state and market as two aggregate entities performing the role of network actors as a

667
whole, further breaking down and looking at the specific composition of sub-level actors
inside the state and market domains were needed in order to provide a thorough
understanding of the nature of the dynamics among the civil society, the state, and the
market domains.
Looking forward in terms of the future works of the NGO actor #24, the director
expressed his hopes for further “policy coordination” from the various branches inside
the government for the development of grassroots civil society actors. This desire in
looking into the positive changes in the larger institutional environment also arose
directly in relation to the field of activities that the actor chose to engage in. For example,
he particularly pointed out the specific needs of attention be paid from the ministry of
forestry, environmental protection, and tourism. On the civil society side, the role for
information exchange platform was emphasized to be critical in facilitating the
collaborations among peers in the civil society domain. He envisioned that through the
assistance of such as a platform, the initial coordination and the later development of
collaboration dynamics between foundations and NGOs, as well as among NGOs
themselves could be facilitated and sustained over time
166
.

166
For details, refer to Appendix 7.1.CaseNGO24.5.

668
The Case of Actor #123 (SG123)
Among the newly established civil society actors shortly after the earthquake,
actor #123 quickly developed more intermediary communication ties toward both the
state and the market to a level that was comparable to its registered organizational
counterparts at this period of time. The source of the cross-sector initiatives being
activated around this actor was explored further.
From 2009 to 2011, the organizer recalled that there were several private
enterprises that had showed interests in financially supporting the group. The exact
functioning mechanism of such interaction was in the form of businesses providing funds
and the group would execute the programs according to the needs in the field. Up until
the time of the interview, these kinds of programs operated in a way that the group
provided training for teachers across all the middle school and elementary school in the
city of Dujiangyan to learn psychology-related topics. Regarding the group’s specific
interactions with the government, the organizer re-emphasized its role in “bridging” the
communication channels between the mass the government. Just as SG123 said: “we
(NGOs) function as an intermediary that ease up the frictions between the two” (SG123-
01-04
167
). Note that the informant specifically distinguished the function of group actions

167
For Chinese script, refer to Appendix 7.1.04.01. Other details refer to Appendix 7.1.CaseSG123.5.

669
of actor SG123 from that of the general public. And the dynamics between the state and
the mass was perceived to be a somewhat frictional one. The role of the civil society
actors, on the other hand, was perceived and preferred to be detached from both the state
and the general public. And only through such a realization would the civil society fully
establish its functioning. Such envisioning would partially explain the actor developing
an increasing number of intermediaries to diversify its communication channels to the
state and market actors.
In the policy-making arena for disaster preparedness, for example, the particular
position for actor #3, #24, #34, and #123 after a disaster can be singled out as a case to
look deeper into ways of designing interventions to boost the information flow across the
entire network. Since each one of first three actors had thirty-two medium flow pathways
that connected them to the state agencies during the response period, one could examine
who these mediums were, their registration status, location, and field of expertise in order
to categorize what types of relationships they have with the government agencies. How
the three actors got connected with the intermediaries in the first place can also be
informative in tracing the origin of structuration of ties over time. These understandings
altogether can build up the foundation of developing policy tools that can enhance the
capability of other actors in the network to connect well with others as well as to the state

670
agencies. At the same time, by doing so, the state could also increase the pathways for it
to obtain information from the civil society actors. This is particularly critical during the
period of emergency response after a disaster. With a well-connected network structure
built in place, the availability of alternative connections for pairs of actors will provide a
“shield” in making communication occur in time and information passed accurately
throughout the network.

Long-term Recovery
During the long term disaster recovery period, the general structure of the flow
pathways among pairs of actors maintained their dynamics from the previous emergency
response period. Civil society actors remained to be active in building up communication
relationships and continued to be less dependent on just one or two alternative routes to
reach out to their target actors. Actors had kept establishing themselves with a wide
variety of medium connections. Actor #51 had a total 28 pathways to communicate with
the state, which immediately followed by actors #3 and #32. Actor #3 remained to have
the highest number of pathways to reach the state by having 31 alternative routes. Actor
#32 and #6 emerged as having the next highest number of pathways towards the state.

