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
/
Media reinvented: the transformation of news in a networked society
(USC Thesis Other)
Media reinvented: the transformation of news in a networked society
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
i
MEDIA REINVENTED:
THE TRANSFORMATION OF NEWS IN A NETWORKED SOCIETY
by
Matthew S. Weber
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
(COMMUNICATION)
August 2010
Copyright 2010 Matthew S. Weber
ii
DEDICATION
To my parents, Linda and Ken. Thank you for always supporting me.
And to Emily, with all my love. When we met, this dissertation was a glimmer of an
idea – I would never have finished without you.
iii
ACKNOWLEDGEMENTS
I started thinking about the ideas contained in this dissertation when I was an
undergraduate at Northwestern University, and I witnessed first-hand much of the
upheaval in the news media community during my time working in industry. This
dissertation brings together those experiences, and research that I have been working
on since I began my Ph.D in August 2006 at the University of Southern California’s
Annenberg School for Communication and Journalism. This dissertation would not
have happened without the help, support and encouragement of many people along
the way.
I would like to acknowledge the support of USC’s Annenberg School for
Communication and Journalism, without which this project would never have been
possible. First, I would like to thank my advisor, Peter Monge, for guiding me
through the world of academia, and always pushing me to develop my ideas and
push my thinking. Thank you for your advice and encouragement, and for treating
me as a colleague. In addition, thank you to my dissertation committee: Manuel
Castells and Mark Kennedy. Your feedback and support was invaluable in this
endeavor.
Outside of my dissertation committee, many other faculty members at USC
were hugely important in supporting this work. Janet Fulk was a willing collaborator
as I explored the world of social networking. Larry Gross, a tremendous advocate for
the graduate student community, supported my work throughout and allowed me the
iv
flexibility I needed to finish this dissertation. Andrea Hollingshead always looked
out for my best interests and provided the critical guidance I needed to make my
work what it is today. Geneva Overholser helped me to expand the scope of my
research, and I have learned quite a bit from her leadership in the journalism
community. Sarah Banet-Weiser offered helpful advice throughout, and was always
willing to listen and offer advice as I navigated the world of research. And lastly, a
special thank you to Ganesh Gaikwad, who helped with much of the programming
behind the Web crawler used in this research, for his help on this project.
Outside of USC, I benefited from the generosity of many. Vladamir Barash
and Douglas Forster were invaluable during the early stages of this research. Their
work at Cornell University helped to inspire what would eventually become the
HistoryCrawl research tool. At the Oxford Internet Institute, Eric Meyer, Ralph
Schroeder, Bill Dutton and Vikki Nash offered support and resources throughout this
project, hosting me during numerous trips to the United Kingdom and always
welcoming me. Thank you as well to David Levy and the staff and fellows of the
Reuters Institute for the Study of Journalism for providing me with an opportunity to
spend time learning from you, and for the opportunity to expand the scope of my
research internationally. I am also eternally gratefully to the Web Science Research
Initiative and Wendy Hall for their ongoing support and funding. The dissertation is
a testament to the potential of Web Science, and the type of research that can be
accomplished when the foundations of computer science intersect with the theory
v
and research of the social sciences. Thank you as well to the University of
Michigan’s Management and Organizations Department, and specifically Gretchen
Spreitzer, for hosting me in the final year of this study, welcoming me into their
community and encouraging my research.
Additionally, I am indebted to the team at the Internet Archive for their help
and advice along the way – particularly Kris Carpenter. In 1996 they undertook the
task of archiving the Internet, and I’m confident that in the year to come we will
begin to realize the full value of this enormous historical record. In the past year, I
have been fortunate to meet many amazing journalists, bloggers and social media
experts. While many have chosen to keep their contributions anonymous, I would
like to thank Rachel Sklar at Mediaite.com, as well as those who helped me at
MSNBC, The New York Times, The Chicago Tribune, The Washington Post, The
Guardian, The Telegraph, The BBC, The San Diego Union Tribune, and many
others.
The process of writing a dissertation is an experience unlike any other. More
than anything, this process has reminded me how lucky I am to have an amazing
family and wonderful friends to rely on for support and encouragement. Chris
Meador pointed me in the right direction and introduced me to many wonderful
people in the media world. Maegan Carberry is a force of nature and helped to
educate me on the ways of the blogger. Barbara Jaffrey is one of the kindest people I
know, and she opened up her home to me on my many trips to Los Angeles. Rachel
vi
Schachter and Steve Caskey have been great friends since college, and are my
academic friends-in-arms. My sister, Stacey, and my brother-in-law, Erik, were
always willing to give me a couch to sleep on, a cold beer to drink, or just some
moral support. My brother, Michael, is my best friend and made sure I kept my sense
of humor throughout. My parents, Ken and Linda, have supported every decision I
have made, even if they didn’t always understand my choices. I am indebted to all of
you, and the many other friends and colleagues who have helped me out along the
way. The experience of researching and writing a dissertation has reminded me of
just how lucky I am to be surrounded by so many amazing people – thank you. Most
importantly, I am forever thankful to my best friend, my love and my soulmate,
Emily. When I moved to Los Angeles to start my PhD, she moved with me knowing
full well the road that lay ahead. And without her patience, support, and
encouragement, this would not have been possible.
Finally, I would never have started this without the encouragement of Dick
Schwarzlose, my mentor when I was a graduate student at the Medill School of
Journalism, Northwestern University. Dick encouraged me to pursue the answers to
the questions that I asked. He passed away in 2003, but his encouragement has
continued to motivate me.
