Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Engaging together: exploring the peer effects of civic engagement
(USC Thesis Other)
Engaging together: exploring the peer effects of civic engagement
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Engaging Together:
Exploring the Peer Effects of Civic Engagement
by
Ada Yue Li Sarain
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(POLITICAL SCIENCE AND INTERNATIONAL RELATIONS)
August 2022
Copyright 2022 Ada Yue Li Sarain
ii
Acknowledgements
This study would not have been possible without the support of many people. First and
foremost, I would like to express my sincere gratitude to my advisor Prof. Jefferey Sellers for his
continuous support and guidance throughout my career at University of Southern California. His
patience, insightful advice, and encouragement have been invaluable, for which I’m
indescribably grateful.
Great thanks to my dissertation committee members, Prof. Ann Crigler and Prof. Aimei
Yang, for offering their support, extensive knowledge, and helpful suggestions. I own them my
heartfelt appreciation. Special thanks to Prof. Stanley Rosen for his insightful feedback on this
project.
Thank you to Prof. W. Wang, Prof. J. Yin, Mr. H. Qing, and Ms. Y. Chen for helping me
recruit participants and administer surveys in China. The comparative component of this study
would not have been possible had it not for their support.
I am deeply indebted to all my participants for generously sharing their time, opinions,
experiences, and perspectives. Their responses contributed to enhancing our understanding of
civic engagement and peer effects. While I cannot thank each of them by name here, I will
always remember what I learnt from them.
Special thanks to all faculty members of POIR, for sharing their knowledge and wisdom;
to faculty members of the Spatial Science Institute and Department of Computer Science, for
broadening my intellectual horizons and inspiring me to pursue interdisciplinary research; to all
administrative staff, for making the life as a graduate student much easier; to my colleagues, for
their support and friendship.
iii
Finally, infinite thanks to my family, especially to my wife, for being the anchor in my
life; to my parents, for their love and support throughout the years; to my grandparents, for
believing that I could be anyone I want to be and do anything I want to do; to Amelia and
Serena, for their emotional support. I am forever grateful for you.
iv
Table of Contents
Acknowledgements .................................................................................................................... ii
List of Tables............................................................................................................................ vii
List of Figures ......................................................................................................................... viii
Abstract ..................................................................................................................................... ix
Chapter 1 Introduction ................................................................................................................ 1
1.1. Overview of the Study.................................................................................................... 3
1.2. Organization of Chapters................................................................................................ 4
Chapter 2 Theoretical Framework and Literature Review ........................................................... 6
2.1. Theoretical Framework .................................................................................................. 6
2.1.1. Rational Choice ..................................................................................................... 7
2.1.2. Socio-ecological Model ....................................................................................... 10
2.1.3. Theoretical Framework of Peer Effects in Civic Engagement .............................. 14
2.2. Civic Engagement ........................................................................................................ 19
2.2.1. Defining Civic Engagement ................................................................................. 19
2.2.2. Explanatory Variables of Civic Engagement ........................................................ 22
2.2.3. Operationalizing Civic Engagement .................................................................... 28
2.3. Peer Effects, Individual Outcome, and Civic Engagement ............................................ 31
2.3.1. Definition of Peer Effects .................................................................................... 31
2.3.2. The Link Between Peers and Individual Outcome................................................ 32
2.3.3. Exploring the Causality of Peer Effects Through Quasi-experiments ................... 36
2.4. Contributions of the Current Study ............................................................................... 43
Chapter 3 Methodology ............................................................................................................ 45
3.1. Research Design .......................................................................................................... 46
3.1.1. A Nested Most Different Systems Comparative Analysis .................................... 46
v
3.1.2. Case Selection ..................................................................................................... 50
3.2. Data Collection Process ............................................................................................... 54
3.2.1. Data Collection for the Quasi-experiment ............................................................ 54
3.2.2. Data Collection for the In-depth Interviews ......................................................... 59
3.3. Measuring Variables .................................................................................................... 61
3.3.1. Measuring Civic Engagement .............................................................................. 61
3.3.2. Background Variables ......................................................................................... 65
3.4. Data Analysis Approaches ........................................................................................... 66
3.4.1. Data Preprocessing .............................................................................................. 66
3.4.2. Analyzing Quantitative Data from the Quasi-experiment ..................................... 69
Chapter 4 Results ...................................................................................................................... 72
4.1. Data and Descriptive Analysis...................................................................................... 72
4.2. Linear-in-Means Analysis: Peer Effects on Overall Civic Engagement ......................... 77
4.3. Linear-in-Means Analysis: Peer Effects on Different Forms of Civic Engagement ....... 81
4.4. Summary of Findings ................................................................................................... 91
Chapter 5 Discussion ................................................................................................................ 93
5.1. Summary and Interpretation of the Quantitative Findings ............................................. 93
5.2. Qualitative Findings and Possible Mechanisms of Peer Effects in Civic Engagement ... 96
5.2.1. Peer Effects Through Increased Perceived Benefits ............................................. 97
5.2.2. Peer Effects Through Changing Perceived Costs ............................................... 101
5.2.3. Why Sometimes Peers Fail to Affect an Individual’s Civic Engagement? .......... 105
5.2.4. Summary ........................................................................................................... 109
Chapter 6 Conclusion ............................................................................................................. 111
6.1. Summary of Findings ................................................................................................. 111
6.2. Implications ............................................................................................................... 113
vi
6.3. Limitations ................................................................................................................. 114
6.4. Future Research ......................................................................................................... 117
References .............................................................................................................................. 120
Appendix A: Descriptive Statistics for Background Variables................................................. 133
Appendix B: Descriptive Statistics for Civic Engagement Variables from the First Wave of
Survey in Fall 2018 ......................................................................................................... 134
Appendix C: Descriptive Statistics for Civic Engagement Variables from the Second Wave of
Survey in Spring 2019 ..................................................................................................... 137
Appendix D: List of Interviews ............................................................................................... 140
Appendix E: Survey Questionnaire for Participants in China .................................................. 141
Appendix F: Survey Questionnaire for Participants in the U.S. ............................................... 147
Appendix G: In-depth Semi-structured Interview Question (Guideline) .................................. 153
vii
List of Tables
Table 1 The 19 Core Indicators of Engagement ........................................................................ 28
Table 2 The Modified 17 Core Indicators of Civic Engagement ................................................ 30
Table 3 Data Collection Summary for the Quasi-experiment..................................................... 58
Table 4 Interviewee Information ............................................................................................... 60
Table 5 Descriptive Statistics for the Four Composite Variables of Civic Engagement ............. 74
Table 6 Univariate Regression of One's Own Civic Engagement on Average Civic
Engagement of Roommates in the First Wave of Survey ................................................... 76
Table 7 Peer Effects on Overall Civic Engagement Among Chinese Participants ...................... 79
Table 8 Peer Effects on Overall Civic Engagement Among American Participants ................... 80
Table 9 Peer Effects on the Civic Participation Indicators of Civic Engagement Among
Chinese Participants .......................................................................................................... 83
Table 10 Peer Effects on the Civic Participation Indicators of Civic Engagement Among
American Participants ....................................................................................................... 84
Table 11 Peer Effects on the Political Voice Indicators of Civic Engagement Among
Chinese Participants .......................................................................................................... 86
Table 12 Peer Effects on the Political Voice Indicators of Civic Engagement Among
American Participants ....................................................................................................... 87
Table 13 Peer Effects on the Formal Political Indicators of Civic Engagement Among
Chinese Participants .......................................................................................................... 89
Table 14 Peer Effects on the Formal Political Indicators of Civic Engagement Among
American Participants ....................................................................................................... 90
Table 15 Statistical Significance of the Quasi-experiment Results ............................................ 91
Table 16 Summary of Cited Interviews ..................................................................................... 96
viii
List of Figures
Figure 1 The Socio-ecological Model ....................................................................................... 11
ix
Abstract
Active civic engagement is the cornerstone of democracy, or any other desirable forms of
political systems. Understanding the determinants of civic engagement is both important for
theory construction and crucial to promoting civic engagement in real life.
Extant theories on civic engagement have proposed a number of factors that are
associated with civic engagement such as macro-level attributes, including political institutions,
civic culture, and economic development, that outline the general environment of civic
engagement. There is also an abundance of research that focuses on micro-level attributes such
as age, gender, political efficacy, and political knowledge that account for individual variation in
civic engagement. However, politics, including citizens’ participation in politics, rarely happens
independently. This dissertation explores the relatively under researched social network
determinants of civic engagement from the perspective of the influence of peers on individual
civic engagement with a focus on the causal mechanism. This study examines the following
research questions: 1) Do one’s peers affect her civic engagement behaviors? 2) Do peers with
high levels of civic engagement increase one’s own civic engagement? 3) Are peer effects
uniform across all forms of civic engagement? 4) Do peer effects differ in different institutional
settings? In addition, the current study also explores possible mechanisms of peer effects.
This dissertation employs a nested research design that combines a large-N quasi-
experiment and a series of semi-structured in-depth interviews to explore and demonstrate the
causal relations between one’s civic engagement and that of their peers’. The quasi-experiment
takes advantage of random roommate assignment in some colleges to eliminate the correlated
effects due to the self-selection into peer groups. Moreover, this dissertation adopts a most
different systems comparative design to demonstrate the robustness of peer effects in different
x
institutional settings. Participants are recruited among first-year college students who live in
dorms with randomly assigned roommates in China and the United States, two countries with
significantly different political institutions.
The results from the quasi-experiment demonstrate the causality of peer effects in civic
engagement. Having randomly assigned roommates with high levels of civic engagement
increases an individual’s own overall civic engagement for both Chinese and American
participants. However, peer effects are not uniform across all forms of civic engagement, which
is often caused by the joint influence of institutional, contextual, and individualistic factors.
Exploring the peer effects in civic engagement provides new approaches for promoting
civic engagement. Unlike the macro-level institutional factors and micro-level personal attributes
that are often either impossible to manipulate or immutable, peer groups provide a more feasible
and flexible way to cultivate civic engagement. The dynamic and information-laden nature of
civic engagement makes it a crucial supplement to formal politics. Understanding peer effects in
civic engagement could potentially contribute to a healthier and more robust political life.
1
Chapter 1 Introduction
Civic engagement has long been one of the focal points of political science research as it
represents the most important ways an individual can provide input to the public realm, both the
civil society and the formal political system. It provides the opportunity for private citizens to
shape their communities, their countries, and even the world.
The importance of understanding civic engagement is two-fold. First of all, civic
engagement is crucial for a health and functional democracy as it is the most important source of
formal and informal political input to the democratic system from the citizens. On the on hand,
the formal electoral outcomes would not be truly democratic, without unbiased and
representative participation in elections. On the other hand, other forms of civic engagement,
while not as direct or efficient as formal electoral participation, provide invaluable influences
over political and public decisions. Formal electoral political participation, even if unbiased and
representative, is not sufficient for a truly healthy democracy. Other forms of civic engagement
provide inclusive ways for all citizens to have their voices and needs heard, protected, and
advanced (Dahl 1998). The dynamic and information-laden nature of civic engagement makes it
a crucial element of a healthy democracy.
Second, civic engagement is also pivotal for non-democracies. While scholars have been
debating whether active civic engagement leads to democratic transitions (e.g. Tismăneanu 1990,
Putnam et al. 1993, Brysk 2000, Skocpol 2003, Fung 2003), stronger control of the non-
democratic state (e.g. Unger and Chan 1997, Howell 2000), or better public goods and services
provision (e.g. Tsai, L. 2007, O’Brien and Li 2006, , Spires 2011, Hildebrandt 2013), these
studies collectively demonstrate that civic engagement in non-democracies still serves as an
2
important channel where citizens can influence public and political affairs and have their voices
heard.
One crucial aspect in the study of civic engagement is to examine its determinants it
provides both scholars and practitioners ways to promote civic engagement. Extant literature has
proposed a list of “usual suspects” that have relatively stable and reliable correlations with civic
engagement, including education, income, political knowledge, political efficacy, sense of civic
duty, political culture, regime type, etc. (e.g. Putnam 2000, Zukin et al. 2006, Delli Carpini and
Keeter 1996, Verba et al. 1995, Uslaner 2004, Lake and Huckfeldt 1998, Pancer 2015) While
these usual suspects predict civic engagement fairly well, there is still a significant amount of
residual unexplained. Previous empirical research primarily focuses on examining and
demonstrating the relationship between micro-level individualistic and macro-level institutional
factors of civic engagement, leaving the social dimension of civic engagement under explored.
Yet, the social dimension could potentially provide valuable contribution to
understanding and promoting civic engagement. In terms of theory, examining the social
determinants of civic engagement contributes to constructing a more generalizable model of
civic engagement that goes beyond the constraints of national regime types or personal attributes.
Moreover, unlike macro-level institutional factors such as political regime and political culture
or micro-level individualistic factors such as age, education, and personal resources that are
essentially impossible or unethical to manipulate, the social dimension of civic engagement such
as peer groups offers a rare opportunity for practitioners to promote civic engagement through
altering one’s social context.
Therefore, this study aims at exploring the social dimension of civic engagement with a
focus on peers. This study bridges the gap in the study of civic engagement by examining its
3
social dimension within a more comprehensive and dynamic model that also considers macro-
level institutional and micro-level individualistic factors. This study proposes that the civic
engagement of one’s peers has causal effects on her own civic engagement and demonstrates it
with a quasi-experiment and a series of in-depth interviews. The following section provides an
overview of this study.
1.1. Overview of the Study
While prior research has demonstrated strong links between peers and individual
outcome in various academic fields (e.g. Christakis and Fowler 2007, Case and Katz 1991,
Imberman et al. 2012), peer effects still remain under-researched in political science. Among the
limited political research on peer effects, most are correlational. This study focuses on exploring
the causal relations between one’s civic engagement and that of their peers’. Specifically, the
research questions of this study are:
1) Do peers’ civic engagement or other characteristics have a causal effect on an
individual’s civic engagement?
2) If causal peer effects indeed exist, how do peers affect an individual’s civic
engagement? Does having peers who have high levels of civic engagement increase or
decrease one’s own civic engagement?
3) Are peer effects uniform across all forms of civic engagement?
4) Do peer effects differ in different institutional settings?
5) What are the mechanisms behind peer effects?
Drawing on the socio-ecological model (Bronfenbrenner 1977, Bronfenbrenner 1979),
this study situates peer effects on civic engagement in a comprehensive theoretical framework
that conceptually accounts for institutional, social, and personal factors. Building upon this broad
4
analytical foundation, I employ a modified utility function based on the rational choice theories
to map the impacts of peers on an individual’s civic engagement as an attempt to further explore
the mechanisms of peer effects.
This study adopts a nested most different systems comparative design to explore the peer
effects in civic engagement with a quasi-experiment and a series of in-depth interviews. Taking
advantage of random roommate assignment in colleges, the quasi-experiment eliminates the
correlated effects due to the self-selection into peer groups and examines the causal relations
between one’s own and her peer’s civic engagement. The most different systems comparative
design uses samples from China and the U.S., two countries with drastically different
institutional settings, to demonstrate the robustness of peer effects. Furthermore, this study
employs in-depth semi-structured interviews with selected participants of the quasi-experiment
to test the quantitative results of the large-N analysis. Chapter 3 offers more discussion on the
data and methodology of this study.
1.2. Organization of Chapters
The rest of this dissertation is organized into the following chapters. Chapter 2 proposes a
theoretical framework that builds upon the socio-ecological model and the rational choice
theories to examine the impacts of peer effects on civic engagement. Then, I review existing
literature on the explanatory variables of civic engagement and on exploring peer effects with
quasi-experimental designs. Chapter 2 also operationalizes the concept of civic engagement for
empirical analysis. Chapter 3 begins with a discussion of the research design that utilizes mixed
methods in a nested most different systems comparative analytical framework. I then report the
case selection criteria, the construction of variables, and data analysis approaches of this study.
Chapter 4 reports the results of the quasi-experiment, including summary statistics, peer effects
5
on overall civic engagement, and peer effects on specific forms of civic engagement for both
samples from the U.S. and China. Chapter 5 first summarizes and discusses the findings reported
in Chapter 4. Next, I propose possible mechanisms of peer effects drawing on the results from
the in-depth semi-structured interviews. As the conclusion of this study, Chapter 6 begins with a
brief summary of the findings. I then discuss the theoretical, methodological, and practical
implications of this study as well as some suggestions for future research.
5
6
Chapter 2 Theoretical Framework and Literature Review
Drawing on the socio-ecological model and the rational choice theory, this study
demonstrates the causal effects of peer groups on individual civic engagement and the
mechanism behind the peer effects. Conventional research on civic engagement tends to focus
either on institutional factors such as political regime, civic culture, and history that outline the
general environment of civic engagement or individualistic attributes such as age, gender, and
political efficacy that account for personal variation in civic engagement, leaving contextual
factors under-researched.
This chapter begins with a discussion of the socio-ecological model and the rational
choice theories and then introduces a theoretical framework that builds upon the strengths of
these two theories that is used to examine the causality and magnitude of peer effects on civic
engagement. Next, this study discusses the definition of civic engagement, reviews the
explanatory variables of civic engagement in existing literature, and provides an
operationalization of civic engagement for empirical analysis. Then, I provide an operational
definition of peer effects, review the link between peers and individual outcome, and discuss
causally exploring peer effects with quasi-experimental designs. Lastly, I conclude this chapter
with a discussion of the contribution of this study.
2.1. Theoretical Framework
Existing research from different disciplines proposes numerous theories that provide
valuable insight in explaining political behaviors. In particular, the socio-ecological theory
presents a way to incorporate a wide range of factors in the analysis of civic engagement. This is
especially valuable since most existing research on civic engagement primarily focuses on
7
individualistic attributes or macro level institutions and fails to examine civic engagement in its
overall context. Moreover, the rational choice theory and its modified versions offer an approach
to empirically test the impact of different factors on civic engagement decisions. In this section, I
first discuss these two theoretical approaches. Then, I build upon both the rational choice and the
socio-ecological theories and present a theoretical framework to examine peer effects in civic
engagement.
2.1.1. Rational Choice
As one of the most prominent schools of thought in social sciences, rational choice theory
offers a framework to model social behaviors. Rational choice theory argues that each individual
has preferences among the available options and seeks to maximize her net benefit through her
choice (Downs, 1957). Under the rational choice model, civic engagement should only occur
when its benefits outweigh its costs. Namely, a rational actor would not engage unless she could
gain positive net benefit from a give form of civic engagement.
Downs (1957) proposes a model with voting as an example where the probability that the
voter will turn out (V) equals the probability of her vote mattering (p) times the benefits of
voting (B) minuses the costs of voting (C). If one only considers direct benefits such as financial
gain and political power, civic engagement should be non-existent according to Downs’ version
of rational choice theory. Similar to voting, the likelihood of an individual’s civic engagement
mattering to the outcome is extremely low, making the benefits of civic engagement no different
than zero and the net benefit negative. In this case, apathy is the only rational choice for a
citizen. However, it is paradoxical that civic engagement does exist widely in the world.
In their effort to explain this paradox and improve the usefulness of rational choice
theory, scholars offer numerous adjustments, clarifications, and additions. Riker and Ordeshook
8
(1968) add another variable (D) to Downs’ model to represent the psychological and civic
benefits one receives from her action, such as the sense of fulfilling her civic duty. This new
variable adds to the benefits of civic engagement, making it possible for one to gain positive
benefits even when the probability of her civic engagement mattering is minuscule. The civic
duty variable also opens up the opportunity to explore the psychological aspects of civic
engagement actions and decisions, making rational choice theory more suitable for real-world
applications.
While Riker and Ordeshook’s addition to the model provides an explanation that
addresses the paradox of voting, their model still considers actions in the political world as
independent decisions made by atomized actors who are not connected to the overall social
structure. There are several modified versions of the rational choice theory that attempt to
incorporate factors beyond the individual level in order to provide more realistic explanations of
political behaviors in the real world. Shepsle (1989) offers a version of rational choice theory
that bridges individuals and the social structure in which they are embedded by introducing
institutional features. In Shepsle’s model, institutions shape individual behavior through
identifying eligible players, determining possible outcomes, deliberating alternative models, and
observing revealed preferences of eligible players. By adding the institutional features, the
adjusted rational choice model can explain why individuals sometimes deviate from their
economic interests in their civic engagement. More importantly, the addition of new
institutionalism allows the rational choice model to expand beyond the individualistic dimension
and incorporate macro-level factors to explain civic engagement.
Other than individual and institutional factors, social factors can also shape civic
engagement under the rational choice model. Wilson (2011) argues that the measurement of
9
individual preferences should include social preferences and that individual behavior consistently
violates predictions made by traditional rational choice models because the traditional rational
choice models fail to include interpersonal comparisons. In addition to interpersonal comparison,
Whiteley and Seyd (1996, 1998) further expand the social features of rational choice model by
incorporating socio-psychological models of participation. Their model includes not only costs,
benefits, and sense of duty, but also social norms and expressive attachments. Like Olson (1965)
argues in The Logic of Collective Action: Public Goods and the Theory of Groups, Whiteley and
Seyd (1996, 1998) also expand the concept of benefits by arguing that there are two distinct
types of benefits: collective benefits that are available to all and selective benefits that are
reserved only for those who participate. While collective benefits are rarely enough to
incentivize participation as demonstrated by traditional rational choice theory and empirical
evidence, selective benefits provide additional incentives for active civic engagement. In this
model, one’s decision regarding civic engagement is shaped by the social norms of people in
their social network. Namely, if a citizen’s family members or close friends highly value civic
engagement, she would also be more likely to engage in civic activities. Moreover, expressive
attachments add to one’s incentives for participation. A citizen is more likely to have a higher
level of civic engagement on behalf of a social group if she has strong attachments to the group
in question.
The modified versions of the rational choice theory provide a way to incorporate a wider
range of explanatory variables of civic engagement. By including both individual and
institutional factors in an individual’s utility function, the recent development in rational choice
theory to a certain extent addresses one of its most common critiques on the unrealistic
assumption that social behavior is the result of the cost-benefit calculation of atomized rational
10
actors without context. In addition, Wilson’s (2011) introduction of interpersonal comparison
and Whiteley and Seyd’s (1996) inclusion of group attachment and group social norms in the
rational choice theory also provide a roadmap of including contextual and social actors in the
model. With these modifications, the model rational choice theory gains more explanatory power
and conceptual flexibility, extending its application from abstract conceptual discussion to
empirical analysis on behaviors that take place in the complexity of the real world. However,
while the modern modifications of the rational choice theory incorporate social and institutional
factors as variables in an individual’s utility function, it still lacks a conceptual framework that
systematically explains how individual, social, and institutional factors shape an individual’s
behavior.
2.1.2. Socio-ecological Model
Bronfenbrenner proposes the socio-ecological model that emphasizes the important role
of environmental factors in human development (Bronfenbrenner 1977, Bronfenbrenner 1979,
Bronfenbrenner and Ceci 1994, Bronfenbrenner and Morris 2006), providing framework that
examines individual attributes and various environmental factors systematically. Challenging the
the prevailing trend that focuses solely on isolated individuals in the field of psychology at the
time, Bronfenbrenner emphasizes the role of an individual’s context argues that human
development is determined by both internal and external factors as well as their interaction.
Therefore, the entire ecological system of a person needs to be considered in order to truly
understand human development. In this model, one’s socio-ecological system consists of four
subsystems: microsystem, mesosystem, exosystem, and macrosystem (Bronfenbrenner 1977).
These four subsystems are nested with the individual in the center, like a Russian nesting doll
(See Figure 1). In addition, Bronfenbrenner later includes chronosystem, a subsystem that
11
represents the dimension of time in the bioecological model, an adapted version of socio-
ecological system (Bronfenbrenner 1994).
Figure 1. The Socio-ecological Model
Building upon the original socio-ecological model, Bronfenbrenner and coauthors
(Bronfenbrenner and Ceci 1994, Bronfenbrenner and Morris 1998, 2006) incorporate the
processes of human development in the Process-Person-Context-Time framework (PPCT). In
this framework, Bronfenbrenner and Morris (1998, p. 572) offer two theoretically interdependent
propositions. The first proposition states the nature of proximal processes: “human development
takes place through processes of progressively more complex reciprocal interactions between an
active, evolving biopsychological human organism and the persons, objects and symbols in its
Macrosystem
(cultural norms,
institutions)
Exosystem
(media, social programs,
public policies)
Mesosystem
(interactions of
microsystems)
Microsystem
(family, peers)
Individual
(behaviors,
attitudes)
12
immediate environment”. Bronfenbrenner and Morris (1998) further emphasize that proximal
processes have to occur “on a fairly regular basis over extended periods of time” in order to be
effective. The second proposition suggests that the impacts of one’s personal characteristics, her
context, and time jointly and simultaneously affect proximal processes. The PPCT framework
allows researchers to examine both proximal processes as well as the internal and contextual
factors that shape proximal processes , providing a suitable structure for empirical investigations
based on the socio-ecological model.
The PPCT framework consists of four key elements, namely, proximal processes, person,
context, and time. Processes, or proximal processes, are the reciprocal interactions between
individuals and their environment, as well as other individuals and objects in their environment.
Bronfenbrenner and coauthors see proximal processes as the foundation of the PPCT framework
because they serve as the engines of human development (Bronfenbrenner and Morris 1998).
Through proximal processes, or reciprocal interactions with everything in their environment,
individuals make sense of their environment, understand themselves in relation to their
environment, and adapt their behaviors and attitudes.
Persons are where the proximal processes begin. A person possesses three types of
characteristics, including demand characteristics, resource characteristics, and force
characteristics (Bronfenbrenner and Morris 1998). Demand characteristics are a person’s most
visible and recognizable attributes that trigger the reactions of other persons in their environment
and affect proximal processes, such as gender, age, skin color, facial expressions, and physical
appearance. Resource characteristics are a person’s physical, mental, social, and material
resources that she brings to proximal processes, such as physical strength, intelligence, life
experiences, relationship with friends, and housing. Force characteristics are a person’s
13
personality traits that can influence proximal processes, such as temperament, efficacy, and
persistence. These three types of characteristics jointly influence when and how a person
interacts with her environment.
Context, or environment, is where proximal processes take place. This element in the
PPCT framework refers to the four nested subsystems in the original socio-ecological model,
namely, the microsystem, the mesosystem, the exosystem, and the macrosystem. The
microsystem, as the inner most nesting doll, is the closest subsystem to an individual and
consists of structures she directly contacts. This subsystem includes relationships and
interactions in an individual’s immediate surroundings such as family and peers (Bronfenbrenner
1977). The interactions and connections of structures in an individual’s microsystem occur in her
mesosystem. The subsystem that is in the middle layer of the socio-ecological system is the
exosystem, which contains larger social systems that do not affect an individual directly, such as
public policies, media, and social programs (Bronfenbrenner 1977). The macrosystem, which is
the outermost layer, contains institutional patterns such as cultural norms (Bronfenbrenner 1977).
Elements in the macrosystem have cascading impacts on all the previously discussed subsystems
as they are the “concrete manifestations” of the macrosystem (Bronfenbrenner 1977). Each of the
four contextual subsystems affects proximal processes within itself as well as in other
subsystems.
Time, an element added to the original socio-ecological model, is a crucial factor in
human development. Bronfenbrenner and Morris (2006) categorize time into three types in
association with the socio-ecological model: microtime refers to what is happening in the course
of proximal processes, mesotime is microtime proximal processes that recur with some degree of
consistency over the course of weeks or months, and macrotime focuses on substantial changes
14
in the society over extended periods of time over one’s life course and across generations. Time
is an essential element as it influences all aspects of the PPCT framework.
As a theoretical model, the socio-ecological model is designed to be conceptually broad
and inclusive to allow for greater generalizability and wider applications in different topics (Xia,
Li, and Tudge, 2020). However, this conceptual flexibility poses a challenge for empirical
research projects as the socio-ecological model essentially encompasses all influences and
interactions that are related to a given individual, which are obviously impossible to exhaust in
the operationalization of the model. Furthermore, while the socio-ecological model provides an
excellent way to explain human behaviors, the conceptual model itself does not include an
empirical testing component. Therefore, in order to apply the socio-ecological model to explore
behaviors such as civic engagement, it is necessary to first operationalize the model by
specifying the elements of interest in the subsystems. In addition, the conceptual framework of
the socio-ecological model needs to be paired with a model that is suitable for empirical testing
in order to examine the specific impact of each element on civic engagement.
2.1.3. Theoretical Framework of Peer Effects in Civic Engagement
As discussed in the previous two subsections, the socio-ecological model provides a
conceptual framework to explore the impact of peer effects on civic engagement in a
comprehensive context. And the modified rational choice models collectively incorporate
individual, social, and institutional factors in an individual’s utility function, making it possible
to empirically measure peer effects on individual civic engagement. Therefore, this study draws
on the theoretical strengths of both the socio-ecological model and the rational choice theories
and construct a theoretical framework to analyze peer effects in civic engagement. In this
theoretical framework, I utilize the socio-ecological model to construct a broad analytical
15
foundation for this study and propose a modified utility equation based on the rational choice
theories to map and measure the impacts of various factors on an individual’s civic engagement.
Since this study focuses on examining the impact of peer effects on individual civic
engagement in a comprehensive context instead of attempting to propose or test a comprehensive
model that explains every aspect of civic participation, it is necessary to first operationalize the
socio-ecological model in a way that narrows it down to suit the needs and scope of this project.