671
The overall structure of the recovery communication network tended to follow similar
patterns of resiliency in terms of information flow as that of the period before. This can
be interpreted as a signal of the level of motivation from the civil society actors’ side.
From the policy-making perspective, this stage of the post-disaster relief process also
signified an active invitation from the part of civil society to become part of the policy-
making process that could facilitate the institutionalization of the tie connections. In this
case, the construction of “resiliency” did not only have to be a one-way route for
government to design what should be “best” for the society to withstand periods of
dramatic change or catastrophic crisis. The process can also be “home-grown” from the
society’s end and such bottom-up dynamic would have the capacity of being endured
over time because a sense of desire for agency action became the primary sources and
motivations. For such incidences, the structural resiliency would thrive in policy
environments that took into account the efforts from both the actors of civil society and
the role of the government for long term disaster mitigation. In the following paragraphs,
I use the examples of actor #51 and #97 to qualitatively illustrate how agency desires
could be understood.

672
The Case of Actor #51 (NGO51)
Interaction with the State Domain
The interactions with the state sector were perceived as a critical factor for the
projects implemented by NGO51 during the disaster recovery stage. The most direct
support that the actor received was the provision of the physical activity space for the
sixteen community service center locations. According to the program officer (NGO51-
01), this did not just include the location itself, but most importantly the permission to
use the buildings and the rooms free of charge. “This in fact was a significant kind of
support. At least when we went negotiating them, they were willing to provide with these
basic facilities
168
” (NGO51-01).
One other kind of support that the program officer would most want to see was
the institutional support provided by the government. Recognizing that efforts of this
kind would take time to take effects, the actor’s position in the role of the government is
revealed as follows:
What we would hope to see the most is some kind of institutional support on the
registration of civil society groups. This is one factor that is relatively difficult for
them as of now. Whenever we are having discussions with local governments
about our programs, we will always mention our hopes for a more relaxed
registration system that can facilitate the process. They might say that they would

168
For further account, refer to Appendix 7.1.CaseNGO51.5(1).

673
consider it. But you know this, this is not something can have observable results
in a short period of time. So what we can do is to ‘dig’ slowly and work on it
gradually. (NGO51-01-13
169
)

Interaction with the Market Domain
The market sector in this case, played an active role in building collaborative
interactions with NGO51. In fact, the finances of all the sixteen local community service
centers were supported by private firm N, a well-known global enterprise in the mobile
phone industry. It was actually the firm that first initiated its contact with NGO51 to
communicate its intention in making a collaborative effort in the area of disaster recovery.
The specific factor regarding actor NGO51 that attracted the firm’s attention lied in its
original model being practiced in Shanghai before the earthquake event. The uniqueness
of the model was its flexibility in drawing together the resources from different civil
society actors as grassroots “social organizations” to implement community service
programs.
Back then, we were doing works related to community services in Shanghai, they
(N) saw that our Shanghai model worked out very well. Our main work there was
to build community service platform. There are currently a lot of community
service centers, but many were led by the government, such as entities like the
residential committees. The government provided us with permission to use
certain available spaces, and would delegate the management power to us. Since
we do have these permissions to manage things, we will introduce various kinds

169
For original Chinese script please refer to Appendix 7.1.05.

674
of local civil society organizations to come in and implement activities….this is
why (N) started to inquire us about the possibility to build such a community
service platform in the disaster areas as well. (NGO51-01-14
170
)
From the participant account of NGO51, I summarized two traits of the role of the market
sector. One was that although limited by the network analysis design of this research in
revealing the outreach ties originated from the market domain, the qualitative
examination showed that actors in the market system were actually active participants in
building collaborations with civil society actors. Second, their desire to perform social
responsibilities after the disaster did not end with simply providing financial support to
the more grassroots civil society actors. A genuine concern for the long term
development for the local communities and the participating civil society actors was also
taken into their decision-making process
171
.

The Case of Actor #97 (SG97)
Observing across the three periods before and after the earthquake, the number of
cross-sector intermediaries that actor #97 created quickly rose from only 3 toward the
state and market domains to 20 and 19 toward the two respectively over the long term
recovery stage.

170
For original Chinese script please refer to Appendix 7.1.06.
171
For further account, refer to Appendix 7.1.CaseNGO51.5(2).