vii
TABLE OF CONTENTS
Dedication ii
Acknowledgements iii
List of Tables x
List of Figures xii
Abbreviations xiv
Abstract xv
CHAPTER 1: INTRODUCTION 1
Purpose of the Study 1
Chapter Summaries 8
CHAPTER 2: THE EMERGENCE OF NEW ORGANIZATIONAL FORMS OF
NEWS: AN ECOLOGICAL PERSPECTIVE 10
Community Ecology 12
Variation, Selection and Retention 17
Emergence of New Organizational Forms 22
Speciation and the Emergence of New Forms 30
A Stable Community: A Brief History of the Modern News Media 40
The Emergence of Online and Social Media as Organizational Forms 55
The Disruption of the News Media Community as a Speciated Event 61
Speciation and Available Resources 67
CHAPTER 3: LEGITIMIZATION OF NEW FORMS – THE RISE OF ONLINE
NEWS 72
New Media and New Forms 72
Legitimation as an Ecological Process 74
Media Coverage 80
Professional Associations 83
Legal Rulings 85
Speciation and Legitimization 87
Online News: Moving Towards Critical Mass 91
viii
CHAPTER 4: TRANSFORMATION IN RESPONSE – THE EVOLUTION OF
TRADITIONAL NEWS 96
Transformation in Response to the Emergence of New Forms 96
Drivers of Transformation 106
Hyperlinking as a Strategic Tool 114
Communities and Populations of Embedded Hyperlinks 117
The Transformation of News: An Ecological Perspective 121
CHAPTER 5: DATA AND METHODS 130
New Approaches to Transformation Research 130
Digital Archive Research Using the Internet Archive 132
History Crawl: A Tool for Crawling the Internet Archive 135
Interview Data 144
Additional Data Sources 147
Variables 147
Newspaper Organization Variables 148
Event Variables 149
Data Coding and Transformation 155
Interview Data 155
Network Data 156
Newspaper Strategies 157
Environmental Level Events 160
Analysis Overview 161
Network Analysis 161
The Effects of Organizational Strategy on Failure 181
The Effects of Linking Over Time 184
CHAPTER 6: RESULTS 187
Descriptive Results 187
Longitudinal Network Analysis 206
Hypotheses 1 to 3 213
Organizational Transformation and Poisson Regression Results 225
Hypotheses 4 and 5 226
Hypotheses 6, 7 and 8 229
The Role of Strategy in Organizational Communities 232
Hypotheses 9 and 10 232
CHAPTER 7: CONCLUSION 240
Discussion 240
Emergence 240
Legitimation 245
ix
Transformation 249
Implications for the News Media Community 256
Limitations and Future Research 261
Directions for Future Research 263
Archival Internet Research 263
Strategy and Online Development 264
The Nature and Death of Organizational Forms 265
Conclusion 265
BIBLIOGRAPHY 267
APPENDICES:
APPENDIX A: SEED LIST OF WEB SITES FOR HISTORY CRAWL 289
APPENDIX B: LIST OF INTERVIEW SUBJECTS 292
APPENDIX C: INTERVIEW PROTOCOL 295
APPENDIX D: STATNET GOODNESS OF FIT
DIAGRAMS 1998 – 2007 298
APPENDIX E: COVARIANCE MATRIX FOR BIRTH AND DEATH 309
x
LIST OF TABLES
Table 1: The Online News Form Compared to Print Newspapers 105
Table 2: Summary of Hypotheses 127
Table 3: Accuracy of HistoryCrawl versus hand crawl 142
Table 4: Page Depth and Link Validity Analysis 143
Table 5: Professional Organizations in the Online News Media Community 151
Table 6: Key Court Rulings Affecting Online Media – 1998 – 2009 153
Table 7: Organizational Linking Strategies and Density Ranges 158
Table 8: Environmental events affecting the online news media community 160
Table 9: Visual representation of estimated ERGM parameters 174
Table 10: Poisson Regression Models and Hypothesis Summary for
Organizational Foundings 180
Table 11: Poisson Regression Models and Hypothesis Summary for
Organizational Failures 183
Table 12: Visual representation of estimated ERGM parameters for
RSiena 186
Table 13: Organizational Breakdown by Type with Descriptive
Attributes (based on final network – 2007) 193
Table 14: Breakdown of Newspaper Population
(for all newspapers) 195
Table 15: Network Densities and Main Component Density 1998 – 2007 198
Table 16: Statnet Baseline Models 1998 – 2007: AIC Indices 208
Table 17: Statnet Baseline Models 1998 – 2002: Fitted Parameters 210
Table 18: Statnet Baseline Models 2003 – 2007: Fitted Parameters 212
xi
Table 19: Convergence Diagnostics for RSiena Model 1998 – 2007 215
Table 20: Estimation for RSiena Model 1998 – 2007 216
Table 21: Estimation for RSiena Model 1998 – 2007
(Attributes for H
1
, H
2
, H
3
) 220
Table 22: Poisson Regression Results for News Community Birth Rate 221
Table 23: Poisson Regression Results for News Community Birth
Rate – Model Fit 222
Table 24: Poisson Regression Results for News Community Failure Rate 223
Table 25: Poisson Regression Results for News Community Failure
Rate – Model Fit 224
Table 26: Estimation for RSiena Model 1998 – 2007
(Attributes for H
9a
, H
9b
, H
10a
, H
10b
) 237
Table 27: Summary of Results 238
Table A-1: List of initial Web sites used for HistoryCrawl 289
Table B-1: List of Interview Subject 292
Table E-1: Correlation Matrix for Birth Rates 309
Table E-2: Correlation Matrix for Failure Rates 310
xii
LIST OF FIGURES
Figure 1: Stages of Community Evolution by Network Link Density 24
Figure 2: Number of Daily Newspapers From 1783 – 2005 46
Figure 3: Newspaper Circulation with Breakouts for Morning and
Evening Newspapers 47
Figure 4: Parapatric Speciation in an Established Community 66
Figure 5: Press mentions of blogging and social networking
(1998 – 2009) 82
Figure 6: Number of employees in traditional and online news 94
Figure 7: Community Transformation as a Multilevel Process 123
Figure 8: HistoryCrawl operating schema 140
Figure 9: Histogram of Hyperlinking Strategies Based on Density Ranges 159
Figure 10: Number of Organizations by Year 189
Figure 11: Organizational Births by Year Based on new Web sites 190
Figure 12: Changes in Newspaper Ownership, Publisher and Editor 191
Figure 12: Network and Main Component Densities by Year 201
Figure 13: Network Diagram – 1999 202
Figure 14: Network Diagram – 2001 203
Figure 15: Network Diagram – 2003 204
Figure 16: Network Diagram – 2005 205
Figure 17: The Wall Street Journal’s Online Format 259
Figure D-1: 1998 Goodness-of-Fit Out-Degree 298
xiii
Figure D-2: 1998 Goodness-of-Fit In-Degree 299
Figure D-3: 1999 Goodness-of-Fit Out-Degree 299
Figure D-4: 1999 Goodness-of-Fit In-Degree 300
Figure D-5: 2000 Goodness-of-Fit Out-Degree 300
Figure D-6: 2000 Goodness-of-Fit In-Degree 301
Figure D-7: 2001 Goodness-of-Fit Out-Degree 301
Figure D-8: 2001 Goodness-of-Fit In-Degree 302
Figure D-9: 2002 Goodness-of-Fit Out-Degree 302
Figure D-10: 2002 Goodness-of-Fit In-Degree 303
Figure D-11: 2003 Goodness-of-Fit Out-Degree 303
Figure D-12: 2003 Goodness-of-Fit In-Degree 304
Figure D-13: 2004 Goodness-of-Fit Out-Degree 304
Figure D-14: 2004 Goodness-of-Fit In-Degree 305
Figure D-15: 2005 Goodness-of-Fit Out-Degree 305