An empirical study employing the socio-ecological model needs to specify the developing
persons of interest, the developing outcome of interest, the factors from the four contextual
subsystems pertaining to the developing persons and outcome of interest, and the proximal
processes that impact the aforementioned elements.
For this study, the developing persons of interest are citizens who have the potential to
participate in civic engagement. The developing outcome of interest is the level of civic
engagement of an individual, which can be further divided into different forms of civic
engagement ranging from volunteering to running for political office. And among the four
contextual systems, this study primarily focuses on the mesosystem as it is where peer
interaction takes place. The main proximal process that impacts peer interaction in the
mesosystem is the contagion mechanism where the interaction among peers leads to the
transmission of attitudes and behaviors. In this process, peer norms and behaviors transmit and
get reinforced among members of peer groups. In a peer group where members actively
participate in the civil society and in political life and consider such participation as rewarding
and beneficial, an individual will acquire similar behaviors and attitudes through her exposure to
her peers.
16
In order to empirically examine the impact of peers on civic engagement, I propose an
adjusted utility function of civic engagement drawing on the different modifications of the
rational choice theory discussed in 2.1.1. The essence of the rational choice theory is that human
behaviors result from an individual’s cost-benefit analyses, namely, the two key elements in the
utility function are perceived costs and benefits.
The benefits can be further categorized into three groups: universal benefits that are
available to all, selective benefits that are restricted to members of a group or a movement, and
personal benefits that apply directly to an individual. Among the aforementioned three types of
benefits, universal benefits are constant and therefore tend not to play a major role in influencing
an individual’s decision. Yet, selective benefits and personal benefits are exclusive to individuals
who participate in civic and political life, and thus potentially have stronger influence on civic
engagement. For instance, joining a political party grants an individual access to selective
benefits like career opportunities in politics and networking resources. Personal benefits
encompass the sense of civic duty variable proposed by Riker and Ordeshook (1968) as it is one
form of personal psychological benefits.
Unlike Dawns (1957) original utility function for voting behavior, the costs of civic
engagement are more complex. An individual perceives the costs of civic engagement in three
ways. The first type of costs is fixed, namely, it always occurs when an individual participates in
a given form of civic engagement. Taking protesting as an example, fixed costs are things like
the time one needs to spend protesting and the effort required for transportation. These costs
always occur and apply to every participant, and thus tends not to have any major impact on
one’s decision. The second type of costs, on the other hand, is uncertain. Uncertain costs do not
necessarily occur when an individual participate in a given form of civic engagement. Yet, the
17
probability of uncertain costs occurring is always above zero. Examples of uncertain costs for
participating in a protest include getting hurt if the protest turns violent, being arrested by the
authorities, and lasting consequences such as losing career opportunities due to arrest records.
Unlike fixed costs, the calculation of uncertain costs is probabilistic, namely, the perceived
uncertain costs of a given form of civic engagement is the product of the perceived probability
and the perceived cost. The third type of costs is different from the first two types of costs in
natural as it is the perceived costs of not participating. In the protest example, an individual can
choose not to participate to avoid both the fixed and uncertain costs, but she would then have to
face the third type of costs. For instance, she might be considered as a social outcast in her peer
group for not confirming to the peer group norm. Or, she might feel inferior when she compares
herself to her peers and suffer from psychological pain. Therefore, this study proposes a
modified utility function of civic engagement:
𝐶𝐸 =𝐵
%
+ 𝐵
(
+ 𝐵
)
− 𝐶
%
− + 𝑝
-
∗ 𝐶
(-
+ 𝐶
)
In this formula, vector B1 represents universal benefits that are available to all, vector B2
represents selective benefits that are exclusive for group members or participants, vector B3
represents personal benefits that directly apply to the individual, vector C1 represents the fixed
costs of participating in a given form of civic engagement, C2i represents each of the uncertain
costs of participating in a given form of civic engagement, pi represents the perceived probability
of a given uncertain cost occurs, and vector C3 represents the costs of not participating in a given
form of civic engagement. If CE > 0, an individual decides to participate in a given form of civic
engagement.
Combining socio-ecological model and the modified utility equation of civic
engagement, this study presents a theoretical framework to explore peer effects in civic
18
engagement. Key elements from different subsystems in the socio-ecological model defines and
affects the variables in the modified utility function. In the macrosystem, political institutions
determine the possible values for B2 and p * C2. For instance, if an individual chooses to join a
major political party in the United States, her B2 primarily consists of the benefit of being able to
cast a vote in that party’s primary election. Yet, if an individual to join a major political party in
China where the only one major political party is also the ruling party, her B2 may include access
to exclusive educational resources and job opportunities, particularly in the public sector. As for
p * C2, if one chooses to participate in a peaceful protest against the leader of a country, her C2 is
likely to include getting hurt by accident, being arrested by the authorities, and suffering from
lasting consequences due to the record of her participation. The probabilities for the latter two of
these costs are usually so low that they are negligible in a country with democratic political
institutions, whereas they can be very high in an autocracy. Political culture affects the possible
values for B3 and C3. For instance, in a society with a political culture that promotes civic
engagement, an individual gains B3 as the sense of civic duty when she votes in an election. In a
society with a political culture of collectivism, an individual gains B3 as the sense of belonging if
she participates in civic engagement and faces C3 in the forms of being seen as a social outcast if
she does not. In the mesosystem, peer groups influence B2, B3, C2, and C2. A peer group where
members actively participate in civic engagement increases an individual’s B2 and B3 through
shared group identity, norms, and the psychological benefits of feeling in harmony with one’s
peers if she also has active civic engagement. In the meantime, in doing so, she can avoid the
penalty of C3 and lower her perceived p * C2 based on her observation of the outcomes of peers
who actively participate in civic and political life.
19
2.2. Civic Engagement
In this section, I first define civic engagement for this study. I then examine the
explanatory variables of civic engagement in existing literature in order to map out the current
landscape of research on civic engagement. Lastly, I provide the operationalization for the
concept of civic engagement in this study.
2.2.1. Defining Civic Engagement
The concept of civic engagement, oftentimes also referred to as civic participation or
social engagement, has attracted immense interest from both the academia and elsewhere,
especially since Putnam popularized it by presenting the positive association between active
civic engagement and a well-functioning democracy (Putnam et al. 1993). Due to its popular and
multi-disciplinary nature, it is very difficult to find a universally agreed-upon definition for the
concept “civic engagement”. This subsection begins with examining existing definitions of civic
engagement, followed by the definition used in this study.
For those who are interested in civic engagement, the lack of a universally agree-upon
definition for the term is nothing surprising. In fact, scholars and practitioners have been defining
and operationalizing these concepts according to their own interests and agendas, which results
in a wide range of definitions (Adler and Goggin 2005). Although scholars started devoting more
efforts in conceptualizing civic engagement since the 1990s, especially after Cohen and Arato’s
theoretical monograph (1992), they have yet to come up with a widely agreed upon definition for
civic engagement. Some would go as far as arguing that civic engagement is an umbrella term
overly used by scholars and media to the point where it has lost its theoretical utility and
therefore is ready for the dustbin (Berger 2009). But does the broadness and blurriness of civic
engagement mean it is meaningless and unsuitable for conceptualization or operationalization?
20
Upon closer examination, the seemingly nebulous term “civic engagement” has two
distinctive aspects. The first aspect of civic engagement focuses on philanthropy, volunteerism,
and charity, which is typically adopted by nonprofit scholars, sociologists, and civil society
organizations. For instance, Diller (2001) defines civic engagement as “an individual’s duty to
embrace the responsibilities of citizenship with the obligation to actively participate, alone or in
concert with others, in volunteer service activities that strengthen the local community”. This
aspect of civic engagement captures the interests of the Independent Sector (also known as the
Voluntary Sector or the Civic Sector) and often emphasizes the local community.
The second aspect of civic engagement focuses on political participation, which is more
common in political science. Unlike the first aspect of civic engagement, the second aspect of
civic participation places no or very little emphasis on voluntary services or local communities.
Instead, it focuses more on the public nature of civic engagement. This aspect of civic
engagement involves influencing political processes, the civil society, and public life through
citizens’ participation in the public sphere (Diller 2001, Ronan 2004). To a certain extant, this
aspect of civic engagement is very similar to Brady’s (1999) definition of political participation:
“[political participation is] action by ordinary citizens directed toward influencing some political
outcomes”.
Other than the definitions that focus on either aspect of civic engagement, there are also a
significant number of broad definitions that aim at providing a more comprehensive
conceptualization. While not offering an explicit definition of civic participation or civic
engagement, Putnam (1993) describes civic engagement in a very broad term that includes
activities ranging from visiting friends and playing cards to voting and running for office in his
seminal work Bowling Alone. Cnaan and Park (2016) defines civic engagement as “any activity
21
of any individual, alone or with others, that is performed outside the boundaries of the family and
household that directly or indirectly attempts to promote the quality of life of others, and that
may make the community or society a better place to live ”, which covers a wide range of formal
and informal, political and non-political activities. Diller (2001) offers an even more expansive
conceptualization of civic engagement that defines it as “all activity related to personal and
societal enhancement which results in improved human connection and human condition”.
Among the definitions of civic engagement discussed above, some are specific yet fall
short of capturing the bigger picture, whereas others are too expansive that they truly are what
Berger (2009) describes as an all-encompassing buzzword. This study leans toward a broader
definition of civic engagement in the sense that both aspects discussed above should be included
in the conceptualization. As argued by Youniss and coauthors (2002), there shouldn’t be a
“definite demarcation” between civil and political participation in the definition of civic
engagement. Rather, civic engagement should be a continuum that ranges from formal political
activities such as voting and running for office to non-political actions and attitudes regarding
issues of public concern. In the meantime, it is also important to avoid “conceptual stretching”
when defining civic engagement. A ideal definition of civic engagement should be broad enough
to cover both the community and political aspects of the concept and specific enough to exclude
generic social behaviors such as playing card games that are not of public concern.
Therefore, this study adopts Adler and Goggin’s (2005) definition of civic engagement:
“Civic engagement describes how an active citizen participates in the life of a community in
order to improve conditions for others or to help shape the community’s future”. In this
definition, Adler and Goggin (2005) conceptualize civic engagement along two dimensions
where the first dimension spans from individual to collective activities and the second from non-
22
political to political activities, creating a conceptual space that allows researchers to explore the
various interrelated forms of civic engagement in different settings.
2.2.2. Explanatory Variables of Civic Engagement
While this study focuses on exploring the influence of peers on civic engagement and
does not attempt to examine and explain every aspect of civic engagement, it is important to first
understand the current state of the research on the determinants of civic engagement. Over the
past decades, scholars have detailed a list of factors that can “do a pretty good job” (Campbell
2013) predicting the civic engagement of individual citizens. In the context of the socio-
ecological model, these “usual suspects” can be divided into three categories based on their
scale: individualistic demographic and psychological factors on the micro level, institutional
factors on the macro level, and social network contextual factors on the contextual level.
2.2.2.1. Individualistic Factors
Among the three categories of explanatory variables, individualistic factors have received
significant scholarly attention since the rise of the Michigan school and large-scale survey
method. Individualistic factors that influence civic engagement can be divided into two groups,
demographic factors and psychological factors. The most reliable demographic factors of civic
participation include socioeconomic status (SES), age, gender, race and ethnicity, religious
affiliation, etc. Among this group of factors, even all variables argued to have correlation with
civic engagement behavior, SES factors are usually seen as the ones that perform the best, just
like other forms of political participation. An overly simplistic summary of the effect of SES on
political participation would be that individuals with higher SES are more likely to participate in
politics and public affairs (Verba and Nie 1972, Wolfinger and Rosenstone 1980, Verba et al
1995).
23
Unlike the almost universally agreed upon effect of SES, the role of other known
demographic factors varies across different contexts. For example, while a number of empirical
studies demonstrate that there is gender difference in political and civic participation (e.g.
Inglehart and Norris 2003), the exact effect of gender on civic engagement behavior differs in
different issue areas and contexts. While demographic factors present a relatively clear
correlation and have better external validity, they have two disadvantages in terms of real world
implications. Firstly, the causal relation between demographic factors and civic engagement
behavior is relatively under explored. Therefore, it would be difficult to use these robust
correlations to promote civic engagement in real life. Secondly, demographic factors tend to be
either immutable or very difficult and costly to manipulate. Thus, while demographic factors are
helpful for predicting who participates in civil society, they provide limited insight for promoting
civic engagement.
The other group of individualistic factors have their root in psychological studies.
Political attentiveness and political knowledge, strength of preference, political efficacy, sense of
civic duty, and mobilization are all explanatory variables of civic participation (Campbell 2013,
Zukin et al. 2006). Psychological factors demonstrate a greater potential to detail the causality
between them and civic engagement since they are easier to manipulate in an experimental
setting. However, in comparison to demographic factors, psychological factors are more difficult
to conceptualize, operationalize, and measure. Meanwhile, the effect of psychological factors is
sometimes harder to generalize to a setting with different social and civic norms. For instance,
while the positive effect on civic engagement by priming the sense of civic duty has been
demonstrated in multiple empirical research projects (e.g. Gerber et al. 2008), there are still
24
obstacles to apply this finding in a society with different norms, civic and political education,
and political institutions.
These two types of individualistic factors contribute significantly on understanding the
pattern of civic engagement. Yet, these individualistic factors are only part of the whole picture.
Interactions with one’s context in the microsystem and mesosystem as well as institutions in the
macrosystem also have considerable influence on civic engagement.
2.2.2.2. Institutional Factors
Institutional factors in the macrosystem, both formal political and informal institutions,
are also relatively well explored and perform reliably in predicting and explaining civic
engagement. These factors tend to work on an aggregate scale and offer insights on a state’s
overall civic engagement. Political regime, history of democracy, and political institutions all
play a significant role in determining the level of civic engagement of a country or society
(Barrett and Zani 2015). These macro-level factors not only provide the frame within which
individuals make their decisions regarding civic engagement, but also shape the costs and
benefits of individual civic engagement through determining the possible outcomes. For
instance, the average level of civic engagement is substantially lower in non-democratic
countries than democratic ones (Cavatorta 2012) as the average costs of civic engagement are
high and the benefits are low in non-democratic countries. Among democracies, different
structures of political institutions in democracies also lead to variations in civic engagement. For
instance, high-cost civic engagement such as demonstrations is more prevalent in more
centralized and stronger democratic states, where as low-cost civic engagement such as signing
petitions has higher rates in more decentralized and weaker states (Kriesi et al 1995).
25
Other than formal political institutions, historical and cultural factors of a given state also
influences civic engagement. The maturity of the democratic institutions also affects
participation in civil society. For instance, citizens in sophisticated Western democracies tend to
be more likely to participate. Yet, while states with a longer and more stable history of
democracy tend to have higher levels of civic engagement, newly democratized states where
civil society played a significant role during the democratic transition, such as some of the post-
communist Eastern European countries, South Korea, and Taiwan, tend to have higher levels of
civic engagement (Bernhagen and Marsh 2007, Armstrong 2007, Ho 2012). In addition, political
culture also shapes civic engagement to a considerable extent. In states with strong associational
culture, such as the United States, individuals are more likely to acquire civic skills, develop
political interest, and thus participate more (Verba 1978, Putnam 2000). Moreover, other cultural
differences such as dominant state religion, individualism, welfare state tradition also correlate
with aggregate level of civic engagement in a country (Berry et al 2002, Inglehart and Norris
2003, Flanagan and Campbell 2003).
Institutional factors are almost impossible to manipulate in comparison to individualistic
factors. Yet, these macro-level factors account for a significant portion of the variation in
individual citizen’s civic engagement. More importantly, they shape an individual’s utility
function of civic engagement by determining possible outcomes, costs, and benefits. Therefore,
it is important to acknowledge and discuss institutional factors in studies that aim at
understanding civic engagement.
2.2.2.3. Contextual Factors
In addition to individualistic and institutional factors, contextual factors constitute
another important, yet relatively under researched dimension of civic engagement. While the
26
research on explanatory variable or determinants of civic and political participation seems to lean
more towards the individualist factors in recent years, the contextual dimension has long been
part of the research agenda (Campbell 2013). The contextual dimension of civic engagement not
only includes unique social and contextual factors, but also the social aspect of some seemingly
individualistic factors. For instance, while SES, one of the most robust factors of civic
engagement, is measured on an individual level, it essentially measures one’s condition in
comparison to other members of the society (Campbell 2013). Moreover, social factors such as
family and parental influence also effects individualistic factors including one’s SES, political
knowledge, political efficacy, and trust in political institutions (Azevedo and Menezes 2007).
Among the literature that examines the contextual dimension of civic engagement, there
has been a considerable amount of literature focusing on exploring the role of social network
(e.g. Lin et al. 2001, Kavanaugh 2002, Metzger et al. 2014, Choi, Nah, and Chung 2021).
Current literature also demonstrates robust correlations between civic engagement and contextual
factors such as political socialization, civic practices in classrooms and workplace, interpersonal
relations, as well as peer groups (e.g. Huckfeldt & Sprague 1995, Putnam 2001, Jennings 2002,
Zukin et al. 2006, Stolle 2007). For instance, Putnam’s work emphasizes heavily on the social
dimension of civic engagement by addressing social capital and social network in a society
(Putnam 2000, 2001). Moreover, individuals whose parents have strong interest in political and
civic issues are more likely to participate themselves (Schulz et al. 2010). Citizens who work or
study at places with more democratic decision-making process also tend to participate in political
and civic activities at a higher rate (Sobel 1993, Greenberg 2008).
In this body of scholarship, scholars have also explored the role of peers in civic
engagement with a focus on the civic socialization of adolescents and young adults. Since young
27
people are on a stage of life where they actively explore the world and themselves, civic
socialization tends to shape and change their civic behaviors and attitudes. Youth become more
similar to their peers in terms of civic engagement behavior and attitude through a series of peer
civic socialization process (Wegemer 2021). For instance, Kandel (1978) demonstrates that
adolescents are likely to adopt similar political attitudes and political ideologies as their friends.
Moreover, adolescents and young adults tend to mimic the civic engagement behaviors of their
friends through the mechanisms of the role model effect (Gordon and Taft 2011). Terriquez and
coauthors (2020) find that peer civic socialization fosters active civic engagement among Latinx
youth. A number of research projects have also demonstrated the strong link between peers’
civic engagement behavior and attitude through frequent civic talk and political discussions (e.g.
Nisbet and Scheufele 2004, Diemer and Li 2011, Klofstad 2011, Hsieh and Li 2014). In addition,
Ryan (2000) argues that the way peers influence one another’s behaviors and attitudes is not
always the same: behaviors and attitudes that are positively received within a given peer group
tend to have a mutual reinforcing effect while behaviors and attitudes that are negatively
received are not likely to reoccur. While this body of research provides promising perspectives in
investigating the role of peers in civic engagement, it tends to focus on a subset of civic
engagement activities that centered on civic participation such as volunteerism. Meanwhile, like
many research projects that explore peer effects, causality is rarely established between peers and
an individual’s behavioral or attitudinal change.
Unlike the macro-level characteristics of a society or the individual attributes of a citizen,
the social network contextual factors tend to be blurrier in terms of conceptual boundaries and
affect civic engagement through a more complicated mechanism. But contextual factors,
including peer groups, have the potential to contribute to a more comprehensive causal model of
28
civic engagement in the sense that they provide a valuable way to explore the mechanisms
through which the external world interacts with an individual’s preferences and shapes her civic
engagement.
2.2.3. Operationalizing Civic Engagement
Following the previous subsections that discusses the definition of civic engagement and
its usual explanatory variables, this subsection focuses on operationalizing the concept of civic
engagement in order to explore civic engagement through empirical research.
In their 2002 report The Civic and Political Health of the Nation, Keeter and coauthors
propose 19 “core indicators of engagement” (see Table 1) and use them to measure the civic
engagement among American adults. These indicators are divided into three primary groups to
capture different forms of civic engagement: 1) civic indicators that measure activities such as
volunteering, and community fundraising, 2) electoral indicators that measure formal political
participation such as voting in an electoral democracy, and 3) indicators of political voice that
measure influence in political and public issues.
Table 1. The 19 Core Indicators of Engagement
Civic Indicators
Community problem solving
Regular volunteering for a non-electoral organization
Active membership in a group or association
Participation in fund-raising run/walk/ride
Other fund raising for charity
Electoral Indicators
Regular voting
Persuading others
Displaying buttons, signs, stickers
Campaign contributions
Volunteering for candidate or political organizations
Indicators of Political Voice
Contacting officials
Contacting the print media
29
Contacting the broadcast media
Protesting
E-mail petitions
Written petitions
Boycotting
Canvassing
Source: Keeter et al. (2002)
While the 19 specific indicators do not cover all forms of civic engagement, they provide
an effective framework to measure civic engagement. The three categories of indicators not only
capture the diversity of civic engagement, but also allow comparisons between and within
different aspects of civic engagement. With this categorization, researchers can examine various
forms of civic engagement according to their own interests and agenda. However, Keeter and
coauthors’ (2002) indicators of civic engagement have two shortcomings. Firstly, since the
indicators are designed for measuring civic engagement of adults in America, a sophisticated
democracy, some of the indicators are unfit to measure civic engagement in non-democracies.
For instance, volunteering for political candidates and making campaign contributions are not
applicable to Chinese citizens because there is not free election to begin with. Secondly, the 19
indicators proposed in 2002 does not include some new forms of civic engagement. For instance,
in an age of rapidly developing information technology, the role of virtual civic engagement
online has been increasingly important.
Therefore, this study adopts Keeter and coauthors’ operationalization of civic
engagement (2002) with some adaptations so that the measurement is suitable for both
democratic and authoritarian regimes. Among the three categories of indicators, this study
replaces the electoral indicators with formal political indicators in order to measure civic
engagement in societies beyond electoral democracies. The specific formal political indicators
30
depend on what formal political participation entails in a given setting. Instead of using
volunteering for political candidates and making campaign contributions, two indicators that are
specific to democratic countries like the US, this study adopts taking an active role in the ruling
political party and attending public hearings as two of the formal political indicators for
individuals in China. In addition, this study adjusts the indicators of political voice to include
civic engagement using social media so that this category can more accurately reflect the forms
of civic engagement in today’s world. Table 2 presents the indicators used in this study to
measure civic engagement for China and the United States.
Table 2. The Modified 17 Core Indicators of Civic Engagement
Civic Indicators Political Voice Indicators Formal Political Indicators
Community problem solving Contacting officials Regular voting
Regular volunteering for a
non-electoral organization
Contacting the media
regarding political issues
Membership in a political party
or organization
Active membership in a
group or association
Discussing political and
public issues on social
media
Running for office
Participation in fund-raising
for charity
Protesting Volunteering for a political
candidate or organization (US) /
Taking an active role in the
ruling political party or the
government (China)
Petitioning Political campaign
contributions (US) /
Attending public hearings and
open meetings (China)
Boycotting
One caveat is that the primary intention of this operationalization is to provide an
approach to capture the overall picture of civic engagement in the two samples from the U.S. and
China, not to conduct one-to-one comparisons over each specific form of civic engagement
between the two countries. Due to the differences in political systems and political culture, even
31
the thirteen shared indicators might not measure the exact same activities in the two samples. For
instance, being a member of a political party or organization in the American context can mean a
number of things, ranging from being a member of the two major political parties to being a part
of numerous political organizations. However, while there are eight official subservient political
parties in the Chinese context, this indicator most likely measures one type of membership for
the vast majority of Chinese people, namely, membership in the ruling Chinese Communist
Party. In addition, the interpretations of what a specific form of civic engagement entails might
also differ in different political contexts. Therefore, this project only employs this
operationalization to longitudinally compare the pre- and post-treatment levels of civic
engagement for the participants, not to horizontally compare between the two samples from the
U.S. and China.
2.3. Peer Effects, Individual Outcome, and Civic Engagement
In this section, I first discuss the definitions of peer effects. I then examine existing
literature that explores the link between peers and individual outcome in the field of political
science as well as other disciplines. Lastly, I review research that demonstrates the causal
influence of peers on civic engagement through Quasi-experiments.
2.3.1. Definition of Peer Effects
While the term “peer effects” seems to be intuitive enough that one does not need a
definition to understand what it entails, it is still important to state what peer effects are and are
not in order to operationalize this concept in empirical studies. Building upon Manski’s (1993)
work, I provide the operational definition of peer effects for this study in this subsection.
Firstly, it is important to understand what peer effects are not – peer effects are not the
correlation due to self-selection. Similarities in peers’ behavior, attitude, and outcome have been
32
commonly observed for a long time. And such similarities are often mistaken as peer effects
even though they might not be the result of a causal relationship. In his widely cited article,
Manski (1993) offers three different mechanisms for the similarities observed among peers:
endogenous effects, exogenous effects, and correlated effects. Among these three mechanisms,
the correlated effects occur when an individual self-select into peer groups where other members
share similar attitude and behavior as herself. Since peer effects are inherently causal, the
definition of peer effects should not include the observed similarity among peers’ behavior and
attitude caused by self-selection.
The other two mechanisms proposed by Manski (1993), endogenous effect and
exogenous effects, are both causal relationships between one and her peers. Therefore, peer
effects can be defined as the phenomenon where one’s behavior and attitude are affected by her
peers’ behavior, attitude, and characteristics (Sacerdote 2001, 2015, Ryan 2017). The current
study adopts this causal definition of peer effects in the empirical analysis.
2.3.2. The Link Between Peers and Individual Outcome
Despite being an important part of the socio-ecological environment that shapes social
behaviors, contextual factors are often neglected in research that explores individual civic
engagement behavior. Yet, as a key element in one’s context, peers have important influence on
one’s behavior. Therefore, I discuss findings on the link between peers and individual outcome
from different disciplines to shed light on the role of peers in civic engagement in this
subsection.
Research in various academic fields has demonstrated strong links between one’s peer
group on one’s behavior. For instance, medical research shows that having at least one obese
friend increases the probability of one’s own obesity by a significant amount (Christakis and
33
Fowler 2007), adolescent studies research demonstrates the positive correlation between a one’s
own adolescent delinquency and the criminal records of their peers (Case and Katz 1991,
Dishion et al. 1995), education research in particular has many projects that argue for a positive
correlation between one’s own academic achievements and their peers’ (Hoxby 2000, Lavy and
Scholosser 2011, Imberman et al. 2012). The link between peers and individual political outcome
also attracts some attention from political scientists. However, unlike the literature in education
and medical science that consistently finds similar peer outcomes, the finds regarding peer
effects in political science research tend to be conflicting and inconsistent.
In the study of political attitude, scholars explore the role of peer effects in attitude
change. Emphasizing the important influence of one’s micro-level social environment such as
peer groups on individual political attitude change, Kenny (1994) finds that an individual’s party
identification is more likely to be the same as their peers, especially when the peers are their
spouses or close friends, the individual believes the peers in the political discussions are
politically knowledgeable, and they discuss political issues frequently. Similarly, Beck (2002)
finds similar peer effects in voting behavior using evidence from the 1992 American presidential
election: an individual is more likely to “defect” and vote for the presidential candidate from the
other major political party if her peers support that person. However, using data from the
American National Election Study, MacKuen and Brown (1987) discover that while contextual
factors have strong and consistent effects on individual political attitude, the peer effects is
mixed as peers can influence citizens’ evaluations of the candidates and parties but not their
political self-identification. Also challenging the notion that peers tend to have similar political
outcome, Campos and coauthors (2017) find that peer political identification does not affect
individual identification in a significant way. Instead, they discover a new mechanism of peer
34
effects where the political identification of an individual who is in a more politically engaged
peer group tends to move away from the extremes and to the center of the political spectrum.
In the study of political behavior, scholars further explore the nuanced mechanisms
through which peers affect individual civic engagement. Using a sample of Australian
adolescents, Da Silva and coauthors (2004) find that an adolescent who has civically and
politically active friends is more likely to have a stronger sense of civic responsibility and higher
level of civic engagement. However, the strength of this peer effects depends on the quality of
the friendship and communication between one and her peers. Similarly, Harell and coauthors
(2008) find that the positive association between more frequent discussion of political and public
issues with peers and higher civic engagement is conditioned on the specific form of civic
engagement as well as the economic, political ,racial, and religious diversity of one’s peer
network. Pancer and coauthors (2007) suggest that peer effects take different forms and
categorize peer effects into four different patterns based on the political involvement of the peers
and the nature of one’s relationship with her peers. Their empirical result from a sample of
Canadian students shows that the positive peer effects on individual civic engagement only
occurs when one frequently discusses political topics with peers who have high levels of
involvement in political or community activities. Taking the time dimension into account, some
research projects explore the effect of peers on individual civic engagement overtime. For
instance, Zaff and coauthors (2008) demonstrate that one’s friends’ view on the importance of
civic life have positive correlation with one’s civic engagement later in their life.
Another approach to examine the influence of peers on ones’ political behavior and
attitude is influenced by the social capital theory and focuses on the collective influence of the
peer network. Among this approach, one group of research projects primarily focuses on
35
exploring the effect of having a high-quality and active peer network in general on individual
outcome. They find that an individual is more likely to have active civic engagement or prosocial
behaviors if she has good interpersonal relationships with the peers in her social (Wentzel and
McNamara 1999, Yates and Youniss 1998). Similarly, Sax (2004) finds that students from
colleges where more other students report high levels of social activism tend to have
significantly higher levels of social activisms themselves. A second group of research effort
focuses on the characteristics of one’s peer network, such as the diversity of the network, in
order to understand the influence of peers on one’s political participation. Yet, the findings in
this group of research are not consistent. Challenging the ideal derived from the social capital
theory that active engagement with peers encourages political participation, a growing body of
literature argues that exposure to peers with different political views tends to decrease individual
citizens’ political participation as the attitudinal diversity in their network leads them to hold
ambivalent political opinions, which discourages political participation (Mutz 2002, Costa and
Kahn 2003, Mutz and Mondak 2006, McClurg 2006). However, other research shows that
interacting with peers who hold different political views or engage in different political activities
promotes one’s own political participation when taking both macro-level and micro-level factors
as well as the specific forms of political participation into account (Leighley 1990, Pattie and
Johnston 2009, Scheufele et al. 2006, Quintelier et al. 2012).