675
For the case of this actor, one of the primary forms of institutional development in
terms of cross-sector interaction was realized through financial mechanisms. And the
sustainability of such a mechanism in place constituted as a critical factor in the “long
term survival of most NGOs” (NGO97-01). The actor particularly focused in community
health development, thus long term financial support beyond the one-year mark would be
important for its action to take effects inside the communities. In fact, according the
organizer, “one year (of funding) is definitely just a beginning. Especially for health-
related activities, the availability of at least multiple years’ of stable support is a must in
order to observe change” (NGO97-01). Because of the type of service being chosen, the
kind of support from foundations would result in “short-term effect” as they often will be
terminated in a year’s time. The support from private enterprises would also involve a
sense of instability as the funding depends on the operational situations and the directions
of the business goals. Essentially, the types of supports from foundations and the market
actors were not perceived to be “sustainable and stable” sources.
So the only long-term alternative (for funding) might be on the government side,
and if it would be committed to support us in the long term. This kind of support
would have to rely on how they think of our work and their ways of conducting
their work in terms of the degree of recognition and support. (SG97-01-11
172
)

172
For original Chinese script please refer to Appendix 7.1.07.

676
As we can see, the role of the state, from the perspective of a civil society actor,
was one that could fulfill the function of being the critical source of its functioning
stability. However, one important format of the type of support that could be provided by
the state sector was expected to be coming from its “recognition and ideological support”
for the works of NGOs. Note that such a position being claimed by the organizer of actor
#97 could be attributed to two types of the possible transformations inside the civil
society domain after the crisis. One was the growth of civil society actors in looking at
themselves as playing a “standing up” role in proactively making contributions to the
larger social “resiliency” of the country against future risks and crisis situations.
Secondly, the desire for the civil society actor to be known and understood by the state
actors reflected a needed re-evaluation of the nature and activities of Chinese civil society
particularly after the 2008 disaster. Therefore, comparing the institutional changes made
inside the civil society domain before and after the earthquake would contribute as
valuable cultural basis for future policy designs preparing the society for future
disasters
173
.
Similar to the nature of cross-sector interaction between actor #49 and the state
actor, the dynamics that actually contributed to the connection between actor #97 and the

173
Further account refers to Appendix 7.1.CaseSG97.5 (1).

677
state was one of latter’s “recognition and acceptance” of the activities conducted by the
civil society actor in the field. Such awareness on the part of the state as a type of
showing its indirect permission towards the functioning of the actor, from the perspective
of the civil society actor, was an alternative way of the state’s provisioning of open
“opportunities” for it to exercise its envisioned functionings. To put it in the words of the
organizer:
It is already such a great improvement when it (local government) actually
allowed us to do it. If they accepted you conducting this field of work, that
already indicated their recognition. This is because they could also prevent you
from doing it, and if that is the case, what can we do? So, since they provide you
with permission to do this, it would indicate the availability of some open
opportunities. And that is good. There were many times when we were organizing
activities at the local communities, and it is the government that provided us with
spaces for our office work and activities. All of these were types of supports.
(SG97-01-12
174
)
Note that the permission to offering the usage of office spaces and public spaces was
perceived as an important form of support from the state actor. This was perceived as a
culturally unique form of cross-sector “collaboration” as there were not a set of
institutional rules guiding the specific actions that needed to be taken on the different
sides of actors. What could be inferred was that in a cultural context where the civil
society domain was emerging and beginning to take shape in both actions and

174
For original Chinese script please refer to Appendix 7.1.08.

678
interactions with the state, the forms of connection-building between actors inside civil
society and the state could take a variety of supportive channels. For example, in the
execution of one of the community development projects in the outer suburbs of Chengdu,
at the same time that actor #97 was collaborating with other NGOs performing health-
related tasks, the spaces for carrying out the needed activities were provided by the local
government. From the perspective of civil society actors, this form of support was further
interpreted as their “collaboration” with the state sector
175
.
In terms of the interactions between actor #97 and private enterprises inside the
market domain, the existing collaborative relationship for this particular civil society
actor was originally established for the purpose of providing educational support for
children in the disaster areas. The project was first initiated by the private enterprise and
the role of actor #97 was one to assist the implementation in the field. However, such
communication and collaboration relationships with the market sector were not set as its
primary functioning focus as compared to its newly start-up program in health-related
community developments, which was supported by actor #51. Although actor #97 was
not involved in any cross-sector project with actors from all three domains joining