Figure D-16: 2005 Goodness-of-Fit In-Degree 306
Figure D-17: 2006 Goodness-of-Fit Out-Degree 306
Figure D-18: 2006 Goodness-of-Fit In-Degree 307
Figure D-19: 2007 Goodness-of-Fit Out-Degree 307
Figure D-20: 2007 Goodness-of-Fit In-Degree 308
xiv
ABBREVIATIONS
ABC American Broadcasting Company
AP Associated Press
Blog Web blog
CBS Columbia Broadcasting System
CNN Cable News Network
ERG Exponential Random Graph
FNN Financial News Network
HTML Hyper Text Markup Language
MUD Multi-User Domain
NBC National Broadcasting Corporation
ONA Online News Association
RAB RAID Advisory Board
SNIA Storage Networking Industry Association
SNS Social Networking Site
URL Uniform Resource Locator
V-S-R Variation – Selection - Retention
WARC Web ARChive file format
WWW World Wide Web
xv
ABSTRACT
This study presents an examination of the organizational process of
transformation, specifically examining how new organizations emerge as the result
of new information communication technology, and how existing organizations
emerge in response. In aggregate, the process of transformation is examined in three
stages. First, this study looks at the nature of organizational forms, and seeks to
understand how organizational forms emerge in rapidly changing competitive
environments. Second, this study examines the process of legitimation in an attempt
to better understand how emerging organizational forms are established as
legitimate. Third, this research examines the process of organizational
transformation, and seeks to introduce organizational strategy as a critical
determinant of transformational success for existing organizations.
In order to answer these theoretical issues, this study specifically looks at
the emergence of social media and socially driven news, such as blogs and social
networking sites, and examines how these organizational forms developed over time.
Simultaneously, the transformation of traditional print-based news organizations in
response is studied. As these different organizations have interacted in the online
environment, a link economy has emerged in which hyperlinking has emerged as a
critical component of online information flow. Thus, in addition to the above, the
effect of hyperlinking is considered as a strategic tool that enables organizational
success in the long run.
xvi
The results of this study introduce the process of speciation as a key
mechanism for understanding the rapid emergence of new organizational forms. In
addition, this study finds that media coverage and professional organizations are
early indicators of developing organizational forms. Finally, with regards to strategy,
this research finds that organizations implementing aggressive strategies incur a
higher risk of failure, but also a high potential for decreasing the risk of failure.
Keywords: organizational evolution, strategy, news media, networks, social network
analysis, longitudinal network analysis, network structure, organizational ecology,
community ecology, online news media
1
CHAPTER 1: INTRODUCTION
Purpose of the Study
Responding to a perceived “famine’ in mass media research, the sociologist
Herbert Gans wrote in 1972 (p. 697) that, “we need to know how the mass media
function as institutions and why they function as they do.” More than thirty years
later, the newspaper industry is at a juncture in its evolution, yet scholars and
practitioners struggle to decipher the mechanisms of change at work within the
industry. Journalism is a profession by which social meaning is created, current
events communicated, and public discussion is framed, but in the past decade the
industry has shifted from a structured hierarchy of production to a scattered network
of varying authenticity (Boczkowski, 2004a). Despite a number of landmark studies
since Gans’ call for renewed vigor (Gans, 1979; Klinenberg, 2005; Tuchman, 1978),
the underlying mechanisms driving the transformation of the news media industry
remain largely unexplored. The continued need for research prompted Zelizer
(2007) to call for new methods and theories to address recent shifts in the production
of news: “We need to develop inquiry that will not only reflect the changing
circumstances in which journalism finds itself but anticipate them as well” (p. 111).
In response to the need for a critical reexamination of the state of the
newspaper industry, this dissertation examines the coevolution of the traditional
news media industry in concert with the recent rise of user-generated and network-
based content production. In this way, the mechanisms influencing the rapid
2
transformation of the news industry are examined through the lens of recent
theoretical and analytical advances. As Tuchman (1978) observed in her
ethnographic research of news organizations, the combination of a daily production
system, division of news by topical categories and an organizational focus on
continual production of information creates a constant flow of information to
consumers, continually molding our comprehension of society. Yet in recent years
both the production process and the organizational structure of this system have
started to transform in notable ways.
This transition, however, has not necessarily been as rapid as reports in the
popular press would indicate (see, for example Arends, 2009). Rather, it has been an
ongoing process of gradual change. Even though circulation for evening newspapers
began to decline as early as 1968, total newspaper circulation across evening,
morning and Sunday newspapers continued to grow steadily through 1990. Since
then, however, circulation has marched steadily downwards, heralding a similar
decline in advertising revenue, which was stagnant as early as the mid-1990s.
Readership statistics, which measure the number of people actually reading and
consuming newspapers, has also started to decline although not as rapidly.
According to the Newspaper Association of America (2007) the average weekday
readership for all newspapers in 2006 was 124 million consumers, equivalent to a
little more than half of the population of the United States. While still a significant
percentage of the market, readership has dropped 11 percent since 2000, illustrating
3
further the decline that is emerging on the print side of the industry. As readership
and circulation has decreased, so too has the number of newspapers in production;
the total number of newspapers has dropped from more than 1,600 in 1990 to
roughly 1,500 in 2006 (World Press Trends 2007, 2007). In all, these statistics
highlight an increasing tread of readers migrating away from print newspapers as a
primary source of news information.