Even though existing literature recognizes the link between peer group and individual
outcome, including civic engagement, the empirical evidence remains limited and inconsistent.
Few research projects explore the mechanism through which peers affect individual outcome. In
fact, the influence of peer characteristics on individual civic engagement is often blended with
the general richness of one’s social network. Research projects using different analytical
36
framework, methods, and data produce widely different results regarding the effect of peers on
one’s civic engagement. This is largely because most existing literature discusses only the
correlation between peers and individual outcome while very little research explores or
convincingly establishes the causality in peer effects. In particular, the research projects that find
positive correlations between peers and individual outcome constantly fail to rule out the
endogenous effect caused by the self-selective nature of peer groups. Therefore, the following
subsection focuses on discussing exploring and measuring the causal relationship between peers
and individual outcome.
2.3.3. Exploring the Causality of Peer Effects Through Quasi-experiments
While understanding peer effects has the potential to answering a wide range of questions
in social sciences, it is notoriously difficult to demonstrate the causal mechanism of peer effects
and to measure their magnitude. Manski (1993) hypothesizes that there are three main
mechanisms behind the observed peer similarities 1) endogenous effects, where one’s outcome is
influenced by their peers’ outcome; 2) exogenous effects, where one’s outcome is shaped by the
exogenous characteristics of their peers’; 3) correlated effects, where one’s outcome is similar to
their peers’ because one tends to self select into a group of peers who share similar individual
characteristics with oneself. In reality, the “peer effects” observed in empirical data is very likely
to be the result of a combination of all three mechanisms instead of the causal peer effects,
making it difficult to tease out and measure the influences through the causal mechanisms behind
peer effects.
The most commonly used method in modeling peer effects is the linear-in-means model
where one’s outcome results from the combination of her peers’ outcomes, her own
37
characteristics, and her peers’ characteristics (Sacerdote 2014). This model can be represented
with the formula below:
𝑌
-
= 𝛼+ 𝛽1∗ 𝑌
3
4-
+ 𝛾1∗ 𝑋
-
+ 𝛾2∗ 𝑋
3
4-
+ 𝜀
-
In this model, 𝑌
-
represents the individual of interest’s outcome. 𝑌
3
4-
is a vector that
represents the average outcome of all relevant peers in an individual’s network, excluding the
individual herself. 𝑋
-
is a vector that represents an individual’s own key characteristics. 𝑋
3
4-
is a
vector that represents the average key characteristics of all relevant peers in an individual’s
network, excluding the individual herself. This model directly includes peer effects derived from
two of the three mechanisms proposed by Manski (1993), namely, 𝑌
3
4-
that measures the
endogenous and 𝑋
3
4-
that measures exogenous effects. In addition, this model can also reflect
correlated effects caused self-selection into peer groups. While intuitive and parsimonious, the
linear-in-means model itself does not guarantee causality as it can still capture observed peer
effects caused by the endogenous, exogenous, and correlated effects.
The conventional method to model peer effects is regression analysis, where one’s
outcome is regressed on one’s peers’ outcomes along with a series of control variables. For
instance, Christakis and Fowler (2007) find that an individual is 57% more likely to be obese if
she has over-weight friends using a linear regression model. Challenging the causality of their
findings, Cohen-Cole and Fletcher (2008) replicate the research while controlling for self-
selection into peer groups and find that the strong peer effects on obesity greatly reduced to the
point that it is no longer statistically significant. This method produces significant results that
draw scholarly attention to peer effects, yet, it also faces compelling challenges in various
substantive fields in recent decades due to the difficulty in identifying 𝛽1, 𝛾1, and 𝛾2. Namely,
38
researchers cannot distinguish the influences of the endogenous effects, exogenous effects, and
correlated effects on an individual’s outcome or the magnitude of their influence.
Therefore, scholars need to utilize the power of research design in order to tease out the
endogenous, exogenous, and correlated effects and individually identify 𝛽1, 𝛾1, and 𝛾2 in the
linear-in-means model and demonstrate the causality of peer effects. The most ideal method to
address issue is through a controlled lab experiment. In a lab experiment with randomly selective
participants who are randomly assigned into treatment and control groups, researchers can easily
identify all parameters in the linear-in-means model. However, this method is almost never
feasible to examine peer effects. Firstly, artificially assigned peers for the sake of an
experimental study are unlikely to interact with each other the same way as actual peers (Carrell
et al. 2013). Secondly, even if researchers can find the extremely rare opportunities to randomly
assign experiment participants into peer groups in the real world, peer effects are only likely to
occur if the peers interact with each other on a relatively regular basis over extended periods of
time (Bronfenbrenner and Morris 1998) and artificial peer groups created for a academic
research project are highly unlikely to hold together and interact with such intensity overtime.
However, this does not mean individually identifying 𝛽1, 𝛾1, and 𝛾2 is completely out of
the question. Moffitt (2001) proposes that field experiments, or the partial population approach,
is suitable for exploring the causality of peer effects in the real world. Field experiments can
effectively introduce exogenous shock through assigning a treatment to randomly selected
members in natural peer groups instead of artificially assigning random peers to the participants.
In addition, researchers can manipulation the proportion of members who are randomly assigned
to receive the treatment in a peer group. This design also avoids the problems of artificially
creating randomized peer groups so that the results provide a more accurate description of how
39
peer effects functions in the real world. Moreover, by providing an instrument for peers’
outcomes, a field experiment allows researchers to differentiate the influence of peer
characteristics (𝛾2) and the influence of peer outcomes (𝛽1) on an individual’s outcome.
Babcock and Hartman (2010) employ this design in their research that explores the role of peer
effects in promoting college students’ participation in physical exercise. They randomly
incentivize some of the participants to go to the university gym and find that those who are
incentivized exercise if they have more peers who are also incentivized and that the positive peer
effects in the control group is significantly weaker. The field experimental design provides a way
to measure the endogenous effects of peers. In the mean time, it also offers an opportunity to
observe the changes in an individual’s outcome itself. Political science literature also frequently
adopts field experiments in order to explore the causality of one particular form of civic
engagement: voting. In attempts to promote voter turnout in the United States, a series of “get
out the vote” field experiments demonstrate that the voting behavior of one’s peers, including
neighbors and members of the same household, positively effect her own voting behavior
(Gerber and Larimer 2006, Nickerson 2008, Sinclair et al. 2012).
Yet, field experiments are only suitable when the treatment does not cause the
participants any harm beyond the minimal risk. Incentivizing college students to use the gym on
campus more often and encouraging voters in a sophisticated Western democracy to vote does
not have any significant negative consequences for the participants. However, due to the nature
of civic engagement, it is a lot riskier to incentivize participants as the random treatment. Some
forms of civic engagement can be dangerous. For instance, protests can turn violent and
confrontations with the law enforcement or other opposing groups can also lead to harm. The
danger of incentivizing civic engagement in a field experimental design is particularly high in a
40
non-democratic regime where more forms of civic engagement are considered as illegal or
strongly unfavorable. Therefore, it would be unethical and irresponsible to incentivize
participants to participate in certain forms of civic engagement, despite its strengths in exploring
the causality of peer effects.
Another approach to examine the causality of peer effects is using natural experiments to
take advantage of the exogenous shock that naturally occurs. Unlike lab experiments and field
experiments, nature experiments usually only have one source of exogenous variation through
the random variation of an individual’s peers. While introducing exogenous shock into one’s
social condition provides an ideal method to exclude the influence of correlated effects on her
outcome, one source of exogenous variation alone does not necessarily allow researchers to
distinguish the impacts on an individual’s outcome caused by her peers’ outcomes (𝛽1) versus
their characteristics (𝛾2). The most commonly used way to address this issue is to estimate the
“reduced-form” of peer effects (Sacerdote 2014), which no longer distinguishes the impacts of
peers’ outcomes and peers’ characteristics. This reduced-form of peer effects focuses on
excluding the influence of correlated effects from an individual’s outcome and acknowledges the
variation in an individual’s outcome resulted from her peers’ characteristics is a part of peer
effects in a more general sense. While this reduced-form of peer effects does not individually
identify 𝛽1, 𝛾1, and 𝛾2 in the linear-in-means model, it still does a good job in exploring how
peers causally affect an individual. Moreover, if the size of all peer groups in a study is known
and fixed, such as in the setting of college dorms, researchers can construct an equation for each
of the group members and use them to calculate 𝛽1 and 𝛾2 (Sacerdote 2001). Therefore, natural
experiments, if applied in a suitable setting, can be very useful in examining peer effects. On the
one hand, the exogenous shock makes it possible to rule out changes in an individual’s outcome
41
due to endogenous effects and correlated effects, allowing researchers to establish the causality
of peer effects. On the other hand, the results of natural experiments provide a more convincing
description of peer effects in the real work as they are observed in the setting of their genuine
peer groups that are not artificially manipulated.
There is a growing body of literature in different academic fields that explores different
natural experiment opportunities in order to causally examine the effect of peers on individual
outcomes. One natural experiment opportunity to explore the causality of peer effects is through
exogenous movements of people, which brings in exogenous shock to the self-selective nature of
peer groups. These movements can be the results of social programs or public policies such as
the desegregation of schools (Angrist and Lang 2004, Billings et al. 2014), relocation after
evacuations caused by natural disasters (Imberman et al. 2012), or relocation to more affluent
neighborhoods through housing vouchers (Kling et al. 2005). Exogenous movements can also
occur without government intervention, including the arrival of political refugees (Gould et al.
2009) or an unconventional religious to a traditionally homogeneous community (Chong 1994).
This approach is suitable for causally exploring how the average outcome of a large number of
peers in classrooms or neighborhoods affect individual outcome. However, this study is not
interested in examining the aggregated influence on individual outcomes resulted from the
mixture of many strong and weak social ties as this study theorizes that genuine peer effects
occurs only through regular and frequent interactions with peer over an extended period of time.
Randomly assigned roommates in a collective living setting, such as roommates in a
college dormitory, also provides a valuable opportunity to explore peer effects with natural
experiments. This approach is first used in research that aims at exploring the peer effects on
students’ performance. Sacerdote (2001) studies the GPA of students who have randomly
42
assigned dormmates at Dartmouth College and finds that having a roommate with high GPA
positively affects a student’s own GPA. Similarly, Stinebrickner and Stinebrickner (2005) and
Hoel et al. (2006) finds positive peer effects on students’ academic performance and West et al.
(2011) discovers strong positive peer effects on new cadets’ physical fitness scores at the United
States Air Force Academy. Yet, some research suggests peer effects may not exist. For instance,
Foster (2006) finds no positive peer effects on students’ academic performance with a sample of
students from the University of Maryland. Other research shows that the influence of peers’
outcome on one’s own outcome is conditional. For instance, Gleason and Siegfried (2003)
suggests that positive peer effects on academic performance only occurs on students who have
good grades with evidence from Vanderbilt University. In their article that explores peer effects
in attitude towards racial minority, Boisjoly et al. (2006) find that having randomly assigned
minority roommates leads white students to be have more liberal attitude regarding racial
politics. Kremer and Levy (2008) find that the negative impacts on a student’s GPA when having
a randomly assigned roommate who drinks alcohol only occur if the student is male, and that the
negative influence is larger when the student himself also drinks alcohol. Duncan et al. (2004)
find that male college students who have a predisposition to drink alcohol and a randomly
assigned roommate who also drinks tend to significantly increase their alcohol consumption. But
they do not find similar patterns in cannabis consumption, sexual behaviors, or female students.
In the field of political science, research that utilizes natural experiments made possible
by randomly assigned roommates remains scarce. Klofstad (2010, 2015) finds that engaging in
more political discussion with one’s randomly assigned roommates leads to her higher level of
civic engagement and that the magnitude of the influence increases over time. While Klofstad’s
study examines the influence of political discussions on individual civic engagement, which does
43
not necessarily reflect peers’ outcome, it successfully demonstrates the possibility of using
natural experiments with randomly assigned roommates to tackle important questions in political
science. In a recent study, Strother and coauthors (2021) finds that a student’s political ideology
becomes more similar to her roommates’ over time and that this effect is particularly strong on
students who initially hold conservative political views.
Even though natural experiments that take advantage of randomly assigned roommates
are highly context-specific and only applicable for a small fraction of the population, they offer a
rare opportunity to study peer effects in identifiable peer groups with specific boundaries where
the intensity and frequency of peer interaction is high enough for genuine influences among
peers’ outcomes.
2.4. Contributions of the Current Study
In this section, I discussion the contribution of this study from four aspects. Firstly, this
study contributes to constructing a more comprehensive and generalizable model of civic
engagement behavior. Conventional research that explores the determinants of civic engagement
tends to focus on individualistic factors such as education, income, and political efficacy or on
institutional factors such as political regime, political culture, political history. These studies
provide important findings that advance the understanding of civic engagement. However, very
few research projects examine the interactions of different determinants of civic engagement in a
comprehensive framework. Moreover, the findings of these studies tend to lack generalizability,
especially outside of liberal democracies. The current study bridges this gap by examining civic
engagement behavior in the context of a comprehensive socio-ecological system that
incorporates not only the individualistic and institutional factors, but also the contextual factors
that are often neglected.
44
Secondly, this study contributes to the civic engagement literature by demonstrating the
causal link between peers and individual civic engagement and bringing scholarly attention to
the role of social context in civic engagement. As discussed previously, existing literature
primarily focuses on examining civic engagement behavior from either the individual or the
institutional perspective. Even among the studies that discuss the link between social contexts
and civic engagement behavior, most describe the correlation instead of causation. Utilizing a
quasi-experimental design, the current study demonstrates the causal relationship between peers,
an important element in one’s social context, and one’s own civic engagement.
Thirdly, this study contributes to the peer effects literature by examining causal peer
effects in well-defined peer groups. Existing literature defines peer groups in various ways, such
as friends, classrooms, cohorts, and sometimes even neighborhoods, leading to inconsistent and
unstable findings on peer effects. Some of these peer groupings are too loosely organized or too
large for peer effects to occur. The current study examines how peers affect individual behavior
in roommate groups where members interact with one another closely and regularly throughout
an extended amount of time.
Moreover, this study contributes to the behavior change literature by demonstrating that
peers can positively influence an individual’s civic engagement behavior, shedding light on
innovative methods of promoting civic engagement.
45
Chapter 3 Methodology
In this study, I aim at addressing the following research questions. Do peers have a causal
effect on an individual’s civic engagement? If causal peer effects indeed exist, how do peers
affect an individual’s civic engagement? Do peers affect an individual differently on different
forms of civic engagement? Do peer effects differ in different institutional settings? What are the
mechanisms behind peer effects? Based on these research questions, this study constructs the
following hypotheses:
H10: Peers do not affect an individual’s civic engagement.
H1A: Peers affect an individual’s civic engagement.
H20: Having peers who have high levels of civic engagement does not increase one’s own
level of civic engagement.
H2A: Having peers who have high levels of civic engagement increases one’s own
level of civic engagement.
H30: Peer effects are uniform across all forms of civic engagement.
H3A: Peer effects are not uniform across all forms of civic engagement.
H40: Peer effects do not differ in different institutional settings.
H4A: Peer effects differ in different institutional settings.
In addition, the current study explores possible mechanisms of peer effects. This chapter
presents the research design of the current study as well as the methods used to examine the
impact of peers on an individual’s level of civic engagement and test the hypotheses listed above.
In the first section, I present the nested most different systems comparative research design that
combines a quasi-experiment and a series of in-depth interviews used in this study. The second
46
section reports the data collection process for this study. The remaining two sections respectively
discuss the construction of variables and the data analysis approaches used in this study.
3.1. Research Design
This section presents the research design of this study. Aiming at exploring the causality
behind peer effects in civic engagement and the mechanisms through which peers shape an
individual’s civic engagement, this study employs a nested most different systems comparative
design. In the large-N analysis of the nested analysis, this study adopted a quasi-experiment
using samples from two countries that are different in many aspects to demonstrate the causality
and generalizability of peer effects. This study then employs in-depth semi-structured interviews
in the small-N analysis to explore the mechanisms of peer effects.
3.1.1. A Nested Most Different Systems Comparative Analysis
This study employs research design that combines the Nested Analysis (Lieberman 2005)
and the Most Different Systems Design (Przeworski and Teune 1970) to explore the causality
behind peer effects in civic engagement. The nested analysis design is a rigorous approach in
comparative politics that systematically combines the strengths of large-N statistical inference
and in-depth small-N investigation proposed by Lieberman (2005). The most different systems
comparative design compares units of research with the same dependent variable but are
otherwise as different as possible in order to identify the key independent variable that causes the
same outcome (Mill 1874, Przeworski and Teune 1970). This study adopts a quasi-experiment
with a most different systems design to quantitatively demonstrate the existence of causal peer
effects in civic engagement and a series of in-depth interviews to explore the mechanisms behind
the causal peer effects.
47
The nested analysis design conventionally consists of large-N analysis of observational
data with regression models and small-N analysis of one or more cases within the large dataset
where the former explores patterns and trends and the latter assesses the plausibility these
patterns and generates theoretical insights (Lieberman 2005). As argued by Lieberman (2005,
436), quantitative analysis and qualitative analysis can “inform each other to the extent that the
analytic payoff is greater than the sum of the parts” in this integrated approach, and thus improve
the research’s prospects of making valid causal inferences.
In exploring the causal relationship between peers and one’s own civic engagement, this
study takes the nested analysis design one step further by employing a quasi-experimental design
as the large-N component. While conventional wisdom indicates that peers play an important
role in shaping an individual’s behavior and attitude, it is notoriously difficult to convincingly
demonstrate the causality of peer effects. On the one hand, conventional regression analysis
cannot differentiate causal effects and the correlated effects stem from the self-selective nature of
peer groups, and therefore lacks internal validity. On the other hand, lab experiments on peer
effects tend to neglect the rich contexts of real-world civic engagement, and thus lack external
validity. Moreover, artificially assigned peer groups in a lab experiment lack the long-term
frequent interactions that real-world peer groups have, which undermines the validity of the
research.
In comparison to using regression models to conduct large-N analysis on observational
data, the quasi-experimental design takes advantage of the naturally occurred exogenous
variation in the explanatory variables that resembles random treatment assignment to pinpoint
causality. Meanwhile, quasi-experiments do not artificially interfere with social behaviors in the
real world, providing the opportunity to examine peer effects in a realistic context without posing
48
threats to participants. Luckily, exogenous variation can sometimes naturally occur in the real
world in the manner that is similar to randomly treatment assignment. This quasi-experiment
takes advantage of randomly assigned roommates in some universities to causally explore peer
effects in civic engagement. The random assignment of roommates eliminates the correlated
effects caused by self-selection and allows for causal investigation of peer effects in civic
engagement.
In addition to incorporating the quasi-experiment design in the nested analysis, this study
also adopts a most different systems design in the quasi-experiment to further strengthen the
validity of the causal claims made in this study. In a most different systems design, the cases
have similar outcomes, or dependent variables, but are as different as possible in their
independent variables. If one can sift through a variety of independent variables and identify the
same independent variable in all different cases, that variable could be identified as the causal
agent the drives the similarity in the dependent variable. Therefore, the gist of using the most
different systems design to demonstrate causality is through falsification (Popper 1959, Anckar
2008). By adopting a most different systems design in selection samples for a quasi-experiment,
this study is able to not only demonstrates the causality of peer effects in civic engagement, but
also strengthens the generalizability of the causal inferences.
Therefore, this study employs a quasi-experiment where the treatment condition is
identified as the average level of civic engagement of an individual’s roommates’ civic
engagement as the large-N quantitative analysis in the nested analysis design in order to pinpoint
the casual effect between peer groups and individual civic engagement. In this quasi-experiment,
the only systematic variation is the treatment condition, namely, an individual’s background
variables such as pre-existing civic engagement, resources, and demographics are uncorrelated
49
with her roommates’ background variables. Moreover, as a contextual variable, roommates
interact with each other regularly and frequently over an extended period of time, which
constitutes a proximal process that can effectively shape an individual’s behavior and attitude
under the socio-ecological framework (Bronfenbrenner and Morris 1998). Furthermore, this
quasi-experiment adopts a most different systems design by selecting samples from China and
the United States, where both the contextual and institutional variables differ drastically, to
demonstrate the generalizability and validity of the causal relationship between one’s peers’ and
one’s own civic engagement.
As this study aims at not only demonstrating the existence of causal peer effects in civic
engagement, but also explaining the mechanisms of how peers causally affect one’s civic
engagement, I adopt a model-testing small-N analysis as the second step of in the nested
analysis. In the model-testing small-N analysis, this study utilizes in-depth semi-structured
interviews with selected participants in the quasi-experiment to test the quantitative model
generated from the large-N analysis. This study selects both “on-the-line” and “off-the-line”
cases for the model-testing small-N analysis. “On-the-line” case selection refers to selecting
cases where the values of the dependent variable are well-predicted by the fitted statistical model
generated in the large-N analysis, namely, cases that fall on the regression line in a plot that plots
the actual values on the predicted values of the dependent variable (Lieberman 2005). Whereas
“off-the-line” case selection refers to selecting cases where the values of the dependent variable
are not well-predicted by the fitted statistical model generated in the large-N analysis (Lieberman
2005). Both “on-the-line” and “off-the-line” cases contribute to verifying and better
understanding the quantitative results. Conducting model-testing small-N analysis on “on-the-
line” cases strengthens causal arguments in two ways. Firstly, it checks and rules out the effects
50
of spurious correlation. Secondly, it enhances the theoretical arguments by elaborating the causal
mechanisms behind the observed causal effects. In addition, the “off-the-line” cases provide
possible explanations of the puzzling findings of the large-N analysis.
3.1.2. Case Selection
This study adopts a set of case selection criteria that aims at producing a feasible way to
collect data that both contributes to the theoretical strength and the practical implications of this
study. Firstly, on the individual level, the quasi-experiment in this study selects relatively similar
candidates to maximize the comparability among participants. In order to demonstrate the
causality of peer effect, the quasi-experiment requires randomly assigned peer groups, which are
unusual since individuals normally have the ability to choose their peers. However, schools with
random dorm assignment provide a unique candidate pool and offers a rare opportunity to
examine peer effects in randomly assigned peer groups with relatively homogeneous members.
Therefore, this study chooses college students whose roommates are randomly assigned as the
candidate pool for the quasi-experiment.
Secondly, this project selects two samples from two drastically different countries, China
and the United States, for the nested most different systems design. Ideally, samples from
multiple different countries would be beneficial to more thoroughly demonstrate peer effects.
However, due to the limited scope and resources of this study, a feasible alternative is to select
samples from two different countries of theoretical and practical significance. Moreover, China
and United States have very different macro institutional settings ranging from regime type,
political institution, political culture, to political history, which provides a valuable opportunity
to demonstrate the robustness of peer effects.
51
Moreover, this study selects universities in China and United States to recruit participants
from based on two criteria: dorm assignment process as well as data collection affability and
feasibility. In terms of the dorm assignment process, universities in the United States and China
have quite different approaches. In the United States, apart from military academies, very few
universities have truly random dorm assignment. However, a number of universities assign dorm
rooms to freshmen using a relatively random process or a mixed process with an option of
random assignment, such as Dartmouth College, Chapman University, Syracuse University, etc.
Some of these universities give continuing students the option of choosing their own roommates
but randomly assign freshmen to dorm rooms, such as University of Southern California. Other
universities do not force all in-coming first year students to have randomly assigned roommates
but allow freshmen the option of having randomly assigned roommates. In this case, the majority
of students tend to choose to be matched with roommates based on preferences of dorm
buildings and room types instead of roommates’ characteristics. While these roommate
assignment processes are not purely random, the fact that freshmen cannot to pick their own
roommates based on their characteristics, especially their social behavioral characteristics, still
guarantees the elimination of self-selection bias.
Whereas in China, it is usually mandatory for all undergraduate students to live in dorm
rooms throughout their academic career in college and most universities do not allow students to
choose their roommates. A common way to assign students into dorm rooms is random
assignment by major or cohort. This dorm assignment process potentially poses one issue that
could challenge the quasi-experiment design as students can self-select into majors. However,
college majors do not have a strong association with civic engagement. Therefore, the choice of
one’s college major is unlikely to introduce self-selection effects in her civic engagement.
52
Moreover, Chinese students do not have a lot of control regarding their college majors for two
reasons. Firstly, in a typical scenario in China, parents, high school teachers, and the student
jointly decide which major a student applies to. Secondly, in college admission, a university
sometimes may choose to not approve a student’s choice of major and assign her to another
major instead. Therefore, a Chinese College student’s choice of major is not a direct reflection of
her self-selection and does not lead to self-selection effects in examining peer effects on civic
engagement.
In terms of data collection affordability and feasibility, this study selects University of
Southern California and Chapman University in the United States, Peking University, Qingdao
University, and Shandong University Qingdao Campus in China to recruit participants. All five
universities meet the case selection criteria discussed above and are affordable and feasible for
data collection considering the scope of this study. This study selects University of Southern
California and Chapman University in the United States primarily due to concerns of
affordability. University of Southern California is a private research university in Los Angeles,
California that admits around 4,000 first-year students every year. Chapman University is a
private research university in Orange, California, with a freshman cohort of approximately 2,000
students. Both universities are within driving distance, which makes data collection more
affordable. In addition, both universities require first-year students to live in university dorms
and have the option of random roommate assignment.
For the Chinese sample, the selection of Peking University, Qingdao University, and
Shandong University Qingdao Campus is primarily based on feasibility issues. Peking University
is a major research comprehensive university in Beijing, China that admits approximately 4,000
undergraduate students annually. Shandong University is a research comprehensive university
53
with campuses in Jinan, Weihai, and Qingdao, Shandong Province, China with a yearly
incoming cohort of around 10,000 undergraduate students. Qingdao University is a provincial
university in Qingdao, Shandong Province, China that admits roughly 8,000 undergraduate
students per year. Many college campuses in China are gated and require school IDs to enter the
campus, dorm rooms, and sometimes also classrooms, making recruiting participants in person
difficult. Moreover, foreign-based scholars tend to encounter numerous issues in their effort to
collect data in China. While small-scale data collection like this study does not need to receive
approval from or have the data collection results checked by local branches of the Chinese
Statistical Bureau like large-scale ones, foreign-based scholars often have difficulty recruiting
student participants in Chinese universities without professional connections with faculty or staff
members. The professional connections I have with faculty members in Peking University,
Shandong University Qingdao Campus, and Qingdao University make recruiting freshmen
participants feasible by demonstrate my academic intention and political trustworthiness to
university authorities.
This study collects data for the quasi-experiment through survey instruments that are
distributed in mail and in person at two different time points and collects qualitative data through
in-person and phone interviews. The hard copy of the survey questionnaire also includes a QR
code so the students can complete the survey on their electronic devices if they prefer. This study
recruits participants for the in-depth interviews from the participants of the quasi-experiments.
The majority of interviews conducted are in-person. Interviews with Chinese interviewees
conducted in June 2019 are all phone interviews.
In sum, this study selects five universities with random roommate assignment for first-
year students, namely, University of Southern California, Chapman University, Peking
54
University, Qingdao University, and Shandong University Qingdao Campus, to recruit
participants based on feasibility and affordability.
3.2. Data Collection Process
In this section, I discuss the data collection process for this study. The first subsection
discusses data collection using survey instruments for the quasi-experiment at two time points.
The second subsection discusses data collection through in-depth in-person interviews after
conducting the two surveys for the quasi-experiment.
3.2.1. Data Collection for the Quasi-experiment
This study employs a quasi-experimental design to explore the causality behind peer
effects on civic engagement. Participants of the quasi-experiment are first-year undergraduate
students who live with randomly assigned roommates recruited from University of Southern
California and Chapman University in the United States and Peking University, Qingdao
University, and Shandong University Qingdao Campus in Mainland China. Data collection for
the quasi-experiment consists of two waves: the first wave that collects pre-treatment data in the
2018 Fall semester and the second wave that collects post-treatment data in the 2019 Spring
semester. Both waves use a survey instrument to collect data (See Appendix E and F for the
survey instruments).
3.2.1.1. Pre-treatment Data Collection
This study collects pre-treatment data using a survey instrument in the beginning of the
2018 Fall semester. Pre-treatment data is collected via both hard copies of survey and online
survey platforms. Each participant receives a hard copy of the survey with a QR code that
provides a link to the online version of the survey. They can choose to complete the survey in
55
whichever format they prefer. For the students who prefer to complete the survey with pen and
paper, a pre-paid return envelope is provided. For the students who prefer to complete the survey
online, they can complete the survey on two online survey platforms: WJX for Chinese
participants and Qualtrics for American participants. The reason for using two separate platforms
for data collection is that the connection to foreign-hosted websites like Qualtrics tend to be
unstable and unpredictable under the Chinese internet regulations and censorship. Therefore, this
study chooses WJX, a Chinese online survey platform that offers similar functions as Qualtrics,
to collect data in China.