175
Details refer to Appendix 7.1.CaseSG97.5(2).

679
together at the same time, the organizer provided her own vision of how a formal
“collaboration” can be defined
176
:
Currently, we are only collaborating with either one side or the other, but haven’t
worked with the two sides at the same time yet…but the ones we have are not in
terms of in-depth collaboration neither. We don’t have any of those kinds of
collaborations as of now. ****seemed to have done some good works in this
regard, the government was providing the funds, private enterprises are
participating, it’s like all sides are putting into financial and human resources to
execute the process. I think this should be the authentic collaboration process
when everyone truly took part in it, participate in joint discussions and implement
the plans together. (SG97-01-13
177
)
In summary, actor #97 generally perceived its own functioning after the
earthquake as one that was complementary to the state actions. As a newly formed
grassroots, it yearned for stability offered by the process of institutionalization. At the
same time, it also recognized the significance of its own capacity development over the
longer term in order to make a contribution to the social resiliency of China since the
earthquake. The connection to the state, from the perception of actor #97, became the key
factor in reaching both functioning stability and growth. The actor not only desired to be
known and understood by the state, but also proactively built communication ties toward
the state with a long-term orientation.

176
Details refer to Appendix 7.1.CaseSG97.5(3).
177
For original Chinese script please refer to Appendix 7.1.09.

680
The Case of Actor NGOLF
The consideration of functioning stability in relationship to the state like that of
actor #97 was also resonated by the experiences of actor NGOLF. From the key
participant’s (NGOLF-01) account, its registration status was assigned by the office
managing civil organizations within Chengdu’s Municipal Civil Affairs Bureau. This
made the actor the first nonprofit registration certificate given to a civil society
organization permitted to perform without the existence of a sponsor work unit. And it
was at the time transitioning from a disaster recovery-oriented organization to one that
conducts regular social work activities under “normal” conditions. To the participant, this
showed a “gesture” on the part of the government, that at least “it would like us to
conduct out activities by becoming public, under its general supervision and
acknowledgement”. Although admitting that there were still gaps between what the civil
society actor hoped for in an ideal institutional environment provided by the government,
the overall climate at the time as compared to before, was “much better” from the
perspective of the participant
178
.

178
Details refer to Appendix 7.1.CaseNGOLF.5.

681
Clustering across Civil Society, State, and Market Domains
Besides the increasing opportunities for civil society actors themselves to be
engaged in their local neighborhoods, I also investigated the clustering behaviors of civil
society actors around the state and the market
179
. Recall that neither of these two actors
had any outgoing connections but only incoming nominations from the civil society
actors. From a policy-making perspective, the node level clustering coefficients for the
two actors can be singled out as indicators reflecting the tendencies for civil society
actors to form clustered neighborhoods. The more tightly weaved-together the local
neighbors for the state and the market actors, the easier and faster the information
exchange, which is a critical factor in designing timely cross-sector initiatives when
providing assistances in response to the disastrous impact of the earthquake.
Based on the incoming nominations that constructed the immediate neighborhood
of the state actor, there were a total of 496 pairs of actors or possible ties within its
neighborhood immediately after the disaster event. And 15% of all the possible ties
among these neighbors were present. Compared to the period before the earthquake, its
number of pairs of neighbors increased almost ten-folds from 55 to 496. It means that the

179
See Appendix 7.1A, B, C for UCINET clustering output measures.

682
crisis situation not only brought forth the emergence of more actors in the civil society
but also an increasing level of recognition to see the state as an important communication
partner, especially after the earthquake. Over time, the size of the state actor’s local
neighborhood stayed at somewhat the similar level with a small amount of drop of ties to
465 ties. However, the percentage of all these possible ties that actually presented during
the long-term recovery stage went up from 15% to 18.9%. On the one hand, this
demonstrated that the civil society actors’ willingness to keep the enduring
communication ties with the state actor both short term and long term after the disaster.
On the other hand, the signs for an institutionalization process in terms of dyadic tie-
building among pairs of civil society actors ensured the state actor to be embedded in
increasingly higher clustered neighborhood. This type of action could be understood as a
representation of trust on the part of civil society actors towards the state in times of
turbulence and this type of effort did not fade away over time. There was a genuine
tendency of endurance of these clustering neighborhood relationships, which signified as
a sense of willingness to institutionalize the structure of embeddedness from the
perspective of civil society actors.
What then, were the opportunities for change for the state actor in this context?
For planners and policy makers who are designing the various ways to involve the civil