Concurrently, as print circulation has declined newspapers have in recent
years turned to online content as an alternative media outlet. The newspaper industry
has experimented with digital forms of content distribution since the 1980s, when
companies tested videotex digital transmission systems for news. In the 1990s
newspapers continued to experiment, testing content distribution via CD-ROM
(Veglis, 2007) and fax (Boczkowski, 2004a), and eventually via the Internet (Greer
& Mensing, 2006). By the late 1990s, and into the new century, the World Wide
Web has firmly been established as the next generation for content distribution, and
companies continue to test a host of operating models for publishing digital content
(Boczkowski & Ferris, 2005). In economic terms, the burgeoning growth of digital
news and media has created new business models; in response, traditional newspaper
organizations have struggled to adapt to these new forms of business. For most of the
first decade of the twenty-first century, online revenue has grown at more than 30
percent, and current industry estimates predict that online advertising will grow at 21
percent a year over the next four years.
4
At the same time that traditional print organizations have struggled to adapt
to online technology, a robust population of users-turned-journalists have emerged
across blogs, social networking sites and portals driven by user-generated content
(Shirky, 2008; Stovall, 2004). While the notion of user participation in the media is
not new, the rapid explosion of content marks a new trend in news production. By
2009, roughly 2.3 million news blogs existed on the Internet according to the blog
tracker Technorati, and while many of these blogs are inconsequentially small the
collective whole represents a sizeable population (Green, 2009).
From a practical perspective, therefore, this dissertation examines a number
of critical questions facing the news media industry. Considering the community of
organizations associated with the production of news, this study first looks at the
means by which the population of user-generated journalism emerged: was this a
sub-population developed from within the existing journalism community or did
user-generated production develop outside the bounds of traditional news
production? Second, as these two communities coevolved, this research looks at the
impact of alliances and partnerships on the long-term survival of organizations. To
what extent do content sharing agreements and co-production partnerships mitigate
the risks posed by new market entrants, and does the strength of the partnership
factor into survival. Third, this research examines the strategies used by traditionally
hierarchical organizations when faced with rapidly emerging competitive forces. The
role of strategic change imperatives is addressed in media organizations by
5
examining the manner in which traditional print organizations develop user-
generated initiatives, or partner with user-generated or social media producers.
As suggested by both Gans (1972) and Zelizer (2007), facets of these
questions have been posed in previous research, yet both theoretically and
analytically there is need for advancement. Thus, this dissertation contributes
advancement in these areas by extending the dual frameworks of organizational
ecology and social network theory. In addition, this work follows previous
suggestions by integrating network theory with previous ecology research
(DiMaggio, 1994; Monge, Heiss, & Margolin, 2008). Finally, this research was
conducted at multiple levels, examining dyadic interactions between organizations,
population specific interactions within networks of news production as well as social
media, and at the community level, looking at broad macro interactions. Therefore,
this research builds not only on a broad body of literature, but also on a multilevel
multitheoretical framework (Monge & Contractor, 2003).
Using evolutionary theories, scholars have examined many aspects of
organizational change including the emergence of new organizations (Baum &
Singh, 1994b; Carroll & Delacroix, 1983; Usher & Evans, 1996), transformation of
existing organizations over time (Hannan & Freeman, 1984; Haveman, 1992), and
the death or failure of other organizations (Amurgey, Kelly, & Barnett, 1993; Baum
& Singh, 1994c; Carroll & Delacroix, 1982b). Organizational ecology examines
transformation processes in groups of organizations, which are studied as
6
communities focused towards a common purpose or function, for example, on the
production and dissemination of news media.
Communities of organizations change over time based not only on their
interactions with one another, but also as a result of changes in the surrounding
environment (Koka, Madhavan, & Prescott, 2006), the availability of raw resources
(Cheng & Kesner, 1997; Swaminathan, 2001), and the introduction of new
technologies (Lawless & Anderson, 1996), among other factors. Much of this work
focuses on populations and subgroups of organizations, setting aside the effects of
competition from other populations, as well as changes in the competitive
environment. In response to early criticism by Astley (1985), scholars in this area
have started to adopt a community level approach (Ruef, 2000). This work will build
on that foundation, examining the emergence of new forms of organizations as a
process of transformation at the community level. In particular, the speciation
mechanism will be developed as a new theoretical perspective for explaining the
process of form emergence. Scholars have previously noted the potential importance
of speciation as a mechanism for understanding social form emergence, yet few
organizational studies have attempted to deal with the associated complexity of
speciation (Betton & Dess, 1985). Examining today’s complex business networks,
this foundation suggests numerous questions about the multi-faceted shifts occurring
in many industries today, due in part to the influx of new information
communication technology. How do alliances and partnerships formed at the dyadic
7
level shape the macro level development of the community over time? How do
existing and new organizations adapt in the context of networks of organizational
interaction?
Beyond examining the role of organizational networks, this study addresses
how established populations of organizations react to the rapid introduction of new
technology and the emergence of new organizational forms. Simultaneously, this
also brings into question the means by which new organizational forms can emerge
from within established populations. Stemming from this tradition of research, this
work looks at the emergence and coevolution of new organizational forms, seeking
to answer a number of theoretically grounded questions. How do new generations of
communication technology spur the formation of new organizational forms? How do
partnerships between organizations change over time in response to new strategies
and new technologies? The concept of new forms as a type of speciation is extended
and applied as a means for understanding the transformation of organizations ot
organizational populations.
In answering these questions, the core contributions to organizational
research are as follows. The study of organizational change is enhanced with a more
accurate understand of the emergence of new organizational forms and
transformation of existing forms. This perspective is developed from the community
level, illustrating the relationship between intra- and inter-organizational strategies
and macro-level changes. The intersection of organizational ecology and network
8
theories is developed both theoretically and analytically; connections between
network theory and transformation processes are developed, as are methods for
analyzing transformation at the global level. From a practical point-of-view a new
repository of knowledge and a new type of analytical research was initiated. This
research builds on early-stage research analyzing archives of online information, and
in doing so a methodology was developed for analyzing and quantifying historical
data from online sources. Much like traditional libraries and museums, the Internet
can serve as a record of previous events and relationships; this work validates this
type of analysis as a communicative record of alliances and partnerships between
organizations. This archival research is analyzed in conjunction with qualitative
interview data based on discussions with experts in the newspaper industry.
Quotations from the interviews are used throughout this dissertation, but the exact
methods for interview data collection are explained in detail in the methods section.