This study collects pre-treatment data for participants in the United States and China
respectively from August to September 2018 and from October to November 2018. While the
academic year starts in the fall in both China and the United States, first-year college students in
China are required to attend military education and emergency training, or junxun, at the
beginning of their first semester in college. For Chinese undergraduate students, the military
education and emergency training usually takes two to four weeks. During this time, students
usually live in military-style barracks rooms at an off-campus military training facility. Since
students live with other students who are not their randomly assigned dormmates during the
military education and emergency training at the beginning of the 2018 Fall semester, this study
postpones the pre-treatment survey to October 2018 when students from all three universities
finish their military education and emergency training and return to campus life.
The current study distributes the pre-treatment survey in three ways. First, this study hires
first-year undergraduate students who live in university dorms at the five universities as research
assistants to put letters containing the survey questionnaires in the mailboxes or distribute them
door to door at their own dorm buildings. Second, these research assistants also put up flyers
56
with participant recruitment information such as the nature of this study, participant
compensation, and participant eligibility in the public areas of the campus. The flyers each have
10 detachable slips with the link to the survey so the participant can tear it off and visit the online
survey at a later time. Third, this study also reaches out to faculty members and teaching
assistants at University of Southern California, Peking University, Qingdao University, and
Shandong University Qingdao Campus to distribute survey questionnaires in lower-division
general education classes. The students in these general education classes do not receive any
additional incentives to participate in this study other than the financial incentives that are
available to all participants. Their participation in this study does not affect how they are
evaluated in the general education classes.
All hard copies of the questionnaire contain a QR code that directs to the online pre-
treatment survey when scanned with a mobile device for participants who prefer to complete the
survey online. Participants in the United States who complete the pre-treatment survey are
compensated with 5-dollar Starbucks e-gift cards that is sent over email. Participants in China
who complete the pre-treatment survey are compensated with 35 Chinese Yuan, approximately
five dollars, via Alipay, a mobile and online platform that is very commonly used in China. The
participants are compensated within 30 days after they complete the survey regardless of their
answers in the pre-treatment survey. In addition to the compensation, all participants are entered
in a raffle for a 100-dollar e-gift card or a 700-yuan cash prize. The raffle is conducted upon the
completion of the post-treatment survey.
Moreover, participants are encouraged to invite their roommates to complete the pre-
treatment survey. If at least one of a participant’s roommates also completes the pre-treatment
survey, both participants’ names are entered into the raffle 10 more times. Since this study does
57
not collect identifiable personal information, each participant is assigned a random 8-digit
questionnaire number. A participant can provide the random 8-digit questionnaire number of her
participating roommate in order to get the 10 additional raffle entries. In addition, the survey
invites distributed by research assistants in dorms and by faculty members and teaching
assistance in classrooms are bundled into sets of two to eight in advance with the questionnaire
numbers documented, which provides another way to identify roommate groups.
3.2.1.2. Post-treatment Data Collection
This study collects post-treatment data using a survey instrument in the end of the 2019
Spring semester. In the pre-treatment survey, participants are asked to provide their email
addresses to receive information about the post-treatment survey. From April to May 2019,
emails containing information regarding this study and the link to the post-treatment survey are
sent out to participants who complete the pre-treatment survey at the beginning of the 2018 Fall
semester. Only participants who complete the pre-treatment are eligible for the post-treatment
survey.
Participants in the United States and China are respectively compensated with 15-dollar
Starbucks e-gift cards via email and with 105 Chinese Yuan via Alipay for the post-treatment
survey. The participants are compensated within 30 days after they complete the survey
regardless of their answers in the post-treatment survey. In addition, for participants who have at
least one roommate who also completes the post-treatment survey, their names are entered 10
more times in the raffle. For participants who complete both the pre-treatment and post-treatment
surveys, their names are entered into the raffle 10 more times. Namely, if a participant and at
least one of her roommates complete both the pre-treatment and post-treatment surveys, she has
31 entries in the raffle. Upon the completion of the post-treatment survey, two winners, one in
58
China and one in the United States, are randomly selected with an algorithm written with the
random library in Python. The winner in the United States is sent a 100-dollar e-gift card of
their choice via email and the winner in China is sent 700-yuan cash prize via Alipay.
As presented in Table 3 below, there are respectively 1,295 and 1,093 eligible
participants in the pre-treatment and post-treatment surveys after dropping incomplete or invalid
survey responses and participants who can not be matched with at least one roommate. The
overall longitudinal response rate of 84.40%. Among the 643 students in China who participate
in the pre-treatment survey, 297 are from Peking University, 185 are from Qingdao University,
and 161 are from Shandong University Qingdao Campus. Out of the 643 students in China who
complete the pre-treatment survey, 570 take the post-treatment survey. Among the 652 students
in the United States who participate in the pre-treatment survey, 244 are from Chapman
University and 408 are from the University of Southern California. 523 of these 652 students
participate in the post-treatment survey. The longitudinal response rates are respectively 88.65%
and 80.09% for the Chinese and U.S. samples.
Table 3. Data Collection Summary for the Quasi-experiment
Pre-treatment
Survey
Post-treatment
Survey
Longitudinal
Response Rate
Peking University 297 258 86.87%
Qingdao University 185 167 90.27%
Shandong University Qingdao Campus 161 145 90.06%
Chinese Sample Subtotal 643 570 88.65%
Chapman University 244 193 79.10%
University of Southern California 408 330 80.88%
U.S. Sample Subtotal 652 523 80.09%
Total 1,295 1,093 84.40%
59
3.2.2. Data Collection for the In-depth Interviews
In order to further explore the causal mechanisms behind peer effects in civic
engagement, this study conducts in-depth semi-structured interviews with some of the
participants in the quasi-experiment selected through a combination of “on-the-line” and “off-
the-line” case selection. The interviews are primarily conducted in May and June 2019 with
several follow-up interviews later in 2019.
This study chooses to use semi-structured interviews because semi-structured interviews
provide both the logical structure and flexibility needed for exploring the causal mechanisms of
peer effects in civic engagement. This study uses a set of questions (see Appendix G) prepared
beforehand to guide the interviews. The participants are also encouraged to share open-ended
responses and elaborate on the reasons behind their answers.
Moreover, the majority of interviews are conducted in person and on a one-on-one basis
in order to establish rapport with the interviewees. Five interviews are conducted by phone due
to scheduling issues. This study does not use any recording devices in order to protect the
identity of the interviewees since some topics of civic engagement are considered as politically
sensitive in China. No identifiable personal data in collected in the interviews. All in-person
interviews are conducted in a coffee shop or tea house on campus or close to campus of the
interviewee’s choice in order to make sure the interviewees feel comfortable to answer the
interview questions and can easily access the venue. Each interviewee is compensated with 40
dollars or 250 Chinese Yuan in addition to a soft drink of their choice.
This study first uses the quantitative model generated from the quasi-experiment to
predict the outcome of each participants and generates the candidate pools for “on-the-line” and
60
“off-the-line” cases. In order to ensure representation of both samples from China and the U.S., I
first send invitations for an in-depth interview to three “on-the-line” participants and three “off-
the-line” participants in both samples. If an invitation is turned down or receives no reply after
three days, I send an invitation to another similar participant. After interviewing the first 12
participants, I continue to conduct interviews with more “on-the-line” and “off-the-line”
participants as well as participants who are selected based on the information provided by
previous interview participants until the participants share the same stories frequently enough
that I can reasonably predict that additional participants can not provide much additional
information of value. This study conducts a total of 32 in-depth interviews (See Appendix D for
the list of interviews). Table 4 below presents basic information of the interviewees.
Table 4. Interviewee Information
Number of Interviewees Female : Male
Peking University 8 4:4
Qingdao University 7 4:3
Shandong University Qingdao Campus 4 1:3
Chapman University 3 1:2
University of Southern California 10 5:5
Total 32 15:17
Each semi-structured in-depth interview begins with an introduction of the researcher
and the study. In order to protect the privacy of the interviewees, no interview is recorded. After
the participants give consent to participate in the interview, I ask a series of questions regarding
their pre-treatment civic engagement, including their civic engagement behaviors, reasons they
participate or not participate in different forms of civic engagement, their opinions on civic
engagement activities, etc. I also ask the participants about their post-treatment civic engagement
and ask them to identify changes in their participation as well as reasons for such changes. In
addition, I ask them to evaluate and comment on their roommates’ civic engagement. Moreover,
61
I also ask participants about their interactions with their roommates. Specific questions used to
guide the interviews are presented in Appendix G.
3.3. Measuring Variables
In this section, I discuss the measurement and construction of each variables used in this
study. The first subsection discusses the measurements for the dependent variable and the
treatment variable, namely, an individual’s civic engagement and her roommate’s civic
engagement. The second subsection discusses the measurements for various background
variables that are used to gain more insight of peer effects.
3.3.1. Measuring Civic Engagement
As presented in Table 2 in subsection 2.2.3, this study measures civic engagement with
17 indicators, among which two are specific for participants from the United States and another
two are specific for participants from China. Measuring civic engagement with these indicators
provides flexibility for the empirical analysis. Each indicator can be used as a variable to
measure one form of civic engagement. Meanwhile, they can also be used to create multiple-item
composites of civic engagement, such as the three types of civic engagement indicators proposed
by Keeter et al. (2002) and an all-encompassing composite variable of civic engagement. The
rest of this subsection outlines the measurements of these indicators in three categories: civic
indicators, political voice indicators, and formal political indicators.
For the four civic indicators of civic engagement, this study measures community
involvement with two survey questions, volunteering for non-political organizations and fund-
raising participation with one question each, and civic group membership with a set of seven
survey questions. Since college life provides students with an abundance of opportunities to
62
participate in the student community, especially for first year students, this study measures
community involvement with two separate questions: one on a student’s involvement in the
student community and the other on her general local community. These two questions are both
measured on a 7-point ordinal rating scale where 1 represents having no prior experience of
community involvement and having no intention of community involvement in the future, 2
represents having no prior experience but are willing to participate in the future, and 3 -7
represent increasingly active levels of community involvement. Volunteering for non-political
organizations is also measured on a similar 7-point ordinal rating scale ranging from having no
prior experience volunteering for non-political organizations and having no intention to
volunteer in the future, to having no prior experience but are willing to participate in the future,
and to increasingly frequent levels of volunteering. For membership in civic groups and
organizations, this study uses seven questions to respectively measure an individual’s
involvement in trade unions, professional and civil associations, religious groups, business
associations, sports groups, volunteer organizations, and student organizations. Each
membership is measured on a 5-point ordinal rating scale ranging from no membership and no
intention to acquire membership in the future, to no membership but open to future membership,
and to increasingly active levels of membership.
For the six political voice indicators of civic engagement, contacting officials, contacting
the media regarding political issues, discussing political and public issues on social media,
protesting, petitioning, and boycotting, this study measures each indicator with one survey
question. All six questions are measured on a 7-point ordinal rating scale ranging from no prior
experience or future intention of participating, to no prior experience but open to participate in
the future, and to increasingly frequent levels of participation.
63
Among the seven formal political indicators, three are the same for both participants from
China and the United States, two are specific to Chinese participants, and two are specific to
American participant due to the drastic difference between the political institutions in the U.S.
and China. Namely, for all participants, this study uses five indicators that are applicable to their
countries as formal political indicators of their civic engagement. The three common indicators
are voting, running for office, and membership in a political party or organization. However,
these three shared indicators do not measure the same activities in China and the U.S. due to
their different political institutions. For American participants, these three indicators measure
voting, running for office, and membership in a political party or organization as Keeter and
coauthors intend (2002) in their operationalization of civic engagement in sophisticated
democracies.
However, since China is not a democracy and have a very different set of political
institutions, the first two of these three shared formal political indicators do not measure the
exact same three activities of the Chinese participants. While voting and running for political
office are also formal forms of political participation in China, citizens can only directly vote and
run for office in local-level elections. Urban residents can vote or run for local People’s Congress
every five years, whereas rural residents can vote or run for village committees every three to
five years. At the time of this study the latest election for the local People’s Congress was in
2016 and the next upcoming election in 2021, which means the urban Chinese participants will
not have to opportunity to vote or run for office in another election between the two surveys.
Meanwhile, running for local People’s Congress as an independent candidate is rare and highly
politically sensitive. If this study measures voting and running for pollical office in China the
same way as the U.S., it is likely that the results will be consistently low among most
64
participants. Therefore, in order to reflect the variations in these two activities, I relax the
definition of election to include elections for Student’s Congress in addition to elections for local
People’s Congress and rural village committees. Student’s Congress, while not a formal political
organization, is the official self-governing student organization in a university.
Therefore, voting and running for office are measured differently in the Chinese and
American versions of the surveys. In the surveys for American participants, voting and running
for office are measured with one question each. Whereas for Chinese participants, voting and
running for office are measured with three questions each: in an election for local People’s
Congress, in an election for rural village committee, and in an election for university-level
Student’s Congress. Each question is measured with a 7-point ordinal rating scale ranging from
no prior experience or future intention of participating, to no prior experience but open to
participate in the future, and to increasingly frequent levels of participation. Membership in a
political party or organization is measured with a 5-point ordinal rating scale ranging from no
membership and no intention to acquire membership in the future, to no membership but open to
future membership, and to increasingly active levels of membership.
For American participants, the other two indicators of formal political participation are
volunteering for political candidates and organizations and making campaign contributions, two
typical indicators of formal political participation indicators in a democracy as suggested by
Keeter and coauthors (2002). Yet, such forms of formal political participation do not exist in
China. For Chinese participants, the other two indicators are taking an active role in the ruling
political party or the government and attending public hearings and open meetings because they
are the two most common ways for Chinese citizens to participate in formal politics other than
voting, running for office, and being a member of a political party or organization. All four
65
activities are measured on a 7-point ordinal rating scale ranging from no prior experience or
future intention of participating, to no prior experience but open to participate in the future, and
to increasingly frequent levels of participation.
3.3.2. Background Variables
In addition to variables that measure civic engagement behavior and interest, this study
also employs a set of measure participants’ background information to gain more insight of the
participants and their interaction and ensure the samples are representative of the population. The
background variables are categorized into two groups: basic personal background information
and information on dorm rooms.
This study collects basic personal background information with variables including age,
gender, ethnicity/race, and year in college. Age is measured with the year a participant was born.
Gender is measured with a binary variable with the values of male and non-male. Due to the
differences between China and the United States, ethnicity/race is measured differently. For
participants in China, where over 90% of the population are Han Chinese, ethnicity is measured
with a categorical variable with the values of Han Chinese and non-Han Chinese. For
participants in the United States, race is measured with a categorical variable with the values of
Caucasian American, African American, Hispanic, Asian American, and other. Year in college is
a screening variable to ensure the participants are first-year undergraduate students.
For information on dorm rooms, this study collects information on roommate assignment
methods, number of roommates, time living in current dorm room, and time living with current
roommates. The roommate assignment method variable is a screening variable to ensure the
participants have randomly assigned roommates. Only participants who select “I chose to have
66
randomly assigned roommates” or “I did not have a choice but to have randomly assigned
roommates” are eligible for this study. Number of roommates is measured with numeric values.
The survey also measures the time a participant lives in her dorm room and with her current
roommates. Only participants of the first survey receive the second survey, which means a
participant will not be matched with any other participants as roommates if a participant changes
her roommates between the two surveys unless her new roommates are also participants of the
first survey. Therefore, the second survey includes a question that asks the participants to list the
random 8-digit questionnaire number for any roommates who fit this description. In addition,
another question asks if the participant attends the same university throughout the course of this
study in order to rule out the influence of other factors.
3.4. Data Analysis Approaches
This section discusses the data analysis approaches employed in this study. This section
first discusses the data preprocessing process for the quasi-experiment data, including data
screening, roommate matching, and civic engagement variable construction. Then, this section
presents the approach to analyze the quasi-experiment data using a modified version of linear-in-
means model.
3.4.1. Data Preprocessing
Before conducting quantitative analysis on the data collected through the quasi-
experiment, this study first preprocesses the data in ensure data quality through data screening
before matching roommates together. Then, this study constructs a set of civic engagement
variables for the data analysis.
67
3.4.1.1. Data Screening
In order to ensure the quality of the data collected for the quasi-experiment in this study, I
screen the participants based on a set of screening variables. Part I of both surveys include
screening questions for the participants to ensure all participants are first-year college students
who live with randomly assigned roommates. The online version of the surveys automatically
ends if a participant’s answer to the screening questions do not meet the criteria of this study.
The hand-filled surveys require manual participant screening and a total of 90 invalid survey
responses are dropped. Moreover, 164 participants are dropped from the dataset because they
cannot be matched with any other participants as roommates in either survey, transfer to another
university, or drop out from their undergraduate program.
3.4.1.2. Matching Roommates
Since this study does not collect identifiable personal information such as addresses from
the participants, this study relies on the 8-digit questionnaire numbers that are randomly
generated for each participant to match roommates together. In both the pre and post-treatment
surveys, participants are invited to provide the 8-digit questionnaire numbers of their roommates
who also participate in this study. A participant is first matched with the roommate(s) she
mentioned in the surveys. Then, if the participant is mentioned in other participants responses as
their roommate, they are also matched as roommates. In addition, as discussed in 3.2.1.1, the
survey invites distributed by research assistants in dorms and by faculty members and teaching
assistance in classrooms are bundled into sets of two to eight in advance with the questionnaire
numbers documented. Therefore, I first use the documented roommate groups to verify the
roommate groups identified through self-reported information and then to match roommate
groups who participate in the surveys but fail to self-report.
68
3.4.1.3. Construction of the Civic Engagement Variables
3.3.1 discusses the survey questions used to measure the 17 indicators of civic
engagement. To better explore peers’ influence on different types of civic engagement, this study
constructs a set of civic engagement variables.
To measure an individual’ civic engagement, this study constructs four composite
variables in addition to the 17 indicators that measures individual forms of civic engagement.
The first three composite variables are respectively the composite variables for civic
participation, political voice, and formal political participation. To ensure the indicators are
weighted equally in the composite variables, this study first constructs intermediate variables that
represent the 17 indicators of civic engagement. For indicators that are measured by only one
survey question, this study directly takes the responses to the survey question and rescales them
on a 1 to 10 scale as the intermediate variable. For indicators that are measured by more than one
survey question, this study first sums the responses to the survey questions and then rescale it on
a 1 to 10 scale in order to construct the intermediate variable.
The composite variable for civic engagement is the summation of the intermediate
variables that measure community problem solving, volunteering for a non-electoral
organization, membership in a group or association, and participation in charitable fundraising.
The composite variable for political voice is the summation of the intermediate variables that
measure contacting political officials, contacting the media regarding political issues, discussing
political and public issues on social media, protesting, petitioning, and boycotting. The
composite variable for formal political participation is the summation of the intermediate
variables that measure voting, membership in a political party or organization, running for office,
volunteering for a political candidate or organization (U.S.) or taking an active role in the ruling
69
political party or the government (China), and campaign contributions (U.S.) or attending public
hearings and open meetings (China).
The last composite variable measures the overall civic engagement. The overall political
engagement variable is calculated on the basis of the aforementioned three composite variables.
This study first rescales the three composite variables on the same 1 to 10 scale to ensure equal
representation of the three aspects of civic engagement and then calculates the summation of
them to construct the overall civic engagement composite variable.
Moreover, this study creates variables to represent treatment by calculating the average
civic engagement of her roommates for each of the 17 indicators and the four composite
variables directly with the self-reported data of the roommates.
3.4.2. Analyzing Quantitative Data from the Quasi-experiment
This study employs a modified version of the linear-in-means model to analyze the
quantitative data generated from the quasi-experiment. As discussed in Chapter 2, the linear-in-
means model is the most commonly used method to measure peer effects. The original linear-in-
means model can be represented with the formula below:
𝑌
-
= 𝛼+ 𝛽1∗ 𝑌
3
4-
+ 𝛾1∗ 𝑋
-
+ 𝛾2∗ 𝑋
3
4-
+ 𝜀
-
In this model, 𝑌
-
represents the individual of interest’s outcome. 𝑌
3
4-
is a vector that
represents the average outcome of all relevant peers in an individual’s network, excluding the
individual herself. 𝑋
-
is a vector that represents an individual’s own key characteristics. 𝑋
3
4-
is a
vector that represents the average key characteristics of all relevant peers in an individual’s
network, excluding the individual herself. If the linear-in-means model is applied to standard
observational data with no exogenous shock, 𝜀
-
also includes the correlated effects caused by self
70
selection into peer groups where one shares similar attitude and behavior with other members.
But since this quasi-experiment takes advantage of the exogenous shock introduced by the
random assignment of roommates, correlated effects can be ruled out.
In addition, this model directly includes peer effects derived from two of the three
mechanisms proposed by Manski (1993), namely, the endogenous peer effects measured by 𝑌
3
4-
and the exogenous peer effects measured by 𝑋
3
4-
. Since both the characteristics and outcomes of
peers reflect the causal effects of peers on an individual’s outcome, it is not necessary to
differentiate the two if the goal of analysis is simply to identify the causal effects of peers on
individuals. Moreover, the theoretical framework of this study does not offer assumptions where
peers’ background characteristics such as gender, age, and ethnicity influence one’s own civic
engagement. Therefore, the linear-in-means model can be reformulated into a reduced form that
does not include 𝑋
3
4-
:
𝑌
-
= 𝛼+ 𝛽∗ 𝑌
3
4-
+ 𝛾1∗ 𝑋
-
+ 𝜀
-
Furthermore, this study collects both pre and post-treatment data in order to calculate
peer effects. Since the pre-treatment data is collected at the beginning of the 2018 Fall semester
when the participants are yet to spend enough time with their new roommates, this study assumes
their civic engagement at the time is not affected by their college roommates. Therefore, the two
formulas that represent an individual’s civic engagement at the beginning of the 2018 Fall
semester (t1) and the end of the 2019 Spring semester (t2) are as the following:
𝑌
-9%
= 𝛼
9%
+ 𝛾1∗ 𝑋
-
+ 𝜀
-9%
𝑌
-9(
= 𝛼
9%
+ 𝛽∗ 𝑌
3
4-
+ 𝛾1∗ 𝑋
-
+ 𝜀
-9(
71
In these two formulas, 𝑌
-9%
represents the pre-treatment civic engagement of an
individual, which is affected by 𝑋
-
, the vector that represents an individual’s own key
characteristics, and 𝜀
-9%
, an all-encompassing error term that accounts for other factors that
influence an individual’s civic engagement, such as previous peer groups. 𝑌
-9(
represents the
post-treatment civic engagement of an individual, which is jointly shaped by 𝑋
-
and 𝑌
3
4-
, the
vectors that respectively represent an individual’s own key characteristics and the average
outcome of all her relevant peers. Since peer effects can be calculated by subtracting the first
formula from the second formula, 𝑌
-9(
can be rewritten as:
𝑌
-9(
= 𝛼+ 𝛽
%
∗ 𝑌
3
4-
+ 𝛽
(
∗ 𝑌
3
-9%
+ 𝜀
-
In this modified formula, an individual’s post-treatment civic engagement is a function of
the average civic engagement of her peers and her own pre-treatment civic engagement. This
study employs this modified formula to explore the peer effects on each of the 17 specific forms
of civic engagement, the three main categories of civic engagement (civic indicators, political
voice, formal political indicators), and the overall civic engagement.
72
Chapter 4 Results
As hypothesized in Chapter 3, this study is designed to answer four questions: 1) Do
one’s peers affect her civic engagement? 2) Do peers with high levels of civic engagement
increase one’s own civic engagement? 3) Are peer effects uniform across all forms of civic
engagement? 4) Do peer effects differ in different institutional settings? In this chapter, I present
the results from the quasi-experiment to address these questions. I start by reporting the summary
statistics of the quasi-experiment in the first section to provide a general idea of the data
collected in this study. To address the first two questions, I show the results of the peer effects on
an individual’s overall civic engagement and the three general aspects of civil engagement,
namely, civic indicators, political voice indicators, and formal political indicators in the second
section. The third section focuses on the third question and presents the results of peer effects on
each of the 17 specific forms of civic engagement. Both Section Two and Three include the
overall results from all participants as well as a comparison between the results from the Chinese
and American samples.
4.1. Data and Descriptive Analysis
This section offers a descriptive analysis on the data collected from the quasi-experiment,
including background variables, composite variables for civic engagement, and variables that
measure each of the 17 specific forms of civic engagement. The descriptive statistics includes the
mean, standard deviation, minimum, maximum, and number of observations for the data. In
addition, this section also provides evidence of the random assignment of roommates using a
series of univariate regressions on a participant’s civic engagement and her roommates’.
73
Appendix A reports the descriptive statistics for the background variables. The average
age is 19.22 for all participants, 19.19 for participants from China, and 19.25 for participants
from the U.S. The range of participants’ age is between 18 to 24 years, indicating the sample
largely consists of young adult participants. The percentage of female participants is around 40%
for both samples. While the percentage of female participants is slightly lower than the national
average in both China and the U.S., the number of female participants in of this quasi-experiment
is still sufficient to ensure a balanced representation due to the relatively large sample size. The
percentage of ethnic minority participants in the sample from China is 9%, which is in line with
the 8.89% ethnic minority in mainland China (National Bureau of Statistics of China, 2020). The
percentage of ethnic minority participants in the sample from the U.S. is 61%, which is higher
than the 42.2% ethnic minority nationwide (United States Census, 2020). This is because all
participants in the U.S. sample are from California, where the percentage of ethnic minority is
63.5% (United States Census, 2020). The average dorm size is 2.88 for the sample from the U.S.
and 4.23 for the sample from China. In Chinese colleges, the most common dorm sizes are four,
six or eight students. Whereas American colleges tend to offer more options, including two-
person dorms.
Appendix B and Appendix C respectively present the descriptive statistics for the civic
engagement variables collected from the first wave of survey in 2018 and 2019, including the
composite variable for all civic engagement indicators, composite variables for the three aspects
of civic engagement, and different forms of civic engagement. Table 5 below contains a brief
summary of the descriptive statistics for the four composite variables of civic engagement,
namely, overall civic engagement, civic participation, political voice, and formal political
participation. As Table 5 indicates, participants from the U.S. have higher levels of civic
74
engagement across the board in the both waves of survey, except for the composite variable for
all civic indicators in the post-treatment survey.
Table 5. Descriptive Statistics for the Four Composite Variables of Civic Engagement
All Participants China U.S.
Variable Mean Std. D Mean Std. D Mean Std. D
First Wave of Survey (Pre-treatment)
Composite variable for all civic
engagement indicators 11.12 4.77 9.71 4.59 12.48 4.54
Composite variable for all civic
indicators 16.93 6.92 16.36 7.92 17.49 5.71
Composite variable for all
political voice indicators 20.18 11.07 17.52 11.03 22.81 10.47
Composite variable for all
formal political indicators 12.52 7.25 9.57 5.12 15.36 7.84
Second Wave of Survey (Post-treatment)
Composite variable for all civic
engagement indicators 12.12 4.21 11.30 4.11 12.91 4.16
Composite variable for all civic
indicators 19.14 6.08 20.20 6.51 18.10 5.42
Composite variable for all
political voice indicators 20.99 10.02 18.46 9.88 23.47 9.53
Composite variable for all
formal political indicators 13.92 6.98 11.57 5.98 16.16 7.13
In terms of overall civic engagement, participants in the U.S. are more active than their
Chinese counterparts in both waves of survey, which is expected considering the difference in
the political institutions. Paired two-sample Wilcoxon tests indicate that the medians of overall
civic engagement of the U.S. sample are significantly greater than the Chinese sample in both
waves of survey, both with a p value that is less than 0.001. Political voice and formal political
participation follow similar pattern as overall civic engagement, where participants in the U.S.
75
are more active in both waves of survey. Yet, in terms of civic participation, while the pre-
treatment survey in 2018 shows that participants in the U.S. are more active, participants in
China have higher levels of civic participation in the post-treatment survey in 2019. An unpaired
two-sample Wilcoxon test shows that the median of civic participation in the Chinese sample is
significantly greater than the U.S. sample, with a p value that is less than 0.001.
The two samples from China and the U.S. also show differences in the pre- and post-
treatment surveys. For both samples, the levels of overall civic engagement, civic participation,
and formal political participation in the post-treatment survey are higher than the pre-treatment
survey. Paired two-sample Wilcoxon tests indicate that the medians of these three composite
variables are significantly higher in the post-treatment survey than the pre-treatment survey.
Participants in the U.S. also tend to be more active in terms of political voice in the post-
treatment survey. However, a paired two-sample Wilcoxon test indicates that participants in
China are not significantly more active in political voice in the post-treatment survey, with a p
value of 0.28.
Table 6 presents the results of a set of univariate regressions between a participant’s own
level of participation in each form of civic engagement and the average level of participation of
her roommates in the pre-treatment survey, providing evidence of the random assignment of
roommates. The quasi-experiment in this dissertation takes advantage of the random roommate
assignment in colleges to eliminate the correlated effects that stem from the self-selective nature
of peer groups. In order to demonstrate the random assignment of roommates indeed resembles a
randomized experiment, this study regresses a participant’s own initial level of civic engagement
on the average initial level of her roommates’ civic engagement. As indicated in Table 6, the lack
of statistical significance on the univariate regression coefficients and the low R-squared values
76
show that one’s own civic engagement and her randomly assigned roommates’ civic engagement
are independent at the beginning of this study. This demonstrates that random roommate
assignment resembles a randomized experiment.