683
society especially at times of uncertainty or after catastrophic disasters, the behavioral
change of actor neighborhood clustering among civil society actors could offer clues
targeting a variety of channels to generate information exchange throughout different
network environments. Specifically, the attributes of those initiated the sustained
structures of clustering within the state’s local neighborhood can be traced. This can be
followed-up by interventions developed to target the key actors in building up the
communication connections, thus forming efficient ways for the state to exchange
information with those in the civil society domain.
The market was perceived to be another critical player in the disaster response
and recovery context. Before the earthquake, the actor as a whole had only 10 pairs of
civil society actors within its immediate neighborhood. And out of these 10 possible ties,
30% were actually present. In other words, few civil society actors saw the private
enterprises inside the market system as a critical communication partner before the
earthquake. This attitude changed drastically after the disaster, when the size of the local
neighborhood for the market actor increased almost twenty-folds from 10 to 231 ties. At
the same time, the percentage of all these possible connections that actually presented
only dropped by 7.5% to 22.5% at the emergency response period. This means that from
the civil society’s perspective, the communication and information exchange ties with the

684
private sector in the market domain were recognized to be critical and the businesses
themselves were being embedded in increasingly clustered local neighborhood. Over the
long term recovery period, the size of its immediate neighborhood dropped to 171 pairs
of civil society actors. But the realization of actual tie presentation among these actors
increased to 26%. Those who had direct ties with the business sector were increasingly
clustered together with each other and the result of this process was that the private sector
being gradually embedded in highly clustered neighborhood. This point can also be
demonstrated by looking at the network graphs comparing before and after the disaster.
Examining figure 7.1 for communication network before the earthquake event, we
can see that actor #2, the aggregate of private businesses (circled in brown color), was
located near the edge of the densely clustered section.

685

Figure 7.1. Pre-earthquake Communication Network (Actor Attributes in Date of
Establishment, Geographic Location, and Registration Status)
The figure provided a visual demonstration that the actor was nominated by relatively
few civil society actors at this stage of time but those who made it into the actor’s local
neighborhood were relatively highly clustered. The result of such embeddedness in the
local structure was that the private sector had some connections to the clustered core of
the network but was not recognized as a key player. Another characteristic of this graph
that is related to the area centered in actor #115. Recall that the actor made a lot of effort
in initiating ties to others and created a relatively large local neighborhood for itself. This
was the reason that we observed the “star-like” feature of its connections with the actors
in its immediate neighborhood. However, the neighbors themselves were actually not that
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
*Color: establishment
Yellow: before EQ; Blue: after EQ
*Shape: location
Circle: Sichuan based; Square: Non-Sichuan
*Size: Registration status
Small: registered; Large: non-registered

686
well connected with each other at all. Almost half of the actor’s (#115) ties were made
towards those actors whose only connection was to actor #115. Approximately the other
half of its ties actually facilitated actor #115 to build communication relationships with
some of those inside the densely clustered section of the graph. For policies that are
intended to bring more those civil society actors who are relatively less engaged in the
communication network, or on the periphery of the network “closer” to the information
exchange activities among those who are embedded in the highly clustered section of the
network, actor such as #115 was unique and critically important in making such a
connection. For actors #136, #92, #90, #113, and #124, the only way for them to be
informed of what’s going on inside the densely clustered part of the graph was through
actor #115. Therefore, this actor was an important intermediate for information to spread
throughout the network.
Observing the distribution of the actors shortly ad long term after the earthquake
(figures 7.2 and 7.3), the most distinguishing feature one can notice is the inclusiveness
of the network with the disappearance of the “isolates” from the previous period.

687

Figure 7.2. Emergency Response Communication Network (Actor Attributes in
Date of Establishment, Geographic Location, and Registration Status)

1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128 129
13
130
131
132
133
134
135
136
137
138
14
15
16
17
18
19
2
20
21
22
23
24
25
26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
*Color: establishment
Yellow: before EQ; Blue: after EQ
*Shape: location
Circle: Sichuan based; Square: Non-Sichuan
*Size: Registration status
Small: registered; Large: non-registered