Thus, this work also provides practical best practices and strategic guidelines for
newspaper managers in an online news environment.
Chapter Summaries
Building on previous theories and research, the above questions are addressed
in the following chapters. Chapters 2, 3 and 4 develop the theoretical framework for
this study while examining the reality of change in the United States newspaper
industry. Chapter 2 introduces the theoretical framework for this study. Theories of
evolution, alliance formation and community ecology are introduced to the reader
9
and paired with network theory to construct a multilevel framework for examining
organizational transformation. Grounded in theories of evolutionary ecology, this
chapter focuses specifically on the process of form emergence. Looking specifically
at the newspaper industry, the emergence of blogs and online newspapers is
examined in parallel to the changing business environment for traditional
newspapers. Chapter 3 builds on evolutionary ecology to explain the processes by
which new organizational forms become legitimized. This process is seen in the
newspaper industry through establishment of online news and user-generated news
as legitimate modes of news production. Chapter 4 discusses the theoretical
processes by which existing organizations transform in response to the emergence of
forms and looks at the changes that traditional newspapers have undertaken in
response to new competition from online news sources. Chapter 5 details the data
collection process, as well as the analytical methods used. Data were collected using
both interviews and historical records. The process of extracting historical network
data from the Internet is detailed, as is the development of a new research tool.
Chapter 6 reports the results of the analysis. Chapter 7 concludes with a discussion
summarizing the research findings and limitations, and setting an agenda for future
research.
10
CHAPTER 2: THE EMERGENCE OF NEW ORGANIZATIONAL FORMS OF
NEWS: AN ECOLOGICAL PERSPECTIVE
“…whilst this planet has gone cycling on according to the fixed law
of gravity, from so simple a beginning endless forms most beautiful
and most wonderful have been and are being evolved.” (Darwin,
2003, p. 460)
Darwin could hardly have conceived that his grand theory of evolution would
provide a basis for conceptualizing the transformational ebb and flow of modern
organizations, yet today evolutionary theories provide a clear framework for the
study of a wide range of organizational phenomena (Hannan & Freeman, 1977).
Darwin suspected that the complexity of nature “guarded against the frequent
discovery of her transitional or linking forms,” (Darwin, 2003, p. 302) and while
recent studies of organizational change reveal much about the nature of
transformation, many of the underlying mechanisms remain unknown. Evolutionary
theories provide an ideal framework for examining mechanisms of transformation
and form emergence by emphasizing the evolutionary process at multiple levels of
organizational interaction. Campbell (1965, p. 26) connected Darwin’s work with
sociocultural evolution, noting that Darwin’s theories of evolution provide an
analogy “from a general model for adaptive fit or quasiteleological processes for
which organic evolution is but one instance.” In this way, traditional models of
ecology provide a foundation through which researchers can understand the
underlying mechanisms that drive the transformation of organizations over time.
11
Following in Campbell’s foundational work on societal processes of
variation-selection-retention, organizational researchers have expanded this theory to
explain organizational evolution by examining interactions between organizations,
parts of organizations, as well as interaction with the surrounding environment
(Aldrich, 1979, 1999; Baum & Singh, 1994a). Just as different species compete for
resources, both against limits imposed by the environment and consumption by other
species, organizations compete against other organizations and the limits imposed by
the surrounding resource space (Ruef, 2000). A primary tenet of this work is the
notion of coevolution: organizations transform over time through interaction with
one another, and as a result of the adaptive pressures driven by competition
(Campbell, 1965). Following this concept, the study of the transformation of
organizations has focused on coevolution at three primary levels of analysis:
organizations (Nelson & Winter, 1982), populations (Hannan & Freeman, 1977) and
communities (Ruef, 2000).
Organizational studies focus on the development of artifacts within an
organization, and the development of those artifacts over time. Nelson and Winter
(Nelson & Winter, 1982) posited that organizations are comprised of a set of routines
that develop over time, with the optimal routines selected and retained over time by
members of an organization. At the organizational level others have looked at
changes in strategy and structure as the result of competitive pressure (Tushman &
Romanelli, 1985b), changes in the function of specific job descriptions (Miner,
12
1991), and managerial influences on the evolutionary process (Boisot & Child, 1988;
Miner, 1994). Population-level research examines mechanisms that affect the
evolution of common groups of organizations and the evolution of those groups over
time. The population approach answers questions regarding issues such as the
development of specialist organizations (Swaminathan, 1995, 2001), competition for
limited pools of resources (Baum & Singh, 1994c; Carroll, 1985; Freeman & Lomi,
1994) and the birth and death of populations over times (Hannan & Freeman, 1977).
Prior research has generally focused on the single population as the unit of
interaction.
Community Ecology
The ecological perspective emphasizes interaction between populations,
communities and the environment (Hannan & Freeman, 1977). Within a given
community environment, various organizational populations interact and compete for
resources. At the same time, populations draw on common resource spaces within
which organizations interact and reinforce one another. The population ecology
perspective specifically focuses on issues of organizational diversity and the
emergence of varying organizational types. The overall focus is on forces of
isomorphism rather than organizational diversity, and as such population ecology
fails to explain the process of new form emergence (Astley, 1985). On the other
hand, the community perspective of organizational transformation focuses on the
assumption that populations of organizations are not self-sustaining systems. Rather,
13
populations are interdependent with one another, and as they evolve they interact
with one another while simultaneously being influenced by the surrounding
environmental space. This view, therefore, allows for the examination of the
formation and dissolution of populations within community contexts (Astley, 1985).
Community ecology, however, views populations as part of a larger ecosystem of
interaction. Populations are the core unit of analysis, and their interactions are
examined within the context of a broader environment (Aldrich & Ruef, 2006;
Astley, 1985). As March (1994) notes, “the convergence between an evolving unit
and its environment is complicated by the fact that the environment is not only
changing but changing partly as apart of a process of coevolution” (March, 1994, p.
43). Previous levels of analysis struggle to fully embody the interaction that occurs
not only between organizations, but also between populations and as a result of
changing levels of resources. Community ecology treats these parts as coevolving
elements of a larger system of interaction.