Table 6. Univariate Regression of One's Own Civic Engagement on Average Civic Engagement
of Roommates in the First Wave of Survey
Variable China US
Civic Indicators
Coef. R
2
Coef. R
2
CI-1
Community problem solving -
overall
0.213 0.003 -0.061 0.001
(0.162)
(0.066)
CI-2
Volunteering for a non-electoral
organization
0.097 0.001 -0.098 0.002
(0.144)
(0.077)
CI-3
Membership in a group or
association - overall
-0.015 0.001 -0.044 0.002
(0.019)
(0.042)
CI-4
Participation in charitable
fundraising
10.022 0.000 -0.128 0.003
(0.184)
(0.089)
Political Voice Indicators
PV-1
Contacting officials
-0.069 0.001 -0.137 0.004
(0.156)
(0.088)
PV-2
Contacting the media regarding
political issues
-0.217 0.003 -0.073 0.001
(0.155)
(0.076)
PV-3
Discussing political and public
issues on social media
0.017 0.001 -0.113 0.003
(0.157)
(0.081)
PV-4
Protesting
0.015 0.001 -0.085 0.003
(0.035)
(0.066)
PV-5
Petitioning
-0.183 0.002 -0.043 0.001
(0.172)
(0.092)
PV-6
Boycotting
0.064 0.001 -0.079 0.002
(0.176)
(0.078)
Formal Political Indicators
FP-1
Voting
0.026 0.001 -0.095 0.002
(0.064)
(0.087)
FP-2
Membership in a political party or
organization
-0.122 0.004 -0.019 0.000
(0.076)
(0.048)
77
FP-3
Running for office
0.126 0.002 0.077 0.001
(0.113)
(0.084)
FP-4
Volunteering for candidate or
political organization (US)
0.011 0.000
(0.082)
Taking an active role in the ruling
political party or the government
(China)
0.035 0.001
(0.099)
FP-5
Campaign contributions (US)
-0.195 0.004
(0.125)
Attending public hearings and open
meetings (China)
0.073 0.002
0.064
Significance: + = p < 0.1; * = p < 0.05; ** = p < 0.01; *** = p < 0.001
4.2. Linear-in-Means Analysis: Peer Effects on Overall Civic Engagement
This section reports the Linear-in-Means results of peer effects on overall civic
engagement and the three categories of civic engagement, namely, civic indicators, political
voice indicators, and formal political indicators. The results presented in this section reject
hypotheses H10 and H20. In addition, these results cannot reject H 40 at the level of overall civic
engagement:
H10: Peers do not affect an individual’s civic engagement.
H1A: Peers affect an individual’s civic engagement.
H20: Having peers who have high levels of civic engagement does not increase one’s own
level of civic engagement.
H2A: Having peers who have high levels of civic engagement increases one’s own
level of civic engagement.
H40: Peer effects do not differ in different institutional settings.
H4A: Peer effects differ in different institutional settings.
As discussed in Chapter 3, this study adopts a modified Linear-in-Means model to
examine the peer effects in civic engagement. Linear-in-means is the most commonly used
78
method in modeling peer effects where one’s outcome is the result of a combination of her own
characteristics, her peers’ outcome, and her peers’ characteristics (Sacerdote 2014). Since this
study focuses primarily on how peers affect individual civic engagement in a quasi-experiment
where roommate assignment is random, it is not necessary to differentiate peers’ outcome from
their characteristics. Namely, the current study sees the variation in an individual’s outcome that
results from her peers’ characteristics also as a part of peer effects. Therefore, this study adopts a
modified version of the Linear-in-Means model where an individual’s post-treatment civic
engagement is a function of the average civic engagement of her peers and her own pre-
treatment civic engagement to explore peer effects in civic engagement.
Table 7 and Table 8 respectively report the results from the modified Linear-in-Means
analysis for Chinese and American participants in this study. The dependent variable examined
in Model (A1) and Model (A5) is overall civic engagement, is a composite variable for all forms
of civic engagement. As these two models indicate, roommate’s pre-treatment overall civic
engagement is a statistically significant predictor of one’s own post-treatment overall civic
engagement for participants both in the U.S. and China. For civic participation, the composite
variable that represents the four forms of civic indicators, Model (A2) and Model (6A) show that
roommates’ pre-treatment civic participation is a statistically significant predictor of one’s own
post-treatment civic participation in both countries. Similarly, for the political voice composite
variable and formal political participation composite variable, the two pairs of models, Model
(A3) and Model (A7), Model (A4) and Model (A8), respectively demonstrate that roommates
pre-treatment participation is a statistically significant predictor of one’s own post-treatment
participation.
79
Table 7. Peer Effects on Overall Civic Engagement Among Chinese Participants
(A1) (A2) (A3) (A4)
Own
Overall
Civic
Engagement
(Post-
treatment)
Own Civic
Participation
(Post-
treatment)
Own
Political
Voice (Post-
treatment)
Own Formal
Political
Participation
(Post-
treatment)
(Intercept)
0.370
3.379*** 1.856+ -1.630*
(0.515) (0.992) (1.042) (0.673)
Own Overall Civic
Engagement (Pre-
treatment)
0.515***
(0.025)
Roommates' Overall Civic
Engagement (Pre-
treatment)
0.618***
(0.053)
Own Civic Participation
(Pre-treatment)
0.487***
(0.024)
Roommates' Civic
Participation (Pre-
treatment)
0.757***
(0.079)
Own Political Voice (Pre-
treatment)
0.558***
(0.026)
Roommates' Political
Voice (Pre-treatment)
0.379***
(0.056)
Own Formal Political
Participation (Pre-
treatment)
0.675***
(0.034)
Roommates' Formal
Political Participation
(Pre-treatment)
0.627***
(0.058)
R-squared
0.441 0.356 0.383 0.390
N
642 642 642 642
Significance: + = p < 0.1; * = p < 0.05; ** = p < 0.01; *** = p < 0.001
80
Table 8. Peer Effects on Overall Civic Engagement Among American Participants
(A5) (A6) (A7) (A8)
Own
Overall
Civic
Engagement
(Post-
treatment
Own Civic
Participation
(Post-
treatment)
Own
Political
Voice (Post-
treatment)
Own Formal
Political
Participation
(Post-
treatment)
(Intercept)
2.395***
7.841***
8.678***
4.621***
(0.722) (1.185) (1.561) (0.948)
Own Overall Civic
Engagement (Pre-
treatment)
0.539***
(0.030)
Roommates' Overall Civic
Engagement (Pre-
treatment)
0.334***
(0.048)
Own Civic Participation
(Pre-treatment)
0.356***
(0.035)
Roommates' Civic
Participation (Pre-
treatment)
0.319***
(0.072)
Own Political Voice (Pre-
treatment)
0.438***
(0.032)
Roommates' Political
Voice (Pre-treatment)
0.234***
(0.058)
Own Formal Political
Participation (Pre-
treatment)
0.528***
(0.029)
Roommates' Formal
Political Participation
(Pre-treatment)
0.208***
(0.046)
R-squared
0.347 0.148 0.223 0.340
N
652 652 652 652
Significance: + = p < 0.1; * = p < 0.05; ** = p < 0.01; *** = p < 0.001
81
The modified Linear-in-Means models that regresses one’s post-treatment civic
engagement on her own pre-treatment civic engagement and the average pre-treatment civic
engagement of her roommates in itself is not sufficient to demonstrate the casual relations of
peer effects. However, as discussed in the previous section, the random roommate assignment for
the participants creates a sample that resembles one from a randomized experiment. Thus, the
results presented above are sufficient to reject H10, indicating peers can affect an individual’s
civic engagement. Furthermore, since the coefficients in the modified Linear-in-Means are all
positive, the results also reject H20, demonstrating peers who have high levels of civic
engagement positively influence one’s own civic engagement. Furthermore, the modified Linear-
in-Means models for both participants in the U.S. and China report similar results, which is
consistent with H40 in terms of the composite variables for civic engagement. Since the results
presented in this section do not reject H40 at the level of overall civic engagement, the positive
peer effects on civic engagement are consistent in China and the U.S., two countries with
significantly different institutional settings.
4.3. Linear-in-Means Analysis: Peer Effects on Different Forms of Civic
Engagement
The previous section reports the quasi-experiment results that demonstrate peer effects on
overall civic engagement and the three main categories of civic engagement using data collected
from both China and the U.S., rejecting the null hypotheses H10 and H20,. This section delves into
peer effects and specific forms of civic engagement in different institutional settings. The results
from the Linear-in-Means analysis suggest that peer effects are not uniform across all forms of
civic engagement. Moreover, peer effects on specific forms of civic engagement also differ in
82
different institutional settings. The results presented in this section rejects H30 and partially
rejects H40 on the level of specific forms of civic engagement:
H30: Peer effects are uniform across all forms of civic engagement.
H3A: Peer effects are not uniform across all forms of civic engagement.
H40: Peer effects do not differ in different institutional settings.
H4A: Peer effects differ in different institutional settings.
Table 9 and Table 10 respectively present the Linear-in-Means results of the four civic
indicators of civic engagement, including community problem solving, membership in an
association, volunteering for a non-electoral organization, and participation in charitable
fundraising. Among these four specific forms of civic engagement, roommates’ pre-treatment
level is a statistically significant predictor of three forms, namely, one’s own community
problem solving, membership in an association, and participation in charitable fundraising both
in the U.S. and China. However, in terms of volunteering for a non-electoral organization,
roommates’ pre-treatment level is only a statistically significant predictor among Chinese
participants, not their American counterparts. In addition, it is worth noting that the magnitude of
positive peer effects is larger in overall community problem solving and charitable fundraising
for the Chinese participants than their American counterparts.
83
Table 9. Peer Effects on the Civic Participation Indicators of Civic Engagement Among Chinese
Participants
(CI-1) (CI-2) (CI-3) (CI-4)
Own
Overall
Community
Problem
Solving
(Post-
treatment)
Own
Associational
Membership
(Post-
treatment)
Own Non-
electoral
Volunteering
(Post-
treatment)
Own
Charitable
Fundraising
(Post-
treatment)
(Intercept)
1.817*** 0.452* 3.948*** 0.602*
(0.296) (0.228) (1.042) (0.287)
Own Community Problem
Solving (Pre-treatment)
0.561***
(0.028)
Roommates' Community
Problem Solving (Pre-
treatment)
0.357***
(0.091)
Own Associational
Membership (Pre-treatment)
0.482***
(0.035)
Roommates' Associational
Membership (Pre-treatment)
0.539***
(0.090)
Own Non-electoral
Volunteering (Pre-treatment)
0.244***
(0.029)
Roommates' Non-electoral
Volunteering (Pre-treatment)
0.540***
(0.105)
Own Charitable Fundraising
(Pre-treatment)
0.486***
(0.029)
Roommates' Charitable
Fundraising(Pre-treatment)
0.704***
(0.080)
R-squared
0.394 0.274 0.135 0.386
N
642 642 642 642
Significance: + = p < 0.1; * = p < 0.05; ** = p < 0.01; *** = p < 0.001
84
Table 10. Peer Effects on the Civic Participation Indicators of Civic Engagement Among
American Participants
(CI-5) (CI-6) (CI-7) (CI-8)
Own
Overall
Community
Problem
Solving
(Post-
treatment)
Own
Associational
Membership
(Post-
treatment)
Own Non-
electoral
Volunteering
(Post-
treatment)
Own
Charitable
Fundraising
(Post-
treatment)
(Intercept)
2.768*** 1.437*** 2.844***
1.565***
(0.323) (0.227) (0.349) (0.330)
Own Community Problem
Solving (Pre-treatment)
0.339***
(0.035)
Roommates' Community
Problem Solving (Pre-
treatment)
0.146***
(0.071)
Own Associational
Membership (Pre-treatment)
0.442***
(0.034)
Roommates' Associational
Membership (Pre-treatment)
0.160***
(0.057)
Own Non-electoral
Volunteering (Pre-treatment)
0.292***
(0.041)
Roommates' Non-electoral
Volunteering (Pre-treatment)
0.083
(0.065)
Own Charitable Fundraising
(Pre-treatment)
0.289***
(0.036)
Roommates' Charitable
Fundraising(Pre-treatment)
0.200**
(0.068)
R-squared
0.136 0.208 0.074 0.097
N
652 652 652 652
Significance: + = p < 0.1; * = p < 0.05; ** = p < 0.01; *** = p < 0.001
Table 11 and Table 12 respectively present the Linear-in-Means results of the six
political voice indicators of civic engagement, including contacting officials, contacting the
media regarding political issues, political discussion on social media, protesting, petitioning, and
85
boycotting. Among these six specific forms of civic engagement, roommates’ pre-treatment level
is a statistically significant predictor of two forms, namely, one’s own political discussion on
social media and petitioning in both in the U.S. and China. Out of the remaining four political
voice indicators, roommates’ pre-treatment level is a statistically significant predictor of only
boycotting among Chinese participants. Whereas for participants in the U.S., roommates’ pre-
treatment level is a statistically significant predictor of three out of these four forms, including
contacting officials, contacting the media, and protesting.
86
87
88
Table 13 and Table 14 respectively present the Linear-in-Means results of the seven
formal political indicators of civic engagement. Since the political institutions differ significantly
in China and the U.S., this study adopts three shared formal political indicators, including voting,
running for office, and membership in a political party or organization. As discussed in Chapter
3, the first two of these three shared indicators do not measure the same activities in China and
the U.S. due to their different political institutions. For the participants in China, the definitions
of voting and running for office is relaxed to include not only participation in local People’s
Congress or rural committee elections, but also participation in Student’s Congress at university
level. In addition, this study also substitutes the last two formal political indicators with attending
public hearings and open meetings and taking an active role in the ruling party or the government
for participants in China, as volunteering for a political candidate or organization and political
campaign contributions are not available in China due to differences in political institutions.
Among the three shared indicators, roommates’ pre-treatment level is a statistically significant
predictor of two forms, namely, one’s own voting and running for office in both in the U.S. and
China. Roommates’ membership in a political party is only a statistically significant predictor for
one’s own membership in a political party among Chinese participants. Peers are not a
statistically significant predictor of either unique formal political indicators for Chinese
participants. Whereas for American participants, peers are a statistically significant predictor of
political campaign contributions, but not volunteering for a political candidate or organization.
89
90
91
4.4. Summary of Findings
To summarize, the results from the quasi-experiment demonstrate that peers’ civic
engagement positively affect an individual’s overall level of civic engagement. Positive peer
effects also apply to each of the three primary categories of civic engagement, namely, civic
participation, political voice, and formal political participation. Moreover, such positive peer
effects are observed in both the Chinese and the American samples despite the differences in the
political institutions of the two countries. Yet, peer effects are not uniform across all specific
forms of civic engagement in the two samples. Table 15 below summarizes the statistical
significance of the results from the quasi-experiment for the four composite variables and the 17
forms of civic engagement measured:
Table 15. Statistical Significance of the Quasi-experiment Results
China U.S.
Overall Civic Engagement Yes Yes
Civic Participation (Composite) Yes Yes
Community Problem Solving Yes Yes
Associational Membership Yes Yes
Non-electoral Volunteering Yes No
Charitable Fundraising Yes Yes
Political Voice (Composite) Yes Yes
Contacting Officials No Yes
Contacting Media No Yes
Political/Public Discussion on Social Media Yes Yes
Protesting No Yes
Petitioning Yes Yes
Boycotting Yes No
Formal Political Participation (Composite) Yes Yes
Voting Yes Yes
Running for Office Yes Yes
Membership in A Political Party/Organization Yes No
Campaign Contributions (U.S.) - Yes
Political Volunteering (U.S.) - No
92
Attending Public Hearings (China) No -
Taking an Active Role in the Ruling Party
(China) No -
The results from the quasi-experiment confirm that peers positively affect an individual’s
overall level of civic engagement and that such effects are beyond the constraints of political
institutions. However, the quasi-experiment also demonstrates that peer effects are not
completely universal across different specific forms of civic engagement in different institutional
settings. Yet, the unexpected variations of peer effects do not invalidate this study on peer effects
in civic engagement. Instead, they provide an opportunity to explore the mechanisms of peer
effects to explore how institutional, contextual, and individualistic variables collectively and
dynamically affect civic engagement behaviors. The next chapter offers more detailed
interpretation of the quasi-experiment results and discusses possible mechanism of peer effects
based on the findings from the in-depth interviews.
93
Chapter 5 Discussion
Understanding the determinants of civic engagement is crucial in the study of politics.
This study aims at examining peer effects in civic engagement within a comprehensive
theoretical framework that builds upon the socio-ecological model and the rational choice
theories, and thus contributes to understanding the relatively under-researched contextual
dimension of civic engagement. As reported in Chapter 4, the results from the quasi-experiment
demonstrate that peer effects exist in civic engagement and that having randomly assigned
roommates who have higher levels of civic engagement is likely to increase one’s own level of
civic engagement in both samples from the U.S. and China. Moreover, quantitative results also
indicate that peer effects are not uniform across different specific forms of civic engagement or
different institutional settings.
This chapter begins with a discussion that interprets the results of the quasi-experiment,
situating the quantitative findings in the theoretical framework of this study. Drawing on the
results of the in-depth semi-structured interviews, the second section proposes possible
mechanisms of peer effects.
5.1. Summary and Interpretation of the Quantitative Findings
The primary goal of this study is to explore the causal relationship of peer effects in civic
engagement. To measure civic engagement, this study first operationalizes the concept of civic
engagement and constructs three groups of variables. The first group is a composite variable that
measures an individual’s overall civic engagement. The second group is three composite
variables that measure the three main categories of civic engagement, namely, civic
participation, political voice, and formal political participation. The third group is the 17
94
variables that measure the specific forms of civic engagement. As reported in Chapter 4, the
Linear-in-Means results demonstrate that having roommates with higher levels of overall civic
engagement is likely to lead to the increase of one’s own overall civic engagement in both the
samples from the U.S. and China. Similar peer effects can also be observed for all three
composite variables that represent the three primary categories of civic engagement in both
samples. However, peer effects are not uniform across all specific forms of civic engagement or
institutional settings.
The quantitative findings demonstrate that peers have a positive effect on one’s overall
civic engagement and the three primary categories of civic engagement, namely, civic
participation, political voice, and formal political participation. In other words, if a participant
has roommates whose average level of civic engagement is higher, she is likely to have a higher
level of civic engagement after a semester’s interaction. This is consistent with a body of
literature that focuses on the social determinants of civic engagement (e.g. Sobel 1993, Huckfeldt
& Sprague 1995, Putnam 2001, Jennings 2002, Zukin et al. 2006, Stolle 2007, Greenberg 2008).
Under the socio-ecological framework, the positive peer effects in civic engagement occur in the
mesosystem, where the reciprocal interactions between roommates take place. As participants
interact with their roommates on a regular basis over extended periods of time, they make sense
of their environment and adapt their behaviors and attitudes accordingly.
It is worth noticing that the positive peer effects in civic engagement apply to both
samples from the U.S. and China, two countries with drastically difference political institutions
ranging from regime type to political culture. In other words, the reciprocal interactions between
peers on a regular basis through an extended period of time can lead to the transmission of
behaviors and attitudes in both participants from a democratic country with a rich participatory
95
political culture and an authoritarian country with a history of subject political culture. This
finding effectively demonstrates the robustness of peer effects in civic engagement.
However, the robustness of peer effects in civic engagement does not mean such effects
would mechanically occur across all specific forms of civic engagement without the influence of
other factors beyond the mesosystem of peer interactions. The results from the quasi-experiment
show variations of peer effect across different specific forms of civic engagement and different
institutional settings. In the Chinese sample, peers’ pre-treatment level is not a significantly
predictor of one’s own post-treatment level in five forms of civic participation, including
contacting political officials, contacting the media regarding political issues, protesting,
attending public hearings and open meetings, and taking an active role in the ruling party or the
government. In the American sample, peers’ pre-treatment level is not a significantly predictor of
one’s own post-treatment level in four forms of civic participation, including non-political
volunteering, boycotting, membership in a political party or organization, and volunteering for a
political candidate or organization.
While the current study does not attempt to provide a comprehensive explanation of how
peers affect an individual’s behaviors and attitudes, the variations of peer effects found in
different forms of civic engagement in both samples calls for further investigation. Moreover,
such variations could potentially shed light on possible mechanism of peer effects in civic
engagement. In this regard, the in-depth semi-structured interviews conducted with selective
participants in the quasi-experiment provide some meaningful findings. The following section
discusses the qualitative findings with the in-depth interviews and reports possible mechanism of
peer effects in civic engagement.
96
5.2. Qualitative Findings and Possible Mechanisms of Peer Effects in Civic
Engagement
As discussed in the previous section, quantitative data from the quasi-experiment
demonstrates 1) one’s peers affect her civic engagement behaviors, 2) peers with higher levels of
civic engagement increase one’s own civic engagement, 3) peer effects do not differ in different
institutional settings for overall civic engagement, 4) but peer effects differ across specific forms
of civic engagement in countries with different institutions. In an attempt to explore the
mechanisms of the aforementioned quantitative findings and the variations of peer effects in
specific forms of civic engagement, this section discusses the qualitative findings from the 32 in-
depth interviews. A list of interviews is provided in Appendix D. Since the rest of this section
frequently cites the interviews in the discussion, Table 16 below summarizes different forms of
civic engagement and the corresponding interviews cited in this section to demonstrate the peer
effects.
Table 16. Summary of Cited Interviews
Forms of Civic Engagement Interviews Cited
Community Problem Solving 2, 5, 6, 16, 19, 21, 29
Associational Membership 2, 4, 9, 11, 19, 25
Non-electoral Volunteering 2, 7, 13, 16, 18, 20, 28
Charitable Fundraising 23, 26
Contacting Officials 1, 4, 18, 29
Contacting Media 1, 29
Political/Public Discussion on Social Media 5, 19
Protesting 5, 6, 21
Petitioning 4, 10
Boycotting 11, 32
Voting 22, 30
Running for Office 8
Membership in A Political Party/Organization 14, 25, 26, 31, 32
Campaign Contributions (U.S.) 23, 27
Political Volunteering (U.S.) 24
97
Attending Public Hearings (China) 18
Taking an Active Role in the Ruling Party
(China) 8
From these in-depth interviews, two primary mechanisms of peer effects emerge. The
first mechanism of positive peer effects in civic engagement is that one’s peers increase her own
perceived benefits of a given form of civic engagement to a point where they outweigh the
perceived costs. The second mechanism is that one’s peers decrease her own perceived costs of a
given form of civic engagement so that the perceived benefits outweigh costs. In addition, the
last part of this section discusses the lack of peer effects in some cases due to limited interactions
between roommates.
5.2.1. Peer Effects Through Increased Perceived Benefits
One important possible mechanism of how roommates affect a participant’s civic
engagement is through increasing the perceived benefits of a given form of civic engagement.
Namely, when a participant interacts with her roommates who participate in a given form of
civic engagement, the perceived benefits of said form gets updated through the interaction. Thus,
it leads the participant to evaluate or re-evaluate civic engagement activities and decide whether
they are suitable for herself. Positive peer effects would then occur if the updated perceived
benefits outweigh the costs.
This mechanism is fairly obvious when a participant is introduced to new forms of civic
engagement through the interaction with roommates. For example, a freshman at Peking
University has a randomly assigned roommate who regularly participates in“plogging”, a form
of volunteering that combines jogging with picking up litter, also decided to participate on a
98
regular basis after trying it out with his roommate. When asked about his reasons for this decide,
he elaborates:
“I’ve never heard of plogging before my roommate told me about
it. It sounds so cool! So, I said yes when he invited me to go
plogging together with him and several other people. It was such a
great experience. We ran together for about a whole hour and
picked up three huge bags of trash before eating the best hotpot in
Beijing. Ever since then, I joined my roommate in plogging and
our plogging group has been bigger and bigger. I get to meet new
friends, explore the really cool parts of Beijing, eat at the hidden
gem restaurants, workout, and do something good for the
environment. It’s just so many birds one stone. It’s part of my life
now.”(Interview 2)
As Interview 2 indicates, positive peer effects occurred when this participant learnt a new
form of civic engagement from his roommate because he evaluated plogging as an activity where
he could benefit from the physical exercise, friendship, and the sense of fulfillment through
protecting the environment. In this case, his personal benefit and selective benefit greatly
outweigh the costs and risks of plogging. Similarly, Interview 6, 7, 9, 21, 23, and 29 also report
positive changes in civic engagement behaviors when introduced to a new civic engagement
activity by their roommates in community problem solving, associational membership, charitable
fundraising, volunteering, etc.
Positive peer effects through increased perceived benefits can also occur when a
participant’s roommate is active in a form of civic engagement she already knows about. For
instance, a participant from Qingdao University decided to become a volunteer tutor for under-
privileged children after she learnt from her roommate that participating in volunteer work could
be accounted towards her extracurricular performance, which is one of the key indicators in the
evaluation of scholarship, student awards, and even graduate school admission (Interview 16). In
this case, the participant’s perceived personal benefits of volunteering greatly increase through
99
the interaction with her roommate and start to outweigh the costs, leading to the positive peer
effects.
Similarly, another participant from Qingdao University decided to become a member of
the Chinese Communist Party (CCP) under his roommates’ influence. In the first survey, this
participant indicated that he was not a member of any political party or organization and that he
had no intention to join one in the future. At the time of the interview, he had already submitted
his application to join CCP and become a probationary member. When asked about this change,
he attributed his change of mind to his roommates:
“Being a [Chinese Community] Party member was never important
for me. Plus, it’s a lot of work. The application process, the
meetings, Party classes and trainings, and all the things one has to
do just seem to be too much for me. I initially thought it wasn’t
worth the effort. But I have three roommates who are either
already probationary members or actively seeking to be one. To be
honest, I was a bit puzzled by them at the beginning because none
of them seemed to be a true believer of Marxism or Leninism or
anything of that sort. But later I learnt from them that being a Party
member can open a lot of doors for you, including getting a job as
a civil servant or working at a big state-owned enterprise. One of
my roommates told me that being a Party member not only gives
you an advantage in the civil servant exam, some positions even
make it a hard requirement [to be a CCP member]. Another of my
roommates believes that being a Party member in itself is like a
stamp of approval on your resume. It’s not about politics, it just
means you are very good in school. It makes sense. If becoming a
Party member can give me an edge on the job market, of course
I’m willing to do it.” (Interview 14)
As Interview 14 indicates, positive peer effects occur when the participant’s perceived
benefits significantly increased after he learnt about the selective benefits that are exclusive for
CCP members and outweighed the costs of going through the application process. Prior to his
interaction with his roommates, this participant’s perceived benefits of being a Party member
were mostly based on political beliefs and ideology. For him, if one is not a Marxist or a true
100
believer of Communism, they cannot benefit much from joining CCP. Yet, his roommates
updated his perceived benefits of being a CCP member by adding multiple selective benefits to
his utility function, leading him to reevaluate his decision.
Another example shows this mechanism also applies to political voice-related civic
engagement. A participant from University of Southern California took part in her first ever
protest with her roommate and became part of the nation-wide 2019 Presidents’ Day
demonstrations because of her roommate. In the first survey, she indicated that she had never
participated in protesting but was willing to consider participating in the future. After President
Donald Trump declared a national emergency in order to build a wall along the border between
the United States and Mexico, she decided to join her roommate and protest in front of Los
Angeles City Hall. One of the reasons why she chose to be part of the protest is her friendship
with her roommate, a second-generation immigrant from Mexico (Interview 21). She elaborates
her reasoning behind her participation in the protest:
“I wanted to be there for my roommate. This is very important to
her. There was no ground for the declaration of a national
emergency what so ever. It was pure xenophobia and racism. Her
family would be affected if this wall was built. We cannot let that
happen.” (Interview 21)
In this case, peer effects occur because the perceived benefits of protesting increased for
this participant. After forming the friendship with her roommate, what matters to her roommate
also becomes more important to the participant herself. Supporting her friend adds to the
participant’s perceived personal benefits of protesting, which prompts her to reevaluate her
decision about protests.
Other interviews also confirm increasing perceived benefits as a mechanism of peer
effects in civic engagement, such as engaging in community problem solving (Interview 6, 16,
101
21, 29), having membership in associations and civic groups (Interview 9, 11, 25), volunteering
for non-political causes (Interview 7, 13, 18, 31), fundraising (Interview 23, 26), discussing
political issues on social media (Interview 19), contacting officials and media (Interview 29),
donating to political candidates (Interview 23, 27), etc.
5.2.2. Peer Effects Through Changing Perceived Costs
Besides increasing the perceived benefits of civic engagement, another main possible
mechanism of how roommates affect a participant’s civic engagement is through decreasing the
perceived costs. As specified in the modified utility function of civic engagement in Chapter 2,
the costs of civic engagement include the fixed cost of participating, the uncertain costs of
participating that is dependent on the perceived risks, and the cost of not participating. The
reciprocal interaction between roommates updates the perceived costs and risks of civic
engagement, and thus prompts the participant to re-evaluate civic engagement activities. Positive
peer effects would occur if the perceived benefits outweigh the updated perceived costs.
Based on the qualitative data collected from the in-depth interviews, decreasing the
perceived costs of a given form of civic engagement is often associated with changing its
perceived risks through interacting with one’s roommates. For instance, a student from Peking
University starts to actively participate in student community problem solving and discussing
public affairs on social media after seeing her roommate’s behaviors. She explains her decision
to join her roommate.