688

Figure 7.3. Long-term Recovery Communication Network (Actor Attributes in Date
of Establishment, Geographic Location, and Registration Status)
The overall structure itself also became very tightly knit together. The state and
the business actors (#1 and #2) were becoming more embedded at the center of the
cluster. Their embedded positions remained during the long term recovery stage (see
figure 7.3) while being closely tied to some of the key civil society actors in the highly
clustered neighborhoods. Therefore, the experiences of the state and the market could be
generalized as one that turned from lacking attention from the civil society actors before
1
10
100
101
102
103
104
105
106
107
108
109
11
110
111
112
113
114
115
116
117
118
119
12
120
121
122
123
124
125
126
127
128
129 13
130
131
132
133
134
135
136 137
138
14
15
16
17
18
19
2
20
21
22
23
24
25 26
27
28
29
3
30
31
32
33
34
35
36
37
38
39
4
40
41
42
43
44
45
46
47
48
49
5
50
51
52
53
54
55
56
57
58
59
6
60
61
62
63
64
65
66
67
68
69
7
70
71
72
73
74
75
76
77
78
79
8
80
81
82
83
84
85
86
87
88
89
9
90
91
92
93
94
95
96
97
98
99
*Color: establishment
Yellow: before EQ; Blue: after EQ
*Shape: location
Circle: Sichuan based; Square: Non-Sichuan
*Size: Registration status
Small: registered; Large: non-registered

689
the earthquake to one that was being perceived as important communication partners to
be drawn into the institutionalization process of relationships among civil society actors.

Registration Institutional Status and Cross-sector Dynamics
Cross-Sector Communication Networks
Pre-earthquake
I now investigate the internal and external-group ties built by civil society actors
with the state and the market sectors. Recall that the study was originally designed to
emphasize the actions of civil society actors, the ties that were connected to both of these
two aggregate actors were taken as approximations of how actors in the civil society
domain perceived and built their communication and collaboration ties towards the state
and market. As a general observation from table 7.1, the total number of ties for
government agencies and private enterprises, regardless of in or between groups, were
relatively small compared to the actions inside the civil society domain discussed earlier.
Table 7.1. Communication Network Pre-Earthquake Inter-sector relationships
Internal External Total E-I
State 8.000 3.000 11.000 -0.455
Market 3.000 2.000 5.000 -0.200

690
Again this confirmed that the communication connections towards the state and the
market were not as prevalent as those among civil society actors during the period before
the earthquake. This suggested that the state and the market domains were operating in a
highly detached manner from the civil society domain at the time. Further examination of
the internal and external ties for the state and the market also indicated that registered
actors were more likely to initiate communication ties or exchange information with
government agencies and private enterprises. Overall, there was a very loose
communication attachment between the civil society and the two aggregate actors. This
kind of state-market-society arrangements contains information flow only among the
actors within each of the domain itself rather than promotes exchanges across sectors. As
a result, neither the state nor the market actor had enough information channels built to
understand what was going on inside the civil society. This is not an ideal type of cross-
sector institutional arrangement particularly for a society to deal with incidences of
dramatic change, such as one triggered by catastrophic disasters.

691
Post-Earthquake
I now look at the inter-sector relationships among the civil society, the state, and
the market exemplified through the actions taken by the civil society domain after the
earthquake. Note that in the context of this study, the registration attributes for both the
state actor and the market actor were originally coded as registered. Therefore, the group
internal ties that were linked to them were actually initiations through registered actors
and the external ties would have to be initiated by non-registered civil society actors. For
the emergency response period (see table 7.2), both the internal and the external ties for
the state and the market increased as compared to before the earthquake.
Table 7.2. Communication Network Emergency Response Inter-sector relationships
Internal External Total E-I
State 21.000 11.000 31.000 -0.313
Market 16.000 6.000 22.000 -0.455

Table 7.3. Communication Network Recovery Inter-sector relationships
Internal External Total E-I
State 22.000 9.000 31.000 -0.419
Market 13.000 6.000 19.000 -0.368

During the recovery stage, government agencies in the state domain had more group
internal ties (22) than group external ties (9). This means that more registered actors
tended to be attracted by the state than those unregistered actors. In other words, the

692
informal groups reached out relatively less for the state to develop communication and
information exchange ties. There was a similar tendency for the ties built with the private
enterprises (13 internal ties and 6 external ties).
Looking at the actor level E-I index for the two aggregate actors, the measure for
the state was -0.313 and the measure for the market was -0.455 for the emergency
response stage. Thus, both state and market actors had certain degree of tendency to
attract ties from registered actors. But the state actor had relatively less tendency towards
group closure with registered civil society actors than the market. During disaster
recovery period, the E-I index measure for the state actor went from -0.313 to -0.419,
while the intensity for group closure for the market actor decreased from -0.455 to -
0.368. Thus, over time, the market actor tended to have less in-group closure while the
state actor continued to draw attention from the registered ones. Therefore, the general
characteristics for inter-sector communication based on the registration status condition
was one with increasing tendency for ties between registered civil society actors with the
government agencies, and with decreasing tendency for private enterprises to have ties
with registered civil society actors. The policy implications for such results can be two
folds. One is that the index measures here can be used as suggestive indicators signaling
the behavior of the different groups within the civil society towards the state during