The community perspective examines interactions across levels, allowing for
a macroscopic analysis of organizational change while still accounting for
mechanisms of interaction at the level of individual organizations. Community
evolution extends from Hawley’s (1986) proposal that ecological analysis must
consider the ecosystem, population and environment within which organization
occurs. Community ecology focuses on the interaction of populations within an open
environmental space. Communities evolve through episodic periods of stability,
14
interrupted by punctuated events that drive organizational populations to transform
(Astley, 1985). In contrast to other perspectives, the community ecology view treats
“variation as an important evolutionary force in its own right. Chance, fortuity,
opportunism, and choice are the dominant factors determining the direction in which
evolution progresses” (Astley, 1985, p. 239). Populations of organizations are thus
part of a broad system of coevolution; populations interact with one another, and are
influenced by the environment within which they are situated. The environment,
including other organizational populations, constitutes the resource space that the
community occupies, and this space impacts all levels of analysis.
The community ecology perspective is thus advantageous for exploring
issues of organizational transformation. Analyzing transformation from the
perspective of a single organization ignores the mechanisms that influence the
development of an organization. Transformation is not an isolated process; it is
important to consider the influence that others’ actions have on the transformation
process, including actions taken by other organizations, and changes in the
surrounding environment. Alternatively, population ecology frames organizational
transformation as a process of selection, whereby organizations within a given
populations compete for various limited resources available within that population’s
niche, and respond over time to changes in different environments. The core question
of population ecology is, “why are there so many kinds of organizations?” (Hannan
& Freeman, 1977, p. 936). Mechanisms of selection drive organizations within a
15
population towards isomorphism, and winnow out unfit organizational forms, yet
this does not explain the emergence of new populations. By looking at the broader
community of populations, community ecology provides a lens through which the
emergence of new forms can be examined. Technological innovations, changes in
norms and values, and the implementation of new regulatory systems are just a few
examples of changes that spur the emergence of new organizational forms and drive
change that cuts across populations (Aldrich & Ruef, 2006). The community
perspective is thus employed here in order to understand the emergence of new
organizational forms, and the subsequent transformation of existing organizations,
Astley’s explication of community draws directly from the biological concept
of synecology, or the study of “whole communities or major fractions of
communities, and ecosystems” (Whittaker, 1974, p. 4). In order to delineate the
different levels of analysis, a community is defined as “a set of coevolving
organizational populations joined by ties of commensalisms and symbiosis through
their orientation to a common technology, normative order, or legal-regulatory
regime” (Aldrich & Ruef, 2006, p. 301). In its original spirit, community was
defined by Whitaker (1974, p. 1) as comprised of various speciated forms interacted
within a common resource space. By Aldrich’s definition, the emphasis is shifted
from geographic boundaries to core functionality. In a technologically distributed
organizational environmental this is increasingly logical as both resources and
organizations are dislocated. This, however, presents an analytical problem, as the
16
community core must be defined in relation to the specific goal at hand. When
implemented in context, community definitions are therefore likely to shift from
study to study (Ruef, 2000).
The community is ultimately comprised of coevolving populations
interacting within a broad environmental space. The environmental space is a
somewhat abstract concept, but refers broadly to the common resources that various
populations draw from. Open environmental spaces are those resources spaces where
competition is at a sufficiently low level in order to enable new organizations to form
with relatively relaxed selective pressures (Astley, 1985). Within the community,
populations coevolve as they transform and react to changes in the environment. A
population of organizations is defined as “bounded sets of organizations with a
common form” (Carroll & Hannan, 2000). Organizational forms are the ecological
synonym for species; forms are the blueprint of the organization, accounting for its
unique structure and the rules, processes and normative order that creates the
organization (Hannan & Freeman, 1977). Through the ecological lens, individual
organizations are viewed as collections of routines that evolve through the variation-
selection-retention process (Miner, 1994). Routines are described by Nelson and
Winter (1982) as programs that constitutes patterns, skills or programs of activities
that are used in the regular activity of an organization. Organizations, and
organizational members, hold a repertoire of routines in reserve, accessing different
routines as required by the functioning of the organization.
17
The adoption of a community perspective provides a framework for a
comprehensive analysis of the mechanisms of organizational transformation. This
encompasses the interactions at the population level, whereby transformational and
emergent processes are driven by changes in the external environment. These
pressures drive organizational populations to form as stable, cohesive collections of
forms (Hannan & Freeman, 1977). The process is driven by the general mechanisms
of variation, selection and retention (V-S-R). Subsequent changes to the goals,
boundaries and activities of organizations, as well as transformations within
communities, populations and singular organizations, can largely be traced back to
these mechanisms. Even processes of symbiosis and commensalism – competition
and cooperation in order to acquire scare resources – occur through variation,
selection and retention, as organizations develop new strategies for resource
acquisition (Aldrich & Ruef, 2006).
Variation, Selection and Retention
From the community perspective, variation, selection and retention are
fundamental processes in the evolution of organizations at any level of analysis.
Communities of organizations move through phases of development towards periods
of stability, or community closure. This stability is interrupted by punctuated events
that disrupt the status quo and drive the emergence of new forms (Astley, 1985). The
process of form emergence is ultimately regulated through mechanisms of variation,
18
selection and retention. While a multitude of factors will impact development over
time, V-S-R lies at the heart of form emergence.
Variation is the driver of the change process, accounting for the generation of
change within and between organizations. There are two primary types of variation:
blind and intentional. Blind variations are independent of both environmental and
selection processes. They result entirely from random processes such as luck, chance
and organizational exploration. Intentional variations, on the other hand, are attempts
by organizations, or people within organizations, to generate change and seek
alternative variations of form (Campbell, 1965; Romanelli, 1991). Selection is the
elimination of certain variations, in favor of other more optimal variations, based on
both internal and external selection process. Internal processes include routines used
within an organization or attempts to promote internal stability. Routines operate as
selection mechanisms when they are enacted; aggregated and enacted routines drive
the essence of the organization, and simultaneously select out sub-optimal routines
(Aldrich & Ruef, 2006). External selection factors include environmental pressures
such as market conditions and competition from other entities (Campbell, 1965;
Romanelli, 1991; Ruef, 2000). Lastly, retention is the process whereby selected
variations are preserved, copied and selected into future generations. Retention
occurs as a process within organizations whereby roles and internal routines become
standardized, and between organizations as managerial processes and technological
competencies diffuse and are copied by other organizations (Romanelli, 1991; Ruef,
19
2000). Routines thus operate as both selection and retention mechanisms: the
enactment of routines is a selection mechanism, but the creation of new routines is a
retention mechanism in that optimal processes are preserved and retained for future
generations.