“I was shocked when she brought up the issue
1
to the department
and even posted it online. I mean, all social media accounts are
linked to your real personal information these days and they will
know who posts what. In fact, a senior told me that one single
incident could cost someone the opportunity to be recommended
1
“The issue” refers to a controversial incident at the participant’s university. To protect the
participants’ anonymity, the current study omits the details of the incident.
102
for graduate school and could even leave a permanent record on
your file. I was concerned for her because it could potentially cost
her future if she gets kicked out for bringing up the issue and even
disseminating it on social media. But I guess it’s not as bad as I
thought it was going to be. She didn’t really get into trouble. I
mean, yes, she was invited to “have tea”
2
with the department
leadership. But they were pretty nice and agreed to solve the
problem as fast as possible. I guess it’s not that bad to stand up for
ourselves. Now that I think about it, the senior who told me about
all the consequences probably intentionally lied so I would self-
censor.” (Interview 5)
As Interview 5 indicates, positive peer effects occur when her roommate’s active
participation in student community problem solving and online political discussion lowers her
own perceived risks of these activities. This mechanism of peer effects is particularly common in
forms of civic engagement that are considered as politically sensitive in China, as many
participants tend to over-estimate the costs and risks due to the political socialization in the
authoritarian regime.
Other interviews provide similar explanations. In Interview 4, the participant decides to
volunteer for a progressive environmental group after his perceived risks are adjusted through his
interaction with his roommate. As a law student, the participant is particularly interested in
environmental laws. In the interview, he shared that he had always wanted to volunteer for an
environmental organization but never did due to concerns of political sensitivity. He used to
consider working with independent nonprofits highly risky because “there’s always news about
lawyers being put under travel restrictions, disbarred, or jailed.” (Interview 4). But after
interacting with his roommate who regularly volunteers for a nonprofit that provides legal aid for
2
“Have tea”, or hecha, is commonly used to refer to situations where someone is approached by
the authorities for a forced appointment, especially when their behavior or speech is not
welcomed by the authorities.
103
labor disputes, which he sees as even more politically sensitive though environmental issues, he
lowered his perceived risks and started volunteering for an environmental nonprofit. He explains:
“I guess I was too focused on the negative cases and didn’t realize
that most people who work on these issues are fine. If anything, the
authorities probably won’t care too much about a college student
volunteer who mostly collects information and processes
paperwork.” (Interview 4)
In this case, Interviewee 4 initially perceived political repression as highly probable for
participating in an environmental advocacy group, which leads him to have a high perceived
uncertain cost of participation. The experience of his roommate adjusted his perceived
probability of political repression and therefore lowered his perceived costs, leading him to
reevaluate his decision to volunteer for the environmental organization.
Meanwhile, this mechanism does not solely apply to forms of civic engagement that are
commonly perceived as politically risky in China. Lowering the perceived costs of participation
also leads to positive peer effects in other forms of civic engagement and in the U.S. For
instance, a participant from University of Southern California becomes more open to participate
in protests, especially the ones for racial equality, under her roommates’ influence (Interview
21). As mentioned in 5.2.1, one of the reasons why Interviewee 21 decided to join the Presidents’
Day protest in 2019 is the increased perceived benefits of participation due to her friendship with
her roommate. Another reason that leads to her eventual participation is the decreased perceived
costs of participation. She was initially concerned about counter-protesters, police actions,
brutality, and other safety issues. Her roommate who regularly participates in protests regarding
immigrant rights and racial justice helped her with safety tips before the protest, which greatly
lowered her perceived uncertain costs of participating in the protest (Interview 4).
104
It is worth mentioning that not all costs are associated with participating in a given form
of civic engagement. One type of cost is the cost of not participating, which is common in peer
effects. In Interview 11, a participant from Shandong University explains why she chooses to
boycott against 85 Degrees Café and Meet Fresh, two Taiwanese dessert restaurants.
“I actually love them both. I used to get either a milk tea or a
tapioca dessert bowl whenever I go shopping in that area. But my
roommate feels really strongly about the political stuff. You see,
she’s a devoted patriot. Of course, she was furious after what
happened at the Golden Horse Awards
3
and started boycotting all
Taiwanese businesses and products. I personally don’t feel very
strongly about this. I’m not saying I agree with those Taiwan
Independence separatists, obviously. It’s just sometimes it’s
difficult to tell which business is Taiwanese, which is Japanese,
which is Korean, all that stuff. But between milk tea and my
friend, I would choose her any day. So, I join her in solidarity and
also boycott Taiwanese businesses. At the end of the day, I can
always get milk tea elsewhere.” (Interview 11)
As Interview 11 shows, the participant’s roommate’s enthusiastic participation in
boycotting Taiwanese businesses increases the perceived cost of not participating, and thus leads
to the participant’s participation in boycotting Taiwanese businesses. Similarly, a participant
cites “not wanting to be left out” as the most important reason why she joins mock U.N. with her
three roommates (Interview 11). A participant chooses to give his signature for a petition
because his roommate “keeps badgering him and wouldn’t leave him alone until he signs”
(Interview 4). There are also participants who casted their first votes in local elections because
“if everyone else does it and I don’t, I would feel like a bad citizen.” (Interview 22, 30).
3
At the 55
th
Golden Horse Film Festival and Awards in 2018, Fu Yue, a Taiwanese director, said
Taiwan should be seen as a separate entity from China in her acceptance speech.
105
Other interviews also confirm decreasing perceived costs as a mechanism of peer effects
in civic engagement, such as engaging in community problem solving (Interview 5, 19), having
membership in associations and civic groups (Interview 2), signing petitions (Interview 10), etc.
5.2.3. Why Sometimes Peers Fail to Affect an Individual’s Civic Engagement?
While the quasi-experiment demonstrates that one’s peers’ civic engagement affects her
own civic engagement in a positive way and such peer effects apply to overall civic engagement
and all three categories of civic engagement, having a roommate who is active in a given form of
civic engagement does not necessarily leads to changes in one’s own participation in said form
of activity. The previous two subsections discuss two possible mechanisms of peer effects in
civic engagement in order to explain how peers influence each other. But it is also important to
analyze why, sometimes, peers do not influence each other. Therefore, this study also utilizes the
in-depth interviews to explore the reasons why sometimes peers fail to change an individual’s
civic engagement through the in-depth interviews.
The first reason is consistent with the two mechanisms of positive peer effects discussed
in the previous two subsections. Lowering the perceived costs and risks of civic engagement
does not always lead to positive peer effects, just like increasing the perceived benefits of civic
engagement also does not guarantee higher levels of civic engagement because there are other
sources of perceived risks and benefits when an individual makes her decisions. The second
reason why peers sometimes fail to change an individual’s civic engagement the lack of
meaningful interaction between roommates. Since the proximal processes have to occur “on a
fairly regular basis over extended periods of time” (Bronfenbrenner and Morris 1998), peer
effects are unlikely to occur when there is not sufficient interaction between roommates.
106
Firstly, peer effects will not occur if one’s perceived benefits of participating in a form of
civic engagement increase, but not to a point where they could outweigh the perceived risks. For
instance, a participant from University of Southern California has a roommate who actively
volunteers for a local political candidate. Yet, the participant himself indicated that he had never
volunteered for a political candidate and would not consider volunteering in the future in both
surveys. In the interview of the off-the-line case, the participant shared his reasoning:
“I totally understand why my roommate volunteers for the
politician. He’s always into politics. Plus, he’s a pre-law student. It
would look good on his resume. But it just doesn’t make sense for
me. If I support someone, I would probably just donate to him and
let the experts figure out how to get him elected.” (Interview 24)
In this case, even though the participant sees the benefits of political volunteering from
his roommate’s experience, the perceived benefits of participation for him do not increase much.
As an engineering student, he does not share the benefit of having an extracurricular activity that
favors his future career or believe he could provide much professional contribution to the
campaign. The personal benefits regarding political volunteering differs significantly between
the utility functions of the participant and his roommate. Therefore, the participant’s perceived
benefits of volunteering for a politician only increase marginally in this case, which is not
enough to outweigh the perceived costs or change his participation behavior.
Similarly, two other participants whose roommates actively participate in non-electoral
volunteering indicated that the volunteering activities were not as meaningful to them as to their
roommates (Interview 20, 28). In addition, a participant from Peking University whose
roommate is among the small number of Chinese participants who have contacted officials and
media about political or public issues also did not change her own behaviors (Interview 1). The
participant’s roommate is passionate about women’s rights and had contacted officials and media
107
regarding issues of domestic abuse legislations multiple times. While the participant shares with
her roommate’s values, she explained her reasons of not doing the same:
“Rationally speaking, we all know contacting officials or media is
not going to lead to any changes in the near future. Are they going
to cancel the 30-day ‘cooling-off period’ when a woman wants a
divorce just because some college students emailed them? Of
course not. Yes, the government publishes drafts of legislations
and solicits comments from the public. But has anyone ever seen a
finalized version that is different from the draft? I don’t think so. I
understand my roommate keeps contacting People’s Congress
representatives because she genuinely believes it would bring
changes. But for me, I don’t think it would achieve anything other
than the good feeling that I’m fighting the good fight. It’s a good
effort, but also a futile effort.” (Interview 1)
Like the case mentioned in Interview 24, the participant of Interview 1 and her roommate
have very different utility functions despite their shared values about women’s rights. The
participant’s perceived benefits of contacting officials or media depends more heavily on the
actual results of the actions, whereas her roommate’s perceived benefits also include the
psychological benefits of enforcing her beliefs. Therefore, while the participant was influenced
by her roommate’s values regarding women’s rights issues, the increased perceived benefits
were not enough to alter her decision.
Secondly, peer effects will not occur if one’s perceived costs of participating in a form of
civic engagement decrease, but not to a point where the perceived benefits outweigh them. For
instance, a participant has a randomly assigned roommate who actively boycotts Starbucks and
tries to convince him to join the boycott. His roommate boycotted Starbucks primarily due to the
Philadelphia arrests in 2018 where Starbucks employees called the police and accused two
African American men of trespassing, which his roommate saw as a sign of the underlying racist
culture at Starbucks (Interview 31, 32). While his roommate explained his reasons for
108
boycotting Starbucks, the participant did not change his decision of not being part of the
boycotting. He explained his decision:
“First, let me make it clear. I don’t support racism, not at all. I just
don’t think boycotting Starbucks is the way to end racism. It’s not
going to lead to any real changes. If I joined him and boycotted
Starbucks, I would just have to walk all the way to the other side of
the campus for a cup of coffee before all my morning classes for
no reason. Most of my classes are in the same building as the
Starbuck! Even then, how do you know you are not inadvertently
supporting some other terrible things like homophobia, child labor,
or whatever? ……I know he gets upset when people go to
Starbucks. I try not to bring my Starbucks drinks back to the dorm.
But that’s all I’m gonna do. ” (Interview 32)
In this case, the participant’s perceived costs of not boycotting Starbucks increased
because his relationship with his roommate was negatively affected. However, the changed
perceived costs of not boycotting Starbucks still were not high enough to outweigh the other
costs of boycotting for the participant to change his behavior.
Thirdly, other than the reasons discussed previously, there is another important possible
reason why sometimes roommates’ active participation in certain specific forms of civic
engagement fails to affect a participant’s participation, namely, regular interaction and
communication between the roommates regarding said forms of civic engagement. If an
individual is not aware of her roommates’ civic engagement behaviors and attitudes, neither of
the two previously discussed mechanisms can take effect.
For instance, the participant in Interview 5 is one of the few Chinese participants who
have participated in protests. Two of her roommates also participate in the in-depth interview,
but neither of them is aware of the participant’s participation in protests
4
(Interview 6, 7). Both
4
The question regarding protest participation is part of a standard list of similar questions about
all forms of civic engagement to ensure the participant’s privacy. The participants were not given
any information regarding their roommates’ participation in protests in the interviews.
109
roommates indicated they had never participated in a protest and would not consider
participating in the future in both surveys. As for the participant who has participated in protests,
she explains,
“I guess it never came up in our conversations. No one just
randomly talks about protesting on the street with her roommates
unless it’s something they do together. Also, while I trust my
roommates one hundred percent, it’s still probably not a good idea
to go about and advertise my participation in protests. I’m not
ashamed of what I did. But I still need to look out for myself.”
(Interview 5)
In this case, Interviewee 5’s participation in protests could not influence her roommates
because they were not even aware of her actions. Therefore, it would be impossible for the
roommates to adjust their perceived benefits or costs of protesting or to change their behavior
regarding protesting. Other than intentional self-censoring, sometimes roommates do not
communicate about certain forms of civic engagement simply because they do not bring it up in
daily conversations. For instance, multiple interviewees never shared their party identification or
noticed their roommates’ (Interview 25, 26, 31, 32).
5.2.4. Summary
In sum, the in-depth interviews reveal two main mechanisms of peer effects in civic
engagement, namely, increasing the perceived benefits and decreasing the perceived benefits of a
given form of civic engagement to the point where the former outweigh the latter. These two
mechanisms offer plausible explanations of the main findings in the quasi-experiment that
positive peer effects exist in overall civic engagement and all three categories of civic
engagement in both the U.S. and China.
In addition, by investigating reasons why peer effects are absent in certain specific forms
of civic engagement in the quasi-experiment, the in-depth interviews also reveal potential
110
mechanisms of how institutional, contextual, and individualistic factors jointly shape one’s civic
engagement. While peers who have higher levels of civic engagement influence an individual’s
utility function by increasing her perceived benefits or decreasing her perceived costs of civic
engagement, the updated perceived benefits might still not outweigh the updated perceived costs
due to low initial perceived benefits or the high initial perceived costs caused by institutional and
individualistic factors. Moreover, peer effects do not occur if peers fail to interact or
communicate on a regular basis about certain forms of civic engagement for various reasons.
111
Chapter 6 Conclusion
The main purpose of this dissertation is to causally examine how peers’ civic engagement
affect an individual’s civic engagement. In addition, this study also investigates the robustness of
peer effects in civic engagement in different institutional settings and explores the mechanisms
of peer effects. Using a quasi-experimental design that takes advantage of random roommate
assignment in several American and Chinese universities, I test the first four hypotheses with
two samples from China and the United States. I also utilize a series of in-depth interviews to
verify the quantitative results from the quasi-experiment and to further explore the mechanisms
of peer effects. This chapter first summarizes the findings of this study. I then discuss the
theoretical, methodological, and practical implications of this study. In the last two sections, I
consider the limitations of this study and provide suggestions for future research.
This chapter begins with a discussion that interprets the results of the quasi-experiment,
situating the quantitative findings in the theoretical framework of this study. Drawing on the
results of the in-depth semi-structured interviews, the second section proposes possible
mechanisms of peer effects.
6.1. Summary of Findings
Evidence from the quasi-experiment shows that one’s peers’ civic engagement indeed
affects her own civic engagement, namely, participants with roommates who have high level of
civic engagement are more likely to become more active in civic engagement as well. Such peer
effects on overall level of civic engagement apply to participants in both China and the U.S., two
countries with drastically different macro-level political institutions. Furthermore, the peer
effects hold among all three categories of civic engagement, including civic participation,
political voice, and formal political participation for samples from both countries. The results of
112
the quasi-experiment demonstrate that the civic engagement of one’s peers has positive causal
effects on her own civic engagement.
Moreover, it is worth noticing that the peer effects observed in the quasi-experiment,
while robustly applicable to overall civic engagement and the three main categories of civic
engagement for both samples, are not uniform across all specific forms of civic engagement. The
quasi-experiment measures 17 forms of civic engagement, among which two are unique for the
Chinese sample and two are unique for the American Sample. Out of the 13 shared forms of
civic engagement measured in this study, roommates’ pre-treatment level is a statistically
significant predictor of seven for both samples. For the rest of indicators, roommates’ pre-
treatment level is a statistically significant predictor of a participant’s volunteering for a non-
electoral organization, boycotting, and membership in a political party only among Chinese
participants. For forms of civic engagement such as contacting officials, contacting the media,
and protesting, roommates’ pre-treatment level is a a statistically significant predictor for
participants in the U.S. sample. In terms of the 4 country-specific formal political indictors, peers
are not a statistically significant predictor of either unique formal political indicators for Chinese
participants. Whereas for American participants, peers are a statistically significant predictor of
political campaign contributions, but not volunteering for a political candidate or organization.
The in-depth interviews further investigate the variations of peer effects found in
different forms of civic engagement in the quasi-experiment and find two primary mechanisms
through which peers influence an individual’s civic engagement, namely, updating her perceive
benefits or costs of a given form of civic engagement. Interactions with one’s roommates who
participate in a given form of civic engagement update her perceived costs and benefits of the
form of civic engagement in question. These mechanisms offer possible explanations for the
113
variations of peer effects observed in different specific forms of civic engagement. Positive peer
effects occur when the updated perceived benefits outweigh the costs. When the updated benefits
fail to offset the updated costs, positive peer effects do not occur. When an individual reevaluates
the costs and benefits of a given form of civic engagement, the macro-level institutional and
micro-level individualistic factors also play an important role as they jointly determine the
baseline costs and benefits of a given form of civic engagement for the individual.
6.2. Implications
The findings of this study have several theoretical, methodological, and practical
implications. In terms of theoretical implications, this study introduces a theoretical framework
that draws on the strengths of the socio-ecological model and the rational choice theories, where
the former provides a broad conceptual context and the latter a comprehensive empirical
guideline, for the analysis of peer effects in civic engagement. Situating peer effects in this
conceptually broad yet empirically feasible framework allows this study to take into
consideration individual, social, and institutional factors as well as their interactions when
examining peer effects in civic engagement. This theoretical framework not only provides a
potential opportunity for future research to comprehensively examine all different determinants
of civic engagement, but also offers a possible approach to dynamically examine one specific
determinant of civic engagement while accounting for its interaction with other factors.
Methodologically, the current study demonstrates the methodological strengths of quasi-
experiments and mixed method analyses in social science studies. Quasi-experimental conditions
such as random roommate assignment provide valuable opportunities to conduct causal inference
while observing phenomena in their natural setting. The exogenous shock introduced by the
random roommate assignment allows this study to eliminate correlational effects caused by self-
114
selection and causally demonstrate peers’ influence on an individual’s civic engagement, paving
the path of causal inference on social and contextual factors in the study of civic engagement.
Moreover, leveraging the strengths of most different systems comparative design, this study is
able to demonstrate the robustness of the positive peer effects observed in the quasi-experiment
with samples from China and the U.S., two countries with drastically different political
institutions. In addition, the nested analysis design that combines a quasi-experiment and a series
of in-depth interviews makes it possible for the current study to both explore patterns and assess
their plausibility.
Furthermore, the current study offers practical implications for promoting civic
engagement. As the findings of this study show, college students with roommates who are more
active in civic engagement tend to become more active in civic engagement as well. This points
towards a new way to promote civic engagement. Governments, non-profit organizations,
schools, and other entities that wish to increase the level of civic engagement can devise
programs where individuals who are more and less active in civic engagement are assigned into
peer groups.
6.3. Limitations
Extant research has extensively studied civic engagement and its determinants, but little
efforts have been devoted to examining the social dimension of civic engagement or causally
demonstrating the links between different factors and civic engagement. By employing a
theoretical framework that builds upon the socio-ecological model and rational choice theories,
this study situates the often neglected peer effects in a more comprehensive context. In addition,
this study takes advantage of random roommate assignment in some colleges and utilizes a
quasi-experimental design to demonstrate the causality in peer effects, extending previous
115
research by drawing causal relationships between peers and one’s own behavior. Furthermore,
the most different systems comparative design ensures the robustness and generalizability of the
findings in this study. However, it is important to note the limitations of this study, particularly
in terms of methodology and scope.
First, since all participants in this study are first-year college students from the U.S. and
China, the generalizability of the findings might be limited. While this study takes advantage of
random roommate assignment in colleges and demonstrates the internal validity of peer effects
civic engagement, the external validity of the results is to a certain extent sacrificed due to the
lack of representativeness in the samples. College students are mostly young and well-educated
adults whose background information tend to be more homogenous than the general public. In
addition, due to their age and other characteristics, it is relatively uncommon for college students
to participate in certain forms of civic engagement, such as running for office, being a
membership of a trade union, etc. Moreover, the students who participated in this study are likely
to be more open to behavioral and attitudinal changes than older adults as exploring new options
is often an important part of college life. Also, on average, it is more common for college
roommates to interact and communicate regularly over an extended period of time than
individuals in many other peer groups, which might potentially make peer effects among college
roommates stronger than in other peer groups. Furthermore, while the most different systems
design demonstrates the robustness of peer effects in civic engagement in drastically different
institutional settings, China and the U.S. are in no way representative of all countries in the
world. Therefore, caution should be taken when generalizing the findings to other demographic
groups and institutional settings until future research replicates this study on different samples.
116
Second, despite the strong internal validity compared to conventional correlational
studies of civic engagement, the causality of peer effects demonstrated in the quasi-experiment in
this study is not as strong or precise as a true randomized controlled experiment. Since this study
is still an observational research, it does not fully control for all extraneous variables and is more
susceptible to confounding variables. Further research is needed to fully rule out such noises in
peer effects. In addition, while this study strives to maintain internal validity through the research
design and demonstrates meaningful findings, the results of this study fail to distinguish between
endogenous and exogenous peer effects. Further research is needed if scholars are interested in
parsing out the endogenous peer effects where one’s civic engagement is solely influenced by
their peers’ civic engagement and the exogenous effects where one’s civic engagement is shaped
by the exogenous characteristics of their peers.
Third, the outcome variables for civic engagement measure both behavioral intention and
actual participation. In other words, the positive peer effects for some participants could be the
shift from having no intention to participate in civic engagement to willing to participate in the
future. Due to the scope of this study, I only collect data at two time points and cannot verify if
the increased behavioral intention will eventually transform to actual participation with long-
term data. Therefore, the strength of the positive peer effects observed in this study might be
weakened by the “intention-behavior gap”, where behavioral intentions fail to translate into
action. However, this limitation does not invalidate the positive peer effects demonstrated in this
study. On the one hand, the shift in civic engagement intentions is neither the sole source nor the
most significant source of positive peer effects. On the other hand, empirical research has
demonstrated that political intentions and actions are tightly correlated and that changes in
intentions sufficiently predict behavioral changes in the future among young adults (e.g. Eckstein
117
et al., 2013, Pancer et al., 2007). Thus, even though the “intention-behavior gap” might exist in
this study, its impact on the findings is likely to be marginal.
The limitations discussed above also offer potential opportunities for future research,
which the next section discusses in further details along with other suggestions for future work.
6.4. Future Research
This section provides suggestions for future research based on this study. First, an
important line of future work would be to replicate this study on samples from more diverse
demographic groups and institutional settings. While this study contributes to the existing
literature by causally demonstrating the relatively under researched peer effects in civic
engagement, the generalizability of the findings in this study remains limited due to the college
student samples from two countries. Replication of this study with college student samples from
other countries would serve as a desirable starting point as it would be relatively easy and
inexpensive. Replication of this study with samples from other demographic groups might be
more challenging as quasi-experiment opportunities such as the random roommate assignment in
college are rare. Yet, it is still of great importance not only because it could help fully
demonstrate the external validity of peer effects in civic engagement, but also because it can
potentially provide more practical implications to promote civic engagement among the general
public.
Second, future research should also seek to employ research designs that distinguish
between the endogenous and exogenous peer effects. The positive peer effects observed in this
study include both the influences of ones’ peer’s civic engagement and their other characteristics.
While this does not invalidate this study, it is theoretically and methodologically interesting for
future work to tease out the endogenous peer effects. Ideally, a randomized controlled trial
118
(RCT) would be the most methodologically rigorous approach the examine the endogenous peer
effects. Yet, since it is essentially impossible to control for all other factors of civic engagement,
a field experiment that manipulates one’s peers’ civic engagement might be more desirable and
feasible to supplement this study than RCT. One caveat is that scholars should be mindful when
manipulating participants’ civic engagement through financial incentives or other means,
especially in authoritarian regimes where certain forms of civic engagement can impose political
risks on the participants.
Third, future research should conduct longitudinal research over a longer period time in
order to test long-term peer effects. For instance, future work could explore how peer effects
change as the time participants live with their roommates increases beyond two semesters. It
would also be interesting to conduct follow-up surveys after participants stop living with their
roommates to examine if the peer effects still hold over time.
Fourthly, future research should put more effort into investigating the mechanisms of
peer effects in different peer groups. While this study proposes two mechanisms of peer effects
through in-depth interviews, it is only a start and a lot more work needs to be done to fully
understand how peers influence each other. Can peers negatively influence one another in certain
scenarios? Are there any conditions for peer effects to occur? Do the mechanisms of peer effects
change in different peer groups? Understanding the mechanisms of peer effects could help
answering these questions.
Lastly, an ambitious yet promising avenue for future research would be to
comprehensively examine how individualistic demographic and psychological factors on the
micro level, institutional factors on the macro level, and social network contextual factors on the
119
contextual level jointly shape civic engagement. With sufficient resources, such research would
be crucial to understanding civic engagement.
Civic engagement is not yet another buzzword. It represents the most important ways an
individual can provide input to the public realm, both the civil society and the political system. It
is a channel through which everyone could take part in shaping the future of their communities,
their countries, and the world. This study demonstrates the positive links between one’s own and
her peers’ civic engagement, contributing new methods and theoretical insights to enhance the
understanding of civic engagement and its determinants. However, there still numerous questions
left unanswered in this field. More scholarly attention should be devoted to causally,
comprehensively, and dynamically understanding civic engagement and its determinants.
120
References
Acemoglu, D. and Angrist, J. (2001) “How Large Are Human-Capital Externalities? Evidence
from Compulsory-Schooling Laws”, in B.S. Bernanke and K. Rogoff (eds.) NBER
Macroeconomics Annual, Vol. 15, MIT Press.
Adler, R. P. & Goggin, J. (2005). “What Do We Mean by “Civic Engagement”? Journal of
Transformative Education, 3(3), 236–253.
Angrist, J.D. (2014) “The Perils of Peer Effects” Labour Economics, 30: 98-108.
Angrist, J.D., Lang, K. (2004) “Does School Integration Generate Peer Effects? Evidence from
Boston’s METCO Program” American Economic Review. Vol. 94: 1613–1634.
Anckar, C. (2008) “On the Applicability of the Most Similar Systems Design and the Most
Different Systems Design in Comparative Research.” International Journal of Social
Research Methodology. 11(5): 389-401.
Armstrong, C. K. (Eds.) (2007) Korean Society: Civil Society, Democracy and the State.
London: Routledge.
Astin, A.W., Sax, L.J., and Avalos, J. (1999). “Long Term Effects of Volunteerism During
Undergraduate Years.” Review of Higher Education, 22(2): 187-202.
Azevedo, C.N., Menezes, I. (2007) “Learning Politics Beyond Cognition: The Role of
Experience and Participation in Political Development” in N. Kryger and B. Ravn (Eds.)
Learning Beyond Cognition. Copenhagen: Danish University of Education Press.
Babcock, P.S. and Hartman J.L. (2010) “Networks and Workouts: Treatment Size and Status
Specific Peer Effects in A Randomized Field Experiment.” National Bureau of Economic
Research. Working Paper. No.16581.
Ballard, P.J., Ni, X., and Brocato, N. (2020). “Political Engagement and Wellbeing Among
College Students.” Journal of Applied Development Psychology. 71: 101209.
https://doi.org/10/1016/j.appdev.2020.101209.
Barber, B. (1999). A Place for Us: How to Make Society Civil and Democracy Strong. NJ:
Princeton University Press.
Barrett, M., & Zani, B. (2015) Political and Civic Engagement: Multidisciplinary Perspectives.
New York: Routledge.
Beck, P.A. (2002) “Encouraging Political Defection: The Role of Personal Discussion Networks
in Partisan Desertions to The Opposition Party and Perot Votes In 1992.” Political
Behavior, 24, 309–37.
Berger, B. (2009). “Political Theory, Political Science, and the End of Civic Engagement.”
121
Perspectives on Politics, 7(2), 335–350.
Berman, S. (1997). “Civil Society and Political Institutionalization.” American Behavioral
Scientist. 40(5), 562-574.
Berman, S. (1997). “Civil Society and the Collapse of the Weimar Republic.” World politics. 49,
401-429.
Bernhagen, P., & Marsh, M. (2007) “Voting and Protesting: Explaining Citizen Participation in
Old and New European Democracies” Democratization. 14: 44-72
Berry, J., Portinga, Y. H., Segall, M. H., Dasen, P. R. (2002) Cross-Cultural Psychology:
Research and Applications, 2nd Edition. Cambridge: Cambridge University Press.
Billings, D. (2014). “School Segregation, Educational Attainment, and Crime: Evidence from the
End of Busing in Charlotte-Mecklenburg.” The Quarterly Journal of Economics, 129(1),
435–476.
Boisjoly, D. (2006). “Empathy or Antipathy? The Impact of Diversity.” The American Economic
Review, 96(5), 1890–1905.
Brady, H. (1999). “Political Participation.” In J. P. Robinson, P.R. Shaver, L. S. Wrightsman
(Eds.). Measures of Political Attitudes, (pp. 737-801). San Diego: Academic Press.
Brady, H., Verba, S., and Scholzman, K. (1995). “Beyond SES: A Resource Model for Political
Participation.” American Political Science Review, 89(2): 271-294.
Brehm, J. and Rahn, W. (1997). “Individual-level Evidence for the Causes and Consequences of
Social Capital.” American Journal of Political Science, 41(3): 999-1023.