693
periods after a disaster. In the Chinese case, attention had to be paid for government to
build further ties with those informal and non-registered groups. Information exchange
activities with the registered and the non-registered group actors are equally important as
the two groups might be occupying different areas of expertise as well as location
practices for the disaster-impacted communities. The other point is that the private
businesses seemed to be gaining attention by more non-registered civil society actors
over time, especially into the disaster recovery stage. The factors that contributed to
building such attraction can be identified and incorporated into the future policy designs
in terms of forging stronger information exchange ties between the market system and the
civil society for periods of long term disaster relief.

Cross-sector Collaboration Networks
Having examined the communication network, let’s go one step further to observe
what happened at the cross-sector level of the collaboration network over time. The
following tables (7.4, 7.5, and 7.6) depicted the actor-level variability in group closure
regarding the collaboration relationships of these two actors.

694
Table 7.4. Collaboration Network Pre-earthquake Inter-sector relationships
Internal External Total E-I
State 6.000 4.000 10.000 -0.200
Market 4.000 2.000 6.000 -0.333

Table 7.5. Collaboration Network Emergency Response Inter-sector relationships
Internal External Total E-I
State 16.000 8.000 24.000 -0.333
Market 9.000 5.000 14.000 -0.286

Table 7.6. Collaboration Network Recovery Inter-sector relationships
Internal External Total E-I
State 18.000 8.000 26.000 -0.385
Market 8.000 6.000 14.000 -0.143
Recall that any of the collaboration ties that the two aggregate actors had were
actually being nominated by the civil society actors. This feature when being applied to
observing the processes of network structural change in this research context, was used as
a tool to interpret behavioral changes of civil society actors towards the state and the
market.
Before the earthquake, we can see that the un-scaled E-I index measure for the
state actor was -0.200 and -0.333 for the market actor. This suggested two sides of the
story. One was that both the state and the market tended to have a mild level of group
closure with registered civil society actors, with the market having a relatively higher
tendency for building collaborative projects with registered actors. Shortly after the

695
earthquake, the degree for registered civil society actors to be associated with the state for
project collaboration increased while the degree for the market to have group closure
decreased to -0.286. Both the state and the market still had tendencies towards group
closure shortly after the disaster. But it seemed that there was an increasing degree of
variability on the market side towards building collaborative projects with non-registered
actors. This inferred that during the short term response stage, the inclination of market
entity to collaborate with only registered civil society actors receded. Such could be a
signal showing that for those non-registered civil society actors, their awareness of the
importance of finding project collaboration relationships with private enterprises
increased. Throughout the long term recovery stage (see table 7.6), such trend persisted
and endured. The tendency for the market to have collaboration ties with registered civil
society actors turned out to be even less than the emergency response period. The state
actor seemed to be consistently gaining attention from the registered actors in the long
term disaster recovery phase. This evidence showed that there was still room for the state
to build collaborative ties with more non-registered civil society actors to facilitate their
long term development. On the other hand, it was encouraging to find that the
collaboration initiatives between the market and the grassroots social groups were
blooming. This was in accordance with the motivation and commitment side civil society

696
actors at this stage of time. In the Chinese context, when it comes to developing sustained
efforts in making a contribution to the disaster recovery process, it was not just the
actors’ motivations and the commitments that count, the collaborative relationships with
both the state and market actors were important for civil society actors when facing long
term growth. Here, I use the example of actor #49 to illustrate this point.

The Case of Actor #49 (NGO49)
For the case of actor #49, a proto-type of cross-sector collaboration was made
possible through one of its community development projects implemented in one of the
disaster-impacted cities. Such a cross-sector initiative was named as “government, private
enterprises, and NGOs three way collaboration model” from the account of the organizer
(NOG49-01). In the project, the collaborative efforts was implemented by government
providing the permission to use the land, a private enterprise supported the financial
means to build a community center, and actor #49 was responsible for executing the
project through the operation of the center. This model was first implemented in the city
of Shifang, and then further being “copied” in another community development project in
Dujiangyan through the funding of a cellphone carrier functioning as the private
enterprise. Essentially, the key motivational driver for this mode of cross-sector