V-S-R drives organizational change at all levels of analysis, but from a
community perspective this introduction of variation is generally initiated by broad
changes in the surrounding environment. Understanding the effects of changes in the
environment and the resulting interactions between populations allows for an
examination of the emergence of both new organizational forms and new populations
(Astley, 1985; DiMaggio, 1994; Monge et al., 2008; Ruef, 2000). Community
ecology treats variation as a driving force of community growth. In the absence of
selective pressure, variation has the potential to create a wide swath of new
organizational forms and populations (Astley, 1985). Populations interact with one
another within the space of the community. As a community moves towards closure,
populations will establish strong ties amongst themselves. Closure occurs when
populations decrease their dependence on environmental resources and rely
increasingly on the exchange of resources through commensalist and symbiotic
bonds. In this way, populations establish mechanisms for resource exchange and
there is a decreased emphasis on variation, moving the community to stability
(Astley, 1985). The emergence of new forms will then occur as the result of a
marked, rapid disruption in the resource space surrounding the population (Tushman
20
& Anderson, 1986). From the perspective of population ecology, the process of
selection then functions based on phyletic gradualism, selecting out unfit variations
one-by-one moving towards a stable, robust form (Gould & Eldredge, 1977). The
community perspective, on the other hand, acknowledges that the environment plays
a complex role in evolution. Environmental pressures and competition to survive in a
given population will drive organizations to converge to isomorphic forms (Hannan
& Freeman, 1977). But environmental changes open new spaces in the environment
rich with untapped resources, and this ultimately leads to the generation of new
forms.
Thus, the community encapsulates interactions of populations and
organizations. Population compete and interact with one another through
commensalist and symbiotic ties, buffering one another, exchanging resources and
increasing chances of survival within a given community (Astley, 1985). Symbiosis
describes relationships where “individuals complement one another in the
performance of their respective assignments; they enter into mutual dependences
based on their functional differences” (Hawley, 1986). An example of this type of
relationship is the dependency of software manufacturers on original equipment
manufacturers (OEM). Populations of software manufacturers depend on OEMs to
include their software on new PCs, whereas OEMs depend on software
manufacturers to provide necessary components for a finished product.
Commensalist relationships, on the other hand, occur when populations of similar
21
forms engage in “co-actions,” and potentially face a situation of competition for
common resources (Rao, 2007, p. 541). Depending on the degree to which the
resource niches of the respective organizations overlap, there are six different types
of commensalist relationships that can occur, ranging from a state of no effect to a
state of full competition (Brittain & Wholey, 1988). Commensalist and symbiotic
relationships provide a basis for understanding how populations structure and
interact within a community space. A community thus can also be defined as a set of
organizational populations that are linked by a mix of symbiotic and commensalist
relationships within a common environment (Ruef, 2000). In addition, the
community provides an environmental buffer: slack resources remain in the
community space as a buffer against unanticipated shocks providing a degree of
insulation to populations. Over time, populations within a community build networks
of ties between one another, moving towards what Astley (1985) describes as
closure, wherein the system becomes self-reliant and insulated from the effects of the
surrounding environment. Of course, significant shocks to the environment, changes
in technology, or poor coordination among populations and organizations can rapidly
shift a population back into a state of change (Barnett & Carroll, 1995).
In summary, the community ecology perspective applies evolutionary
theories at three levels of analysis: individual organizations, populations of
organizations and communities of organizational populations. Framed from the
community perspective, this work adopts a community perspective to examine
22
transformation in response to the emergence of new organizational forms. The core
mechanisms that drive the transformation process are those of variation, selection
and retention, but the process is also driven by the existing inertia and external
influences (Aldrich, 1999; Hannan & Freeman, 1977), punctuated change (Eldredge
& Gould, 1972), the size, age and density of existing organizations (Hannan &
Freeman, 1977, 1984) and ecological processes of symbiosis and commensalisms
(Aldrich & Ruef, 2006; Hawley, 1986). The emergence of new forms is driven by
many of the same factors, but also by processes of niche formation and resource-
partitioning (Carroll, 1985; Carroll & Hannan, 2000), as well as imprinting,
landscapes and resource availability at the environmental level (Baum & McKelvey,
1999; Kauffman, 1993; Ruef, 2000). These additional processes are each explored in
more detail in the following sections.
Emergence of New Organizational Forms
Forms develop over a period of time through an ongoing cycle of birth and
death. Bryant and Monge (2008) demonstrate that organizational communities move
through four stages of development: emergence, maintenance , self-sufficiency and
transformation. During the period of emergence, populations within the community
are dependent upon resources and legitimacy, and the community is relatively
volatile. When a community reaches the maintenance phase, the community is
expected to grow in terms of cohesion, establishing ties mutual ties between
organizations and populations. Self-sufficiency occurs when the community reaches
23
a general level of stability; populations within the community are shielded from
dramatic change, and the community is able to draw on a stable pool of resources. If
a community fails during the self-sufficiency stage, populations will lose their
cohesion and effectively fail.
The process of transformation can occur at any point, particularly if the
community is subjected to a fundamental disruption. The entrance of a new
population can thus drive a community that has reached self-sufficiency back into a
process of emergence. In line with the logic of Bryant and Monge (2008), it follows
that a community facing the development of a new population would thus reenter the
emergence stage. Bryant and Monge demonstrated this through a staged model of
community development in the children’s television community. Figure 1 illustrates
this staged model of community development. In the model, stages of growth are
determined by the density of communication links between organizations within the
community.
24
Figure 1: Stages of Community Evolution by Network Link Density (Bryant &
Monge, 2008)
In this research, the applicability of the stage model of community development is
first examined to determine its applicability to the online news community:
RQ
1
: To what extent does the evolution of the online news community adhere
to the stage model of community development?