Bronfenbrenner, U. (1977). “Toward an Experimental Ecology of Human Development”.
American Psychologist, 32(7), 513-531
Bronfenbrenner, U. (1979). The Ecology of Human Development. Cambridge, MA: Harvard
University Press.
Bronfenbrenner, U. (1995). “Developmental Ecology Through Space and Time: A Future
Perspective”. In P. Moen, G. H. Elder, Jr., & K. Lüscher (Eds.), Examining Lives in
Context: Perspectives on The Ecology Of Human Development (pp. 619-647).
Washington, DC, US: American Psychological Association.
Bronfenbrenner, U. (2009). The Ecology of Human Development: Experiments by Nature and
Design. Cambridge, MA: Harvard University Press.
Bronfenbrenner, U., & Ceci, S. J. (1994). “Nature-Nuture Reconceptualized in Developmental
Perspective: A Bioecological Model”. Psychological Review, 101(4), 568-586.
Bronfenbrenner, U., & Morris, P. A. (1998). “The Ecology of Developmental Processes”. In W.
122
Damon & R. M. Lerner (Eds.), Handbook of Child psychology: Theoretical Models of
Human Development (pp. 993–1028). John Wiley & Sons Inc.
Bronfenbrenner, U., & Morris, P. A. (2006). The Bioecological Model of Human
Development. In R. M. Lerner and W. Damon (Eds.), Handbook of child psychology:
Theoretical models of human development (pp. 793–828). John Wiley & Sons Inc.
Campos, C.F.S., Heap, S.H., and de Leon, F.L.L. (2017) “The Political Influence of Peer
Groups: Experimental Evidence in the Classroom.” Oxford Economic Papers. 69(4), 963-
985.
Campbell, D. E. (2013) “Social Networks and Political Participation”. Annual Review of
Political Science. Vol. 16: 33-48
Carrell, S.E., Fullerton, R.L., West, J.E. (2009) “Does Your Cohort Matter? Measuring Peer
Effects in College Achievement”. Journal of Labor Economics. Vol. 27: 439–464.
Carrell, S.E., Malmstrom, F.V., West, J.E. (2008) “Peer Effects in Academic Cheating” The
Journal of Human Resources. Vol. 43: 173–207.
Carrell, S.E., Sacerdote, B.I., West, J.E. (2013) “From Natural Variation to Optimal Policy? The
Importance of Endogenous Peer Group Formation”. Econometrica. Vol. 81:855–882
Case, A.C., Katz, L. F. (1991) “The Company You Keep: The Effects of Family and
Neighborhood on Disadvantaged Youths” National Bureau of Economic Research.
Working Paper No. 3705.
Cavatorta, F. (Eds.) (2012) Civil Society Activism Under Authoritarian Rule: A Comparative
Perspective. Routledge.
Chan, A. (1993) “Revolution or Corporatism? Workers and Trade Unions in Post-Mao China.”
The Australian Journal of Chinese Affairs, 31-61.
Chen, W. and Wellman, B. (2009). “Net and Jet: The Internet Use, Travel and Social Networks
of Chinese Canadian Entrepreneurs.” Information Communication & Society. 12(4): 525-
547.
Cho, W. (2003) “Contagion Effects and Ethnic Contribution Networks.” American Journal of
Political Science, 43: 368-387.
Cho, W., Gimpel, J., and Dyck, J. (2006) “Residential Concentration, Political Socialization, and
Voter Turnout.” The Journal of Politics, 68: 156-167.
Choi, D.H., Nah, S., and Chung, D.S. (2021). “Social Media as Civic Mobilizer: Community
Storytelling Network, Social Media, and Civic Engagement in South Korea.” Journal of
Broadcasting & Electronic Media. 65(1): 46-65.
Chong, D. (1994) “ Tolerance and Social Adjustment to New Norms and Practices.” Political
123
Behavior. 16(1): 21-53.
Christakis, N.A., Fowler, J. H. 2007. “The Spread of Obesity in A Large Social Network Over 32
Years”. The New England Journal of Medicine. Vol. 357:370–379
Cnaan, R. A., & Park, S. (2016). “The Multifaceted Nature of Civic Participation: A Literature
Review.” Voluntaristics Review, 1(1), 1–73.
Cohen, J. L., & Arato, A. (1992). Civil Society and Political Theory. MIT Press.
Cohen-Cole E. and Fletcher J.M. (2008) “Is Obesity Contagious? Social Networks Vs.
Environmental Factors in The Obesity Epidemic.” Journal of Health Economics.
27:1382–1387
Costa, D. L., & Kahn, M. E. (2003). “Civic Engagement and Community Heterogeneity: An
Economist’s Perspective.” Perspectives on Politics, 1(1), 103-111.
Da Silva, L., Sanson, A., Smart, D., & Toumbourou, J. (2004). “Civic Responsibility Among
Australian Adolescents: Testing Two Competing Models.” Journal of Community
Psychology, 32(3), 229-255.
Dahl, R. A. (1998). On Democracy. New Haven, CT: Yale University Press.
Delli Carpini, M.X., Cook, F.L., and Jacobs, L.R. (2004). “Public Deliberation, Discursive
Participation, and Citizen Engagement: A Review of the Empirical Literature.” Annual
Review of Political Science. 7: 315-344.
Delli Carpini, M.X., & Keeter, S. (1996). What Americans Know About Politics and Why It
Matters. Yale University Press.
Dey, E. (1996). “Undergraduate Political Attitudes: An Examination of Peer, Faculty, and Social
Influences.” Research in Higher Education, 37(5): 535-554.
Diamond, L. (1999). Developing Democracy: Toward Consolidation. Johns Hopkins University
Press.
Diemer, M.A. and Li, C.H. (2011). “Critical Consciousness Development and Political
Participation Among Marginalized Youth.” Child Development. 82(6): 1815-1833.
Diller, E. C. (2001). Citizens in Service: The Challenge of Delivering Civic Engagement
Training to National Service Programs. Washington, DC: Corporation for National and
Community Service.
Ding, X. L. (1994). “Institutional Amphibiousness and The Transition from Communism: The
Case of China.” British Journal of Political Science. 24(3), 293- 318.
Dishion, T. J., Andrews, D. W., & Crosby, L. (1995). “Antisocial Boys and Their Friends in
Early Adolescence: Relationship Characteristics, Quality, and Interactional Process.”
124
Child Development, 66(1), 139-151.
Downs, A. (1957) An Economic Theory of Democracy. New York: HarperCollins.
Duke, N.N., Skay, C.L., Pettingell, S.L., and Borowsky, I.W. (2009). “From Adolescent
Connections to Social Capital: Predictors of Civic Engagement in Young Adulthood.”
Journal of Adolescent Health. 44(2): 161-168.
Duncan, G.J., Boisjoly, J. Kremer, M., Levy, D.M., and Eccles, J. (2005) “Peer Effects in Drug
Use and Sex Among College Students.” Journal of Abnormal Child Psychology. 33(3):
375-385.
Eckstein, K., Noack, P., Gniewosz, B. (2013) “Predictors of Intentions to Participate in Politics
and Actual Political Behaviors in Young Adulthood” International Journal of Behavioral
Development, 37(5): 428-435.
Evans, C.B., Cotter, K.L., Rose, R.A., Smokowski, P.R. (2016). “Substance Use in Rural
Adolescents: The Impact of Social Capital, Anti-social Capital, and Social Capital
Deprivation.” Journal of Addictive Diseases. 35(4): 244-257.
Flanagan, C.A. (2004). “Volunteerism, Leadership, Political Socialization, and Civic
Engagement.” In R.M. Lerner and L. Steinberg (Eds.) Handbook of Adolescent
Psychology. (pp. 721-745). Hoboken, NJ: John Wiley.
Flanagan, C.A., Campbell, B., Botcheva, L., Bowes, J., Csapo, B., Macek, P., Sheblanova, E.
(2003) “Social Class and Adolescents’ Beliefs about Justice in Different Social Orders”
Journal of Social Issues, 59: 711-732.
Foster G. (2006). “It’s Not Your Peers, and It’s Not Your Friends: Some Progress Toward
Understanding the Educational Peer Effect Mechanism.” Journal of Public Economics.
90:1455–75
Fung, A. (2003). “Associations and Democracy: Between Theories, Hopes, and Realities.”
Annual Review of Sociology. 515-539.
Gallagher, M. (2004). “The Limits of Civil Society in a Late Leninist State”. Civil Society and
Political Change in Asia: Expanding and Contracting Democratic Space, 419-52.
Gerber, A.S., Green, D. P., Larimer, C. W. (2008) “Social Pressure and Voter Turnout: Evidence
from A Large-Scale Field Experiment” American Political Science Review. Vol. 102,
No.1
Gil de Zúñiga, H., Valenzuela, S., and Weeks, B.E. (2016). “Motivations for Political
Discussion: Antecedents and Consequences on Civic Engagement.” Human
Communication Research. 42: 533-552.
Gleason, M. A., and Siegfried, J. J. (2003). “Academic Roommate Peer Effects.” Unpublished
manuscript, Vanderbilt University.
125
Goldman, M. (2005). From Comrade to Citizen: The Struggle for Political Rights in China.
Harvard University Press.
Gordon, H.R. and Taft, J.K. (2011). “Rethinking Youth Political Socialization: Teenage Activists
Talk Back.” Youth and Society. 43(4): 1499-1527.
Gould, L. (2009). “Does Immigration Affect The Long-Term Educational Outcomes of Natives?:
Quasi-Experimental Evidence.” The Economic Journal (London), 119(540), 1243–1269.
Green, D. P., & Gerber, A. S. (2004) Get Out the Vote! How to Increase Voter Turnout.
Washington, DC: Brookings Institution Press.
Greenberg, E.S. (2008) “Spillover from Cooperative and Democratic Workplaces: Have the
Benefits Been Oversold?” in B. A. Sullivan, M. Snyder, J.L. Sullivan (Eds.),
Cooperation: The Political Psychology of Effective Human Interaction (pp. 219-239).
Malden: Blackwell.
Hassid, J. (2012). “Safety Valve or Pressure Cooker? Blogs in Chinese Political Life.” Journal of
Communication. 62(2): 212-230.
Harell, A., Stolle, D., & Quintelier, E. (2008, July). “Network Diversity and Political
Participation: A Complication or Asset?” Paper presented at the Conference on
Comparative Perspectives on Political Socialization, Bruges, Belgium, July 3-4, 2008.
Hildebrandt, T. (2011). “The Political Economy of Social Organization Registration in China.”
The China Quarterly, 208, 970-989.
Hildebrandt, T. (2013). Social Organizations and the Authoritarian State in China. Cambridge
University Press.
Ho, M. (2012) “Sponsoring Civil Society: State and Community Movement in Taiwan”
Sociological Inquiry. Vol. 83(3): 404-423.
Ho, P. & Edmonds, R. L. (Eds.) (2008) China’s Embedded Activism: Opportunities and
Constraints of a Social Movement. Routledge.
Hoel, Jessica, J. Parker and J. Rivenburg. (2006). “A Test for Classmate Peer Effects in Higher
Education.” Reed College Working Paper.
Hoxby, C. (2000). “Peer Effects in The Classroom: Learning from Gender and Race Variation”.
National Bureau of Economic Research. Working Paper No. 7867.
Hsieh, Y.P. and Li, M. (2014). “Online Political Participation, Civic Talk, and Media
Multiplexity: How Taiwanese Citizens Express Political Opinions on the Web.”
Information, Communication & Society. 17(1): 26-44.
Huckfeldt, R.R., Spragues, J. (1987). “Networks in Context: The Social Flow of Political
Information.” American Political Science Review. 81: 1197-1216.
126
Huckfeldt, R.R., Spragues, J. (1995) Cambridge Studies in Political Psychology and Public
Opinion. Citizens, Politics, and Social Communication: Information and Influence in an
Election Campaign. New York: Cambridge University Press.
Imberman, S.A., Kugler, A.D., Sacerdote, B.I. (2012) “Katrina’s Children: Evidence on The
Structure of Peer Effects from Hurricane Evacuees”. American Economic. Vol. 102:
2048–2082.
Inglehart, R., & Norris, P. (2003) Rising Tide: Gender Equality and Cultural Change Around the
World. New York and Cambridge: Cambridge University Press.
Jennings, M.K. (2002) “Generations Units and the student Protest Movement in the United
States: An Intra- and Intergenerational Analysis” Political Psychology. 23: 303-324.
Kandel, D.B. (1978). “Homophily, Selection, and Socialization in Adolescent Friendships.”
American Journal of Sociology. 84: 427-436.
Kavanaugh, A. (2002) “Community Networks and Civic Engagement: A Social Network
Approach: Social Networks and Computer Networks.” The Good Society. 11(3): 17-24.
Keeter, S., Zukin, C., Andolina, M., & Jenkins, M. (2002). “The Civic and Political Health of
The Nation: A Generational Portrait.” College Park, MD: Center for Information and
Research on Civic Learning and Engagement (CIRCLE), School of Public Policy,
University of Maryland.
Kenny, C.B. (1994) “The Microenvironment of Attitude Change.” The Journal of Politics, 56,
715–28.
Kling, L. (2005). “Neighborhood Effects on Crime for Female and Male Youth: Evidence from a
Randomized Housing Voucher Experiment.” The Quarterly Journal of
Economics, 120(1), 87–130.
Klofstad, C.A. (2009) “Civic Talk and Civic Participation: The Moderating Effect of Individual
Predispositions.” American Politics Research, 37: 856-878.
Klofstad, C.A. (2010). “The Lasting Effect of Civic Talk on Civic Participation: Evidence from
A Panel Study.” Social Forces, 88, 2353–2375.
Klofstad, C.A. (2011). Civic Talk: Peers, Politics, and the Future of Democracy. Philadelphia,
PA: Temple University Press.
Klofstad, C.A. (2015). “Exposure to Political Discussion in College is Associated with Higher
Rates of Political Participation Over Time.” Political Communication. 00: 1-18.
Kreisi, H., Koopmans, R., Duyvendak, J. W., Guigni, M. G. (1995) New Social Movements in
Western Europe: A Comparative Analysis. Minneapolis: University of Minnesota Press.
Kremer, L. and Levy, D. (2008). “Peer Effects and Alcohol Use Among College Students.” The
127
Journal of Economic Perspectives, 22(3), 189–206.
Lake, R., & Huckfeldt, R. (1998). “Social Capital, Social Networks, and Political Participation.”
Political Psychology, 19(3): 567-584.
Lavy, V., Schlosser, A. (2011) “Mechanisms and Impacts of Gender Peer Effects at School”.
American Economic Journal: Applied Economics. Vol. 3: 1-33.
Leighley, J. (1990) “Social Interaction and Contextual Influences on Political Participation.”
American Politics Research. 18: 459-75.
Lieberman, E. S. (2005). “Nested Analysis as A Mixed-Method Strategy for Comparative
Research.” American Political Science Review, 99(3), 435-452
Lin, N., Cook, K., and Burt, R.S. (2001). Social Capital: Theory and Research. New York:
Aldine de Gruyter.
Lyle, D. (2007) “Estimating and Interpreting Peer and Role Model Effects from Randomly
Assigned Social Groups at West Point.” Review of Economics and Statistics, 89: 289-
299.
MacKuen, M. and Brown, C. (1987) “Political Context and Attitude Change.” American
Political Science Review, 81, 471–90.
Manski, C. F. (1993) “Identification of Endogenous Social Effects: The Reflection Problem” The
Review of Economic Studies. 60: 531-542.
McClurg, S. (2006) “The Electoral Relevance of Political Talk: Examining Disagreement and
Expertise Effects in Social Networks on Political Participation.” American Journal of
Political Science, 50: 737-54.
Metzger, A., Oosterhoff, B., Palmer, C.A., and Ferris, K. (2014). “Dimensions of Citizenship:
Associations Among Adolescents’ Socio-political Values and Civic Judgements.” PS:
Political Science & Politics. 47(2): 443-448.
Mill, J. S. (1874). System of Logic. New York: Harper & Brothers.
Moffitt, R.A. (2001) “Policy Interventions, Low-Level Equilibria, and Social Interactions.” In
S.N. Durlauf and H.P. Young (Eds.), Social Dynamics (pp. 45–82). Washington, DC:
Brookings Inst.
Mutz, D.C. (2002a) “The Consequences of Cross-Cutting Networks for Political Participation.”
American Journal of Political Science, 46, 838–55.
Mutz, D.C. and Mondak, J. (2006) “The Workplace as A Context for Cross-Cutting Political
Discourse.” The Journal of Politics, 68, 140–55.
Nickerson, D.W. (2008) “Is Voting Contagious? Evidence from Two Field Experiments.”
128
American Political Science Review, 102, 49–57.
Norris, P. (2001). Digital Divide: Civic Engagement, Information Poverty, and the Internet
Worldwide. Cambridge: Cambridge University Press.
Nisbet, M.C. and Scheufele, D.A. (2004). “ Political Talks as A Catalyst for Online Citizenship.”
Journalism & Mass Communication Quarterly. 81(4): 877-896.
O'Brien, K. J. (1996). “Rightful Resistance”, World Politics, 49, 31-55.
O'Brien, K. J., & Li, L. (2006). Rightful Resistance in Rural China. Cambridge University Press.
Olson, M. (1965) The Logic of Collective Action: Public Goods and the Theory of Groups.
Cambridge MA: Harvard University Press.
Oosterhoff, B., Alvis, L., Deutchman, D., Poppler A., and Palmer, C.A. (2021). “Civic
Development within the Peer Context: Associations Between Early Adolescent Social
Connectedness and Civic Engagement.” Journal of Youth and Adolescence. 50: 1870-
1883.
Pancer, S.M. (2015) The Psychology of Citizenship and Civic Engagement. Oxford University
Press.
Pancer, S. M., Pratt, M., Hunsberger, B., & Alisat, S. (2007). “Community and Political
Involvement in Adolescence: What Distinguishes the Activists from The Uninvolved?”
Journal of Community Psychology, 35(6), 741-759.
Passy, F. (2003). “Social Networks Matter, But How?” in M.D. Diani and D. McAdam (Eds.)
Social Movements and Networks: Relational Approaches to Collective Action. (pp. 21-
48). New York, NY: Oxford University Press.
Pattie, C. J., and Johnston, R.J. (2009) “Conversation, Disagreement and Political Participation.”
Political Behavior. 31: 261-85.
Popper, K. (1959). The Logic of Scientific Discovery. New York: Basic Books.
Przeworski, A. and Teune, H. (1970). The Logic of Comparative Social Inquiry. New York: John
Wiley & Sons.
Putnam, R. (2000). Bowling Alone: The Collapse and Revival of American Community. New
York: Simon & Schuster.
Putnam, R.D. (2001) “Social Capital: Measurement and Consequences” Isuma: Canadian
Journal of Policy Research. 2: 41-51.
Putnam R.D., Leonardi R, Nanetti R. (1993) Making Democracy Work: Civic Traditions in
Modern Italy. Princeton: Princeton University Press.
Quintelier, E., Stolle, D., and Harell, A. (2012) “Politics in Peer Groups: Exploring the Causal
129
Relationship between Network Diversity and Political Participation.” Political Research
Quarterly. 65(4): 868-881.
Riker, W. H. and Ordeshook, P. C. (1968) ‘A Theory of the Calculus of Voting’, American
Political Science Review, 62, 25–42.
Ronan. B. (2004). Testimony at the White House Conference on Aging Public Forum on Civic
Engagement in an Older America, Phoenix, Arizona, February 25, 2004.
Rossi, G., Lenzi, M., Sharkey, J.D., Vieno, A., and Santinello, M. (2016) “Factors Associated
with Civic Engagement in Adolescence: The Effects of Neighborhood, School, Family,
and Peer Contexts.” Journal of Community Psychology. 44(8): 1040-1058.
Ryan, A.M. (2000). “The Peer Group As A Context for the Socialization of Adolescents’
Motivation, Engagement, and Achievement in School.” Educational Psychologist. 35:
101-111.
Ryan, C. (2017) “Measurement of Peer Effects.” Australian Economic Review. 50(1): 121-129.
Sacerdote, B. (2000) “Peer Effects with Random Assignment: Results for Dartmouth
Roommates”. National Bureau of Economic Research. Working Paper No. 7469.
Sacerdote, B. (2001) “Peer Effects with Random Assignment: Results for Dartmouth
Roommates”. The Quarterly Journal of Economics. Vol. 116: 681–704.
Sacerdote, B. (2014) “Experimental and Quasi-Experimental Analysis of Peer Effects: Two
Steps Forward?” The Annual Review of Economics. Vol. 6: 253-272.
Sax, L.J. (2004) “Citizen Development and the American College Student.” In Dalton, J.,
Russell, T., and Kline, S. (Eds.), Assessing Character Outcomes in College. San
Francisco: Wiley Periodicals.
Scheufele, D. A., Hardy, B.W., Brossard, D., Waismel-Manor, I.S., and Nisbet, E. (2006)
“Democracy Based on Difference: Examining the Links between Structural
Heterogeneity of Discussion Networks, and Democratic Citizenship.” Journal of
Communication 56: 728-53.
Schulz, W., Ainley, J., Fraillon, J., Kerr, D., Losito, B. (2010) Initial Findings from the IEA
International Civic and Citizenship Education Study. Amsterdam: IEA.
Schwadel, P., Cheadle, J.E., Malone, S.E., and Stout, M. (2016). “Social Networks and Civic
Participation and Efficacy in Two Evangelical Protestant Churches.” Review of Religious
Research. 58: 305-317.
Settle, J.E., Bond, R., and Levitt, J. (2011). “The Social Origins of Adult Political Behavior.”
American Politics Research. 39(2): 239-263.
Shepsle, K.A. (1989) “Studying Institutions: Some Lessons from the Rational Choice Approach”
130
Journal of Theoretical Politics. 1(2), 131-147.
Sinclair, B. (2012). The Social Citizen: Political Networks and Political Behavior. Chicago, IL:
University of Chicago Press.
Skocpol, T. and Fiorina, M. (Eds.) Civic Engagement in American Democracy. Washington
D.C.: Brookings Institute.
Skocpol, T. (2003). Diminished Democracy: From Membership to Management in American
Civic Life. University of Oklahoma Press.
Steinberg, L., and Monahan K.C. (2007). “Age Differences in Resistance to Peer Influence.”
Development Psychology. 43(6): 1531-1543.
Stinebrickner, R. and Stinebrickner, T.R. (2006). “What Can Be Learned About Peer Effects
Using College Roommates? Evidence from New Survey Data and Students from
Disadvantaged Backgrounds.” Journal of Public Economics, 90(8), 1435–1454.
Stolle, S. (2007) “Social Capital” in R.Dalton and D. Klingemann (Eds.) Oxford Handbook of
Political Behavior (pp. 655-674). Oxford: Oxford University Press.
Sobel, R. (1993) “From Occupational Involvement to Political Participation: An Exploratory
Analysis.” Political Behavior. Vol. 15: 339-353.
Spires, A. J. (2008). China's Un-Official Civil Society: The Development of Grassroots NGOs in
an Authoritarian State. Yale University.
Spires, A. J. (2011). “Contingent Symbiosis and Civil Society in an Authoritarian State:
Understanding the Survival of China’s Grassroots NGOs.” American Journal of
Sociology. 117(1), 1-45.
Strother, L., Piston, S., Golberstein, E., Gollust, S.E., and Eisenberg, D. (2021). “College
Roommates Have A Modest but Significant Influence on Each Other’s Political
Ideology.” Proceedings of the National Academy of Sciences of the United States of
America, 118(2), e2015514117. https://doi.org/10.1073/pnas.2015514117
Tismăneanu, V. (1990) In Search for Civil Society: Peace Movements in the Soviet Bloc.
Routledge.
Tocqueville, A. (1988) Democracy in America, translated by George Lawrence. Edited by J. P.
Mayer. New York: Harper Perennial.
Tsai, L. (2007). Accountability Without Democracy: How Solidary Groups Provide Public
Goods in Rural China. Cambridge University Press.
Unger, J., & Chan, A. (1995). “China, corporatism, and the East Asian model.” The Australian
Journal of Chinese Affairs. (33), 29-53.
131
Uslaner, E.M. (2002). “Religion and Civic Engagement in Canada and the United States.”
Journal of Scientific Study of Religion. 41(2): 239-254.
Uslaner, E.M. (2004). “Trust, Civic Engagement, and the Internet”. Political Communication.
21(2): 223-242.
Verba, S. (1978) Participation and Political Equality. Cambridge: Cambridge University Press.
Verba, S., & Nie, N. H. (1972) Participation in America: Political Democracy and Social
Equality. Chicago: University of Chicago Press.
Verba, S., Scholzman, K.L., Brady, H. (1995) Voice and Equality: Civic Voluntarism in
American Politics. Cambridge: Harvard University Press.
Wegemer, C.M. (2021). “Service, Activism, and Friendships in High School: A Longitudinal
Social Network Analysis of Peer Influence and Critical Beliefs.” Journal of Youth and
Adolescence. 51(1): 1-15.
Wellman, B., Salaff. J., Dimitrova, D., Garton, L., Gulia, M., and Hythornthwaite, C. (1996).
“Computer Networks as Social Networks: Collaborative Work, Tele-work, and Virtual
Community.” Annual Review of Sociology. 22: 213-239.
Wentzel, .R., McNamara, C.C. (1999) “Interpersonal Relationships, Emotional Distress and
Prosocial Behavior in Middle School” Journal of Early Adolescence. Vol. 19: 114-125.
West, C., Carrel, S.E., and Hoekstra, M. (2011). “Is Poor Fitness Contagious?: Evidence from
Randomly Assigned Friends.” Journal of Public Economics, 95(7), 657–663.
Whiteley, P. and Seyd, P. (1996) ‘Rationality and Party Activism: Encompassing Tests of
Alternative Models of Political Participation’, European Journal of Political Research,
29, 215–34.
Whiteley, P. and Seyd, P. (1998) ‘The Dynamics of Party Activism in Britain: A Spiral of
Demobilization?’, British Journal of Political Science, 28, 113–37.
Wilson, Rick K. 2011. “The Contributions of Behavioral Economics to Political Science”.
Annual Review of Political Science. 14: 201-23.
Wojcieszak, M. and Mutz, D.C. (2009). “Online Groups and Political Discourse: Do Online
Discussion Spaces Facilitate Exposure to Political Disagreement?” Journal of
Communication. 59: 40-56.
Wolfinger, Raymond E., & Rosenstone, Steven. (1980) Who Votes? New Haven: Yale
University Press.
Xia, M., Li, X., and Tudge, J.R.H. (2020) “Operationalizing Urie Bronfenbrenner’s Process-
Person-Context-Time Model.” Human Development. 64: 10-20.
132
Yang, G. (2009). The Power of the Internet in China: Citizen Activism Online. New York:
Columbia University Press.
Yates, M., Youniss, J. (1998) “Community Service and Political Identity Development in
Adolescence”. Journal of Social Issues. Vol. 54: 495-512.
Youniss, J., Bales, S., Christmas-Best, V., Diversi, M., McLaughlin, W., and Silbereisen, R.
(2002). Youth Civic Engagement in the Twenty-first Century. Journal of Research on
Adolescence, 12(1), 121-148.
Zaff, J.F., Malanchuk, O., Eccles, J.S. (2008) “Predicting Positive Citizenship from Adolescence
to Young Adulthood: The Effects of a Civic Context”. Applied Developmental Science.
Vol. 1: 38-53.
Zaff, J.F., Hart, D., Flanagan, C.A., Youniss, J., and Levine, P. (2010). “Developing Civic
Engagement within A Civic Context.” In M.E. Lamb and A.M. Freund (Eds.) Social and
Emotional Development: The Handbook of Life-span Development, Vol 2. (pp. 590-630).
Hoboken, NJ: Wiley.
Zukin, C., Keeter, S., Andolina, M., Jenkins, K., Delli Carpini, M. X. (2006) A new
Engagement? Political Participation, Civic Life, and the Changing American Citizen.
New York: Oxford University Press.
133
Appendix A: Descriptive Statistics for Background Variables
Variable Code Variable Mean Std. D Min Max Obs.
All Participants
B-1 Age 19.22 0.64 18.00 24.00 1294
B-2 Female 0.40 0.49 0.00 1.00 1294
B-3 Ethnic Minority 0.35 0.48 0.00 1.00 1294
B-4 Dorm Size 3.55 1.60 2.00 8.00 1294
Sample from China
B-1 Age 19.19 0.62 18.00 21.00 642
B-2 Female 0.41 0.49 0.00 1.00 642
B-3 Ethnic Minority 0.09 0.29 0.00 1.00 642
B-4 Dorm Size 4.23 1.57 3.00 8.00 642
Sample from the U.S.
B-1 Age 19.25 0.65 18.00 24.00 652
B-2 Female 0.40 0.49 0.00 1.00 652
B-3 Ethnic Minority 0.61 0.49 0.00 1.00 652
B-4 Dorm Size 2.88 1.32 2.00 7.00 652
Note: The variable age is measured as the age of a participant in 2019. The variable Ethnic Minority refers to
non-Han Chinese participants in the Chinese sample and non-White participants in the American sample.
Dorm size is the number of students in a dorm, including the participant.
134
Appendix B: Descriptive Statistics for Civic Engagement Variables from the
First Wave of Survey in Fall 2018
Sample from China
Variable Code Variable Mean Std. D Min Max Obs.