697
collaborative initiative was one that emphasized on the role of the NGO actor exercising
its specialty in providing social services promoting community development. The role of
the state and the market were considered to be more of a supportive type to facilitate the
functioning of the actor #49. Since 2008, as recalled by the organizer (NGO49-01), the
actor #49 as an organization had provided social services through this cross-sector model
at a regional level every year.
We started this project since 2008, and it has been a quite successful one. This is
also a good collaboration model. As of now, our headquarters has already moved
to Shifang, and the new office building has being completed and current in use.
The usage of this land is also permitted by the government. Every year, we will
conduct social service related works within the entire region every year. I think
this was a relatively new model, with private enterprises providing financial
support in the backdrop. (NGO49-01-07
180
)
From a long term recovery perspective, this particular model of cross-sector collaboration
had transitio

Asset Metadata

Creator Lu, Jia (author)

Core Title The Wenchuan earthquake recovery: civil society, institutions, and planning

Contributor Electronically uploaded by the author (provenance)

School School of Policy, Planning and Development

Degree Doctor of Philosophy

Degree Program Policy, Planning, and Development

Publication Date 11/27/2013

Defense Date 06/14/2013

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag civil society,governance,institutions,OAI-PMH Harvest,resilience,social networks,sustainability

Language English

Advisor Banerjee, Tridib K. (committee chair), Cooper, Terry L. (committee member), Lynch, Daniel C. (committee member)

Creator Email jlu1@usc.edu,lujia78@yahoo.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-354926

Unique identifier UC11288062

Identifier etd-LuJia-2192.pdf (filename),usctheses-c3-354926 (legacy record id)

Legacy Identifier etd-LuJia-2192.pdf

Dmrecord 354926

Document Type Dissertation

Rights Lu, Jia

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Abstract (if available)

Abstract The importance and the theoretical significance of the civil society construct in the public sphere and its involvement in the policy decision-making process have long been emphasized by scholars in policy and planning. The theory itself has yet to deal with the role of a particular set of actors in civil society through a process of social change. My research approaches this piece of the social justice issue by defining a set of foundational problems called the ""theoretical paradoxes of action"": 1) If the institutions of planning exist to reduce uncertainty in our lives and thus provide social order, how do they deal with unexpected change? 2) If by definition, institutions exist to provide stability and meaning to social life, to what extent can they contribute to the ability of society to learn, adapt, and reorganize to meet urban challenges? This dissertation tackles this problem set from the perspective of civil society actors through a procedural-action-oriented approach, while taking into consideration of the diversity of planning cultures across countries. ❧ I investigated the role of civil society in developing long-term collaborative efforts for urban settlements to cope with risks and uncertainties associated with catastrophic disasters. Using emerging citizen groups and non-governmental organizations (NGOs) as the main unit of analysis, the primary intention of the study is to examine their role in forming communication and collaboration governance networks during the post-earthquake response and recovery period. It seeks to explore the experiences of social groups and organizations' participation during the initial three-year recovery process after the 2008 May12 Wenchuan earthquake in Sichuan province, China. I adopted a case study methodology with a mixed-methods research design. The quantitative method utilizing social network analysis to look at the emergence and evolution of institutions inside the domain of civil society, their relational arrangements among each other, as well as to the state and the market domains at different points of time. The supplemental qualitative study of key participants representing group/organizational actors focused on in-depth understanding of the meaning and the driving forces of institutional change inside the civil society domain. Two types of contextual environments--communication and collaboration network structures--were being investigated along with a longitudinal study of their evolution over three time periods: before the 2008 Wenchuan earthquake, immediate emergency response and short-term recovery period, and the longer-term (up to three years) recovery stage after the earthquake. ❧ Using UCINET network analysis software program, I focused on understanding the macro and micro structural characteristics of how social groups and organizations built their communication and collaboration networks over three distinct periods of time. Each of the six structural environments (three communication networks and three collaboration networks) was investigated separately to look at how actors were connected and embedded within their local and global network environments. Accompanied by qualitative data collected from in-depth interviews and ethnographic field observations, the analysis demonstrated primary evidences in understanding the formation, persistence, and the sustenance of the action structures for both communication and collaboration network environments before and after the Wenchuan earthquake. ❧ I further developed longitudinal network models in discovering the rules that governed the dynamic network behavior over the specified three periods of time. I utilized the SIENA program implemented in the R statistical system to longitudinally investigate: 1) Whether the institutional status in terms of actor registration had an effect on communication and collaboration behavior