Civic Indicators
CI-1
Community problem solving
(overall) 4.33 2.45 1.00 10.00 642
CI-2
Volunteering for a non-
electoral organization 5.54 2.99 1.00 10.00 642
CI-3
Membership in a group or
association (overall) 2.38 1.15 1.00 7.75 642
CI-4
Participation in charitable
fundraising 4.12 2.97 1.00 10.00 642
CI-Composite
Composite variable for all
civic indicators 16.36 7.92 4.00 34.05 642
Political Voice Indicators
PV-1 Contacting officials 2.92 2.53 1.00 10.00 642
PV-2
Contacting the media
regarding political issues 2.91 2.52 1.00 10.00 642
PV-3
Discussing political and public
issues on social media 3.39 2.55 1.00 10.00 642
PV-4 Protesting 1.45 1.31 1.00 10.00 642
PV-5 Petitioning 3.39 2.79 1.00 10.00 642
PV-6 Boycotting 3.45 2.86 1.00 10.00 642
PV-Composite
Composite variable for all
political voice indicators 17.52 11.03 6.00 53.25 642
Formal Political Indicators
FP-1 Voting (all) 2.18 1.89 1.00 10.00 642
FP-2
Membership in a political
party or organization 1.99 1.97 1.00 10.00 642
FP-3 Running for office (all) 2.09 1.23 1.00 10.00 642
FP-4
Taking an active role in the
ruling political party or the
government (China) 1.64 1.87 1.00 10.00 642
FP-5
Attending public hearings and
open meetings (China) 1.51 1.60 1.00 10.00 642
135
FP-Composite
Composite variable for all
formal political indicators 9.41 5.12 5.00 32.80 642
Total-Composite
Composite variable for all
civic engagement indicators 9.71 4.59 3.00 24.85 642
Sample from the U.S.
Variable Code Variable Mean Std. D Min Max Obs.
Civic Indicators
CI-1
Community problem solving
(overall) 4.95 2.08 1.00 10.00 652
CI-2
Volunteering for a non-
electoral organization 3.99 2.43 1.00 10.00 652
CI-3
Membership in a group or
association (overall) 3.45 1.31 1.00 7.75 652
CI-4
Participation in charitable
fundraising 5.11 2.79 1.00 10.00 652
CI-Composite
Composite variable for all
civic indicators 17.49 5.71 4.80 33.65 652
Political Voice Indicators
PV-1 Contacting officials 3.97 2.76 1.00 10.00 652
PV-2
Contacting the media
regarding political issues 2.84 2.31 1.00 10.00 652
PV-3
Discussing political and public
issues on social media 3.79 2.50 1.00 10.00 652
PV-4 Protesting 3.85 2.62 1.00 10.00 652
PV-5 Petitioning 5.20 2.88 1.00 10.00 652
PV-6 Boycotting 3.17 2.46 1.00 10.00 652
PV-Composite
Composite variable for all
political voice indicators 22.81 10.47 6.00 57.75 652
Formal Political Indicators
FP-1 Voting 3.92 2.73 1.00 10.00 652
FP-2
Membership in a political
party or organization 3.12 2.53 1.00 10.00 652
FP-3 Running for office 1.96 1.50 1.00 10.00 652
136
FP-4
Volunteering for candidate or
political organization (US) 3.23 2.53 1.00 10.00 652
FP-5 Campaign contributions (US) 3.12 2.52 1.00 10.00 652
FP-Composite
Composite variable for all
formal political indicators 15.36 7.84 5.00 42.12 652
Total-Composite
Composite variable for all
civic engagement indicators 12.48 4.54 3.24 26.32 652
137
Appendix C: Descriptive Statistics for Civic Engagement Variables from the
Second Wave of Survey in Spring 2019
Sample from China
Variable Code Variable Mean Std. D Min Max Obs.
Civic Indicators
CI-1
Community problem solving
(overall) 5.29 2.26 1.00 10.00 642
CI-2
Volunteering for a non-electoral
organization 7.13 2.35 1.00 10.00 642
CI-3
Membership in a group or
association (overall) 2.89 1.19 1.00 8.88 642
CI-4
Participation in charitable
fundraising 4.90 2.76 1.00 10.00 642
CI-Composite
Composite variable for all civic
indicators 20.20 6.51 4.16 37.75 642
Political Voice Indicators
PV-1 Contacting officials 2.86 2.20 1.00 10.00 642
PV-2
Contacting the media regarding
political issues 2.93 2.37 1.00 10.00 642
PV-3
Discussing political and public
issues on social media 3.97 2.57 1.00 10.00 642
PV-4 Protesting 1.42 1.15 1.00 10.00 642
PV-5 Petitioning 3.71 2.64 1.00 10.00 642
PV-6 Boycotting 3.57 2.65 1.00 10.00 642
PV-Composite
Composite variable for all political
voice indicators 18.46 9.88 6.00 52.20 642
Formal Political Indicators
FP-1 Voting 3.59 2.10 1.00 10.00 642
FP-2
Membership in a political party or
organization 2.18 2.17 1.00 10.00 642
FP-3 Running for office 2.50 1.41 1.00 10.00 642
FP-4
Taking an active role in the ruling
political party or the government
(China) 1.98 2.24 1.00 10.00 642
FP-5
Attending public hearings and
open meetings (China) 1.92 2.12 1.00 10.00 642
138
FP-Composite
Composite variable for all formal
political indicators 12.17 5.98 5.00 39.07 642
Total-
Composite
Composite variable for all civic
engagement indicators 11.30 4.11 3.00 28.87 642
Sample from the U.S.
Variable Code Variable Mean Std. D Min Max Obs.
Civic Indicators
CI-1
Community problem solving
(overall) 5.02 1.98 1.00 10.00 652
CI-2
Volunteering for a non-electoral
organization 3.81 2.40 1.00 10.00 652
CI-3
Membership in a group or
association (overall) 3.47 1.28 1.00 7.27 652
CI-4
Participation in charitable
fundraising 5.81 2.70 1.00 10.00 652
CI-Composite
Composite variable for all civic
indicators 18.10 5.42 4.80 31.08 652
Political Voice Indicators
PV-1 Contacting officials 3.85 2.62 1.00 10.00 652
PV-2
Contacting the media regarding
political issues 3.00 2.21 1.00 10.00 652
PV-3
Discussing political and public
issues on social media 3.72 2.47 1.00 10.00 652
PV-4 Protesting 4.02 2.57 1.00 10.00 652
PV-5 Petitioning 5.44 2.62 1.00 10.00 652
PV-6 Boycotting 3.45 2.37 1.00 10.00 652
PV-Composite
Composite variable for all political
voice indicators 23.47 9.53 6.00 54.38 652
Formal Political Indicators
FP-1 Voting 5.02 2.81 1.00 10.00 652
FP-2
Membership in a political party or
organization 3.13 2.49 1.00 10.00 652
FP-3 Running for office 2.05 1.55 1.00 10.00 652
139
FP-4
Volunteering for candidate or
political organization (US) 3.04 2.28 1.00 10.00 652
FP-5 Campaign contributions (US) 2.93 2.33 1.00 10.00 652
FP-Composite
Composite variable for all formal
political indicators 16.16 7.13 5.00 39.39 652
Total-
Composite
Composite variable for all civic
engagement indicators 12.91 4.16 3.41 26.90 652
140
Appendix D: List of Interviews
Interview University Gender Note
1 Peking University Female
2 Peking University Male Roommate of Interviewee 3
3 Peking University Male Roommate of Interviewee 2
4 Peking University Male
5 Peking University Female Roommate of Interviewee 6 and 7
6 Peking University Female Roommate of Interviewee 5 and 7
7 Peking University Female Roommate of Interviewee 5 and 6
8 Peking University Male
9 Shandong University Male Roommate of Interviewee 10
10 Shandong University Male Roommate of Interviewee 9
11 Shandong University Female
12 Shandong University Male
13 Qingdao University Male Roommate of Interviewee 14
14 Qingdao University Male Roommate of Interviewee 13
15 Qingdao University Male
16 Qingdao University Female Roommate of Interviewee 17
17 Qingdao University Female Roommate of Interviewee 16
18 Qingdao University Male
19 Qingdao University Female
20 University of Southern California Female
21 University of Southern California Female Roommate of Interviewee 22
22 University of Southern California Female Roommate of Interviewee 21
23 University of Southern California Male
24 University of Southern California Male
25 University of Southern California Male Roommate of Interviewee 26
26 University of Southern California Male Roommate of Interviewee 25
27 University of Southern California Female
28 University of Southern California Male
29 University of Southern California Female
30 Chapman University Female
31 Chapman University Male Roommate of Interviewee 32
32 Chapman University Male Roommate of Interviewee 31
141
Appendix E: Survey Questionnaire for Participants in China
5
您好！我是⼀名来⾃南加州⼤学的博⼠候选⼈，我正在进⾏⼀项有关⼤学⽣课余活
动的调查研究，这项研究的⽬的是了解⼤学⽣参与公共事务的情况。
本研究包含两个问卷，第⼀个问卷于 2018 年秋季学期末发放，第⼆个问卷于 2019
年春季学期中。这两个问卷分别仅会花费您 15 分钟左右的时间。如果您同意参与并完成
第⼀个问卷，您将获得 35 元的报酬。如果您同意参与并完成第⼆个问卷，您将获得 105
元的报酬（两次问卷共计 140 元）。报酬将会在您完成⼀份问卷后 30 天内发放 (以⽀付宝
的形式⽀付)。同时，您还有机会赢得 700 元的抽奖。如果您和您的至少⼀位室友都完成
了本次问卷，您赢得抽奖的概率将增加⼗倍。如果您和您的至少⼀位室友都完成了两次问
卷，您赢得抽奖的概率将增加三⼗倍。问卷的参与者需满⾜以下条件：
• 您必须是北京⼤学、山东⼤学青岛校区、或青岛⼤学的全⽇制⼤⼀在校⽣
• 居住在学校宿舍
• 同住的室友是学校随机分配的
• 年满⼗八岁的中国公民
本问卷答案没有对错之分，请按您的真实情况和想法填写。参与本研究是完全⾃愿
的，您随时可以退出本研究。这项调查是完全匿名的，我不会问及您的隐私或者个⼈信息
，所采集的数据不包含您的姓名、学校、地址等。您的邮箱、⽀付宝信息仅会被⽤于发放
报酬和第⼆次问卷，在本研究结束后，您的邮箱和⽀付宝信息会被彻底从数据中清空删除
。在删除上述信息后，所有数据将会被加密存储，只做研究之⽤。如果您不希望您的匿名
回答被⽤于将来的研究，请不要参与本研究。如果您选择完成本问卷，将被视作您同意参
与本研究。您的参与⼗分重要，感谢您为开拓新知做出的贡献。
出于对您的保护，南加州⼤学⼈类研究保护计划可以获取本研究产⽣的匿名信息。
HRPP 是⼀个学术研究监督机构，它保障研究参与者的权利和福利。
研究者联系⽅式：yli257@usc.edu
研究伦理委员会联系⽅式：
University Park Institutional Review Board (UPIRB), 3720 South Flower Street #301, Los
Angeles, CA 90089-0702, (213) 821-5272 or upirb@usc.edu
5
The questionnaire in this appendix is the printout version for students who prefer hard copies. The layout of this questionnaire
is adjusted in the online version.
142
QR Code
6
问卷编号: ______
7
如果您选择通过⼿写的⽅式完成本问卷，请将完成的问卷装到提供回信信封⾥，并投入任
何中国邮政邮筒。如果您选择通过⼆维码完成本问卷的电⼦版，则在您完成问卷后，无需
保留纸质版。
第⼀部分
1. 您是北京⼤学⼤⼀全⽇制本科⽣吗︖
a. 是
b. 否
2. 您是山东⼤学青岛校区⼤⼀全⽇制本科⽣吗︖
a. 是
b. 否
3. 您是青岛⼤学⼤⼀全⽇制本科⽣吗︖
a. 是
b. 否
4. 您是中国公民吗︖
a. 是
b. 否
5. 您是如何选择⽬前的室友的︖
a. 我选择让学校随机分配室友
b. 我们学校不提供选择，只能随机分配室友
c. 我⾃主选择了我的室友
d. 其他，请说明_______________
6
A personalized QR code that is associated with the questionnaire number is provided here for
participants who wish to complete the survey on their electronic devices.
7
Each questionnaire has a pre-filled out questionnaire number, which is used to identify
respondents without collecting personal information and to match participants and their
roommates.
143
如果 1-3题中，您的回答都是“否”，或者您第 4题选择了“否”，又或者您第 5题选择了“c”
或“d”选项，则您不符合本研究的参与条件，请终⽌作答。
第⼆部分
6. 请提供⼀个下学期收取第⼆次问卷调查的电⼦邮箱：__________________
7. 请提供您的⽀付宝账号：__________________
*请注意确保您的邮箱和⽀付宝账号填写清晰、正确。因为本研究不采集个⼈信息，您
提供的电⼦邮箱和⽀付宝账号将是您获得报酬和第⼆次问卷的唯⼀渠道。
8. 如果您的室友也参与了本研究，请写下他们的问卷编号。问卷编号在本问卷第⼀页
的左上角。如果您填写室友的问卷编号，您最终赢得 700 元抽奖的概率会增加⼗
倍。如果您和您的至少⼀位室友都完成了两次问卷，您赢得抽奖的概率将增加三⼗
倍。
a. 室友 1：________
b. 室友 2：________
c. 室友 3：________
d. 室友 4：________
e. 室友 5：________
f. 室友 6：________
g. 室友 7：________
h. 如果您需要添加更多室友的问卷编号，请在本页任何空⽩处填写。
9. 您在上⼀题中填写的室友中，有没有⼈在 2018 年秋季学期不是您的室友︖如有，
请列明（如：“室友 1”）
8
：
________________________________________________________________________
_______________________________________________________________________
________________________________________________________________________
10. 您现在还就读于您 2018 年秋季学期就读的⼤学吗︖
a. 是
b. 否，我转学到其他⼤学了
8
Questions 9 and 10 are for the second wave survey in Spring 2019 only.
144
c. 否，我退学了
11. 您在哪年出⽣︖________
12. 您的性别是︖
a. 女
b. 男
c. 其他，请说明________
13. 您的民族是︖
a. 汉族
b. 少数民族
14. 您在您⽬前的宿舍住了有多久︖
a. ⼩于 1 个⽉
b. 1-3 个⽉
c. 3-6 个⽉
d. 多于6 个⽉
15. 您有⼏位室友︖ ____
16. 您和您⽬前的室友住了有多久︖
a. ⼩于 1 个⽉
b. 1-3 个⽉
c. 3-6 个⽉
d. 多于6 个⽉
第三部分
下列问题是关于您在⼀系列活动中的参与情况，请您从 1-7 中选取⼀个整数作答：
1 代表您从来没有参与过这项活动，且未来也不会参加︔
2代表您从来没有参与过这项活动，但未来可能考虑参加︔
3代表您参与过这项活动，但⽬前为⽌只参与过⼀次︔
4代表您参与过这项活动，不过不经常参加（但参与次数多于⼀次）︔
5代表您参与过这项活动，有时参加︔
6代表您参与过这项活动，比较常参加︔
145
7代表您参与过这项活动，非常积极参加。
17. 您参加过下列活动吗︖如果参加过，您参加的频率如何︖
18. 您参加过下列活动吗︖如果参加过，您参加的频率如何︖
19. 您参加过下列活动吗︖如果参加过，您参加的频率如何︖
1 2 3 4 5 6 7
（从不参
与，未来
也不会参
与）
（从不参
与，但未
来可能考
虑参与）
（参与过
一次）
（不经常参
加）
（有时参加） （比较常
参加）
（非常积极
参加）
学生社群的问题解决(比如学
生团体、校园节能、校园安全
等问题)
本地社群的问题解决（比如邻
里公共设施、小区环境等）
志愿者活动或志愿者团体
慈善筹款（比如慈善骑行等）
1 2 3 4 5 6 7
（从不参
与，未来
也不会参
与）
（从不参
与，但未
来可能考
虑参与）
（参与过
一次）
（不经常参
加）
（有时参加） （比较常
参加）
（非常积极
参加）
联系政府官员或者人大代表咨
询问题、了解情况、反映问题
等
联系传统媒体表达关于政治政
策和公共事务的看法
在社交媒体上表达对政治政策
和公共事务的看法
游行、抗议活动
联署请愿书、申诉书等
抵制某公司或某国产品或服务
1 2 3 4 5 6 7
（从不参
与，未来
也不会参
与）
（从不参
与，但未
来可能考
虑参与）
（参与过
一次）
（不经常参
加）
（有时参加） （比较常
参加）
（非常积极
参加）
在基层人大选举中投票
作为候选人参与基层人大选举
在村委会选举中投票
146
下列问题是关于您在组织和团体中的参与情况，请您从 1-5 中选取⼀个整数作答：
1 代表您不是这个团体的成员，且以后也不会加入︔
2代表您不是这个团体的成员，但未来可能考虑加入︔
3代表您是这个团体的成员，但并不积极参与活动︔
4代表您是这个团体的成员，有时参与活动︔
5代表您是这个团体的成员，积极参与活动。
20. 您是下列组织或团体的成员吗︖如果是，您的参与情况如何︖
作为候选人参与村委会选举
在校级学生代表大会（学代会）
选举中投票
作为候选人参与校级学生代表
大会（学代会）选举
在党和政府中担任职务
参与政府听证会、协商会等
1 2 3 4 5
（不是成员，以后
也不加入）
（不是成员，但未
来可能加入）
（是成员，但不积
极参与活动）
（是成员，有时参
与活动）
（是成员，积极参
与活动）
政治党派（包括中国共产党和
其他民主党派）
工会
行业协会、专业协会
宗教团体
商业协会、商业团体
体育团体
志愿者团体
学生团体
147
Appendix F: Survey Questionnaire for Participants in the U.S.
9
Hi! I am a researcher at the University of Southern California. I would like to invite you
to take part in my study regarding political and civic participation. This research studies include
only people who voluntarily choose to take part. This document explains information about this
study. Please feel free to ask questions about anything that is unclear to you.
This research explores civic engagement. Your opinion is very important to this research.
This study consists of two surveys: one at the beginning of this semester and the other at the end
of the next semester. Each survey would take only approximately 15 minutes of your precious
time. If you agree to take part in this study, you will receive a $5 gift card upon the completion
of the first survey, an additional $15 gift card upon the completion of both surveys ($20 total).
You will also have a chance to win a $100 gift card in a raffle at the end of the next semester.
Your roommates also received this survey. If you and at least one of your roommates complete
this survey, your chance of winning the raffle will increase 10 times. If you and at least one of
your roommates complete both surveys, your chance of winning the raffle will increase 30 times.
To qualify for this study, you need to meet the following criteria:
• A first-year full-time undergraduate student at University of Southern California
or Chapman University
• U.S. citizen who is least 18 years old
• Reside in a university dormitory
• Have randomly assigned roommate(s)
There’s no right or wrong answer. Please answer this survey according to your true
opinion. You can withdraw from this study at anytime. Participation in this study is completely
voluntary and your responses are strictly kept anonymous. There will be no identifiable
information obtained in connection with this study. Your name, address or other identifiable
information will not be collected. Your responses will be stored on a password protected
computer with security software and only authorized research personnel can access the data. A
copy of the data collected without any identifiable information will be retained at the discretion
of the researcher. The anonymous data may be used in future research studies. If you do not want
your data used in future studies, you should not participate in this study. By completing the
survey, you are agreeing to participate in the research. Thank you for being a part of the
production of human knowledge. Your participation is greatly appreciated!
The members of the research team and the University of Southern California’s Human
Subjects Protection Program (HSPP) may access the data. The HSPP reviews and monitors
research studies to protect the rights and welfare of research subjects.
INVESTIGATOR CONTACT INFORMATION
The Principal Investigator can be contacted via email at yli257@usc.edu
IRB CONTACT INFORMATION
University Park Institutional Review Board (UPIRB), 3720 South Flower Street #301, Los
Angeles, CA 90089-0702, (213) 821-5272 or upirb@usc.edu
9
The questionnaire in this appendix is the printout version for students who prefer hard copies. The layout of this questionnaire
is adjusted in the online version.
148
QR Code
10
Questionnaire Number: ______
11
If you choose to complete this survey by hand, please seal the completed survey in the provided
return envelope and drop it off at any USPS office or post box. If you choose to complete the
electronic version of this survey via the QR code above, you can discard this hard copy after
you complete the survey.
Part I
1. Are you a first-year full-time undergraduate student at University of Southern California?
a. Yes
b. No
2. Are you a first-year full-time undergraduate student at Chapman University?
a. Yes
b. No
3. Are you a U.S. citizen?
a. Yes
b. No
4. How did you choose your roommate(s)?
a. I chose to have randomly assigned roommates
b. I did not have a choice but to have randomly assigned roommates
c. I chose to select my roommates myself
d. Other, please specify _______________
If your answers to questions 1 and 2 are BOTH “No”, OR your answer to question 3 is “No”,
OR your answer to question 3 is “c” or “d”, you are not eligible for this study. Please
disregard this survey.
Part II
5. Please provide an email address where you prefer to receive your e-gift card and the
second part of this survey next semester: __________________
10
A personalized QR code that is associated with the questionnaire number is provided here for
participants who wish to complete the survey on their electronic devices.
11
Each questionnaire has a pre-filled out questionnaire number, which is used to identify
respondents without collecting personal information and to match participants and their
roommates.
149
*Please make sure the email address is correct and legible as it is the only way for you to
receive compensation since this survey does not collect any identifiable personal
information about you.
6. If your roommate(s) also participate in this survey, please provide their questionnaire
number(s) below. You can find the questionnaire number on the upper left corner on the
first page of this survey. Please note your chance of winning the $100 raffle will increase
10 times if you provide this information. If you and at least one of your roommates
complete both surveys, your chance of winning the raffle will increase 30 times.
a. Roommate 1: ________
b. Roommate 2: ________
c. Roommate 3: ________
d. Roommate 4: ________
e. Roommate 5: ________
f. Roommate 6: ________
g. Please add more roommates to this question if needed. You can write their
questionnaire on any empty space of this page.
7. Were any of the roommates listed in the question above not your roommate in Fall 2018
when you participated in the first survey of this study? If so, please provide their
information (e.g. “Roommate 1”)
12
:
________________________________________________________________________
________________________________________________________________________
________________________________________________________________________
8. Are you still attending the same university as you did in Fall 2018?
a. Yes
b. No, I transferred to another university/college
c. No, I dropped out
9. When were you born? (year)________
10. What is your gender?
a. Female
b. Male
c. Other, please specify ______
11. What is your race/ethnicity?
a. African American/Black
b. Asian/Asian American/Pacific Islander
c. Caucasian American/White
d. Hispanic/Latino
e. Other, please specify______
12. How long have you been living at your current dorm room?
12
Questions 7 and 8 for the second wave survey in Spring 2019 only.
150
a. Less than 1 month
b. 1-3 months
c. 3-6 months
d. more than 6 months
13. How many roommates do you have? ______
14. How long have you been living with your current roommate(s)?
a. Less than 1 month
b. 1-3 months
c. 3-6 months
d. more than 6 months
Part III
For the following questions, please choose a number from 1 to 7 to indicate your participation
in the activities:
1 represents you have never participated in a given activity AND would never consider
participating in the future;
2 represents you have never participated in a given activity BUT would consider participating in
the future;
3 represents you have participated in a given activity only once in your lifetime;
4 represents you have seldom participated in a given activity (but have participated more than
once);
5 represents you have sometimes participated in a given activity;
6 represents you have often participated in a given activity;
7 represents you have always participated in a given activity.
15. Have you ever participated in the following activities? If so, how often do you
participate?
1 2 3 4 5 6 7
(Never,
would
never)
(Never,
maybe in
the future)
(Once) (Seldom) (Sometimes) (Often) (Always)
Problem solving in your
student community (e.g.
student government)
Problem solving in your local
community (e.g. neighborhood
watch, HOA)
Volunteering for a non-
electoral organization
Fundraising activities (e.g.
fundraising run/walk/ride/sale)
151
16. Have you ever participated in the following activities? If so, how often do you
participate?
17. Have you ever participated in the following activities? If so, how often do you
participate?
The following questions are about your membership status in the listed political or civic
organization. please choose a number from 1 to 5 to indicate your participation in the
activities:
1 represents you are not a member of a given organization AND would never consider becoming
a member in the future;
2 represents you are not a member of a given organization BUT would consider becoming a
member in the future;
3 represents you are a member of a given organization, but are not active in said organization;
4 represents you are a member of a given organization and are somewhat active in said
organization;
5 represents you are a member of a given organization and are very active in said organization.
1 2 3 4 5 6 7
(Never,
would
never)
(Never,
maybe in
the future)
(Once) (Seldom) (Sometimes) (Often) (Always)
Contacting political officials
(e.g. requesting information,
expressing your opinion, etc.)
Contacting media (newspaper,
TV, etc.) regarding
political/public issues
Expressing your opinions on
political or public issues on
social media
Protests and demonstrations
Signing a petition
Boycotting
1 2 3 4 5 6 7
(Never,
would
never)
(Never,
maybe in
the future)
(Once) (Seldom) (Sometimes) (Often) (Always)
Voting
Running for office
Volunteering for a political
candidate or organization
Donating to a political
candidate or organization
152
18. Are you a member of the following political or civic organizations? If so, how active are
you?
1 2 3 4 5
Not a member,
would never be a
member)
(Not a member,
maybe in the
future)
(Non-active
member)
(Somewhat active
member)
(Very active
member)
Political parties
Trade unions
Professional and civil
associations
Religious groups
Business associations
Sports groups
Volunteer organizations
Student organizations
153
Appendix G: In-depth Semi-structured Interview Question (Guideline)
1. Introduction and Consent
a. Researcher self-introduction
b. Introduction of the study, including the nature of the study, participant
compensation, privacy protocol and data security, and the rights of participants.
c. Do you agree to voluntarily participate in this study?
2. Participant’s pre-treatment civic engagement:
a. In Fall 2018, what forms of civic engagement did you participate in?
b. Why do you participate in these forms of civic engagement (go over each form of
civic engagement)?
c. What do you think of these civic engagement activities? What do they mean to
you?
d. What do you think are the advantages and disadvantages of participating?
e. What are some most memorable experiences in civic engagement?
f. Why do you not participate in other forms of civic engagement? Have you heard
of them before?
3. Perception of roommates’ civic engagement:
a. Do you think your roommates have active civic engagement? What forms of civic
engagement they participate in?
b. Have any of your roommates’ civic engagement behaviors changed in the past
year? If so, how?
c. What do you think about your roommates’ participation in these activities?
4. Participant’s post-treatment civic engagement:
a. What forms of civic engagement do you participate in now?
b. Has your civic engagement behavior changed in the past two semesters? How?
c. What are the reasons for the changes in your civic engagement?
d. Do you think you and your roommates influence one anther in terms of civic
engagement? If so, how?
5. Do you and your roommates talk often? What do you normally talk about?
6. Do you and your roommates do things together often? What do you normally do?
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Exploring student motivations toward civic engagement: an application of expectancy-value theory
PDF
Providing a new low-cost primary care facility for under-served communities: a site suitability analysis for Service Planning Area 6 in Los Angeles County, California
PDF
The affective framing of tort reform: toward a theory of the mediating effects of emotion on attitude formation
PDF
Civic engagement in American schools: an evaluation study
PDF
Pathways to immigrant civic engagement: the storytelling network and church participation
PDF
In pursuit of liberty: the civic liberalism of Frederick Douglass
PDF
Essays on treatment effect and policy learning
PDF
Civic engagement of Teach For America alumni in Los Angeles
PDF
Identity salience and political violence: centering social identity in the study of intrastate conflict
PDF
Participatory public culture and youth citizenship in the digital age: the Medellín model
PDF
Powellism, Thatcherism and the Conservative Party, 1945-87: the party as the site of ideological transition
PDF
Considering the effects of disfluent speech on children’s sentence processing capabilities and language development
PDF
The social making of authoritarian environmentalism: protest-litigation nexus and policy changes in China
PDF
Factors contributing to the civic engagement of rural community college students
PDF
The #Revolution will be liked: an examination of civic expression among Hawaiian youth
PDF
Social policy and presidential ideology in Latin America: the political economy of social spending and anti-poverty programs
PDF
The role of local actors in international development: how incentive types affect foreign aid effectiveness
PDF
Sustaining open source software production: an empirical analysis through the lens of microeconomics
PDF
Forced migration as a cause and consequence of conflict
PDF
Mexican hometown associations: Zacatecanos in Los Angeles, gendered participation and Mexican state ties
Asset Metadata
Creator
Li Sarain, Ada Yue
(author)
Core Title
Engaging together: exploring the peer effects of civic engagement
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Political Science and International Relations
Degree Conferral Date
2022-08
Publication Date
07/22/2022
Defense Date
06/23/2022
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
China,civic engagement,North America,OAI-PMH Harvest,peer effects,quasi-experiment,random roommate assignment
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Sellers, Jefferey (
committee chair
), Crigler, Ann (
committee member
), Yang, Aimei (
committee member
)
Creator Email
adalisarain@gmail.com,lisarain@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC111373983
Unique identifier
UC111373983
Legacy Identifier
etd-LiSarainAd-10912
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Li Sarain, Ada Yue
Type
texts
Source
20220722-usctheses-batch-961
(batch),
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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.
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
Repository Email
cisadmin@lib.usc.edu
Tags
civic engagement
peer effects
quasi-experiment
random roommate assignment