Yet the nature of organizational forms, and the process by which they emerge, is
theoretically murky. From the perspective of community ecology, organizational
forms are the core unit of analysis for ecological studies, providing the basis for
examining both populations and communities of firms. While the focus is on the
Link Density
25
interaction of populations of organizations, the forms that comprise them define the
exact nature of populations. Populations are defined as coherent aggregates of
organizations based on a set of identifying characteristics defined as a form (Hannan
& Freeman, 1986). Communities, on the other hand, are aggregates of populations
built of variants of similar forms (Ruef, 2000). Form constitutes the central
ingredient of ecological studies, yet the concept is vaguely defined and is
problematic as a unit of analysis (McKelvey, 1982; McKendrick & Carroll, 2001;
Romanelli, 1991).
Stinchcombe (1965) first noted that organizational forms exist in diverse
arrays in response to varied market conditions. Subsequently, ecologists have
emphasized the foundations of organizational forms as the source of new
organizational populations (Hannan & Freeman, 1977). Romanelli (1991) provided a
comprehensive review of previous studies, defining the concept of form as the
characteristics that provide an organization with a unique identity, and
simultaneously classifying it as a member of a larger population. Yet despite
previous research no common definition or classification system exists. McKelvey
(1982, p. 458) emphasizes the structural components of an organization, defining
form as “the internal structure and process of an organization and the interrelation of
its subunits which contribute to the unity of the whole organization and to the
maintenance of its characteristic activities, function or nature.” At a more esoteric
level, Hannan and Freeman (1977) advocate classification based on the objective of
26
study, leaving specific determination of what constitutes form up to the researcher.
Both extremes offer valid perspectives; however Romanelli (1991) suggests that a
new theory of forms will likely emerge in the midst of these points of view.
In recent years, researchers have taken a number of different approaches to
specifying forms. One approach has focused on common features of organizations as
a defining element of form (Carroll & Hannan, 2000). This notion of forms as
clusters of features has deep roots in social science, extending back to Weber’s
(1984) conception of bureaucratic organizations as a collection of rationalized rules.
Carroll and Hannan (2000) summarize forms as a special kind of identity,
combining common features recognizable across organizations, and a clearly
delineated boundary that establishes clear limits to the form. More recently, Polos et
al. (2002) define form as the external identity code of an organization, where code is
a pattern that is recognizable by those outside the organization. Forms represent
identities that are established and embedded in organizational communities, and
becomes identifiable through public directories and institutions that monitor
organizations. Thus, previous research has shifted from focusing on forms as core
features to examining forms as a comprehensive identity code. For the purposes of
this study, McKelvey’s definition is used as a starting point given his focus on
organizational structure and associations, and the focus of this research on
relationships between organizations. This definition is expanded slightly to consider
the focus on intangible elements that constitute form. Thus, an organizational form is
27
defined as the structure and associations between different parts of an organization,
but more recently others as well as the intangible codes that comprise the
organization’s identity.
Forms are therefore the building blocks of community-level analysis. Forms
are selected over time and the most successful forms typically are retained and
reproduced. As organizations reproduce a given form, populations are established
and related populations build into communities. The V-S-R process repeats itself,
and populations establish symbiotic and commensalist ties, sharing resources and
build towards stability and closure. Throughout this process, forms evolve and
optimize. For example, with regards to identity, forms have a number of core
features include goals, hierarchies or managerial structures and technologies (Rao &
Singh, 1999). These vary in the ease with which that can be changed, but some
components such as managerial hierarchies can be modified with relative ease
without significantly harming the organization (Carroll, 1984). Further, firms will
develop within populations, moving to occupy specialist and generalist roles
dependent on the particular variation of a form and the availability of resources in a
space. This distinction illustrates that organizations of a particular form can still
vary: specialists and generalists are organizations of a common form that occupy
different resource spaces (Swaminathan, 1995, 2001).
But how, then, do new forms emerge? One primary mechanism is referred to
as a punctuated equilibrium (Astley, 1985). In a punctuated equilibrium,
28
communities build towards stability, which is ultimately interrupted by punctuated
events driving new form creation. The abrupt nature of the disruption drives new
branches (lineages) of organizations to develop at marked points in the community
development (Eldredge & Gould, 1972). Disruptions that lead to changes in resource
availability can drive change either through entire communities, or specific
populations, depending on the level of disruption. Variation in the availability of
resources – capital, employees, knowledge, among others – will have differential
effects depending on the stability of the community and the nature of the change in
resources. When the availability of these resources change, existing forms will vary
in order to adapt to changed environmental conditions. Disruptions include changes
in norms and values, changes in regulatory regimes and new technological
innovations.
Changing environmental conditions create opportunities for the emergence of
new types of organizational forms. When a community of organizational populations
moves into a period of stability, existing organizational forms become entrenched
and learn to work with the existing set of resources available to the community. If
new resources become available to the community during this period of stability,
existing organizations will not take advantage of these resources and therefore new
forms will emerge in open resource spaces (Romanelli, 1991). For example, the
advent of television as a competitor for viewers and advertisers in the 1950s was
succeeded by major changes in organizational form in the motion picture industry,
29
which moved from vertically integrated hierarchies to project-based network
organizations (Waterman, 2005). The adoption of television as a standard form of
media distribution introduced a technological change that drove the emergence of
new forms not only in the motion picture industry, but also in the broad media
environment. This technological innovation essentially spurred the creation of a vast
body of new resources driving competitive forms to vary and evolve. In today’s
business cycle, technological disruptions are increasingly prominent drivers of
change and innovation. Anderson and Tushman (1990) expound on this notion
through their examination of competence-destroying and competence-enhancing
technological innovations, building on previous work by Schumpeter (1945).
Competency-destroying innovations are essentially disruptive events that create new
regimes of technology. Innovations that are competency destroying render
previously existing organizational routines obsolete and force an organization to
adapt to new market conditions. In this way, competency-destroying innovations
drive organizations to establish new forms based on new technology.
Of course not all disruptions are technology driven; as noted, changes in
regulatory regimes and societal norms also created punctuated change. For example
the 1970 passage of the Alternative C proposal established HMOs as an
organizational form of health insurance (Ruef, 2000). Punctuated equilibria can be
examined at both the population and community levels; a community perspective,
however, encompasses the emergence of new forms in open resource spaces. It is in
30
these uncontested spaces that new forms begin to emerge in the aftermath of
punctuated change. However, Ruef (2000) cautions against too much emphasis on
punctuated equilibria. In his study of community ecology, he modeled the
development of healthcare community overtime and accounted for punctuated
equilibria in one model, while disregarding them in another. The results show that
less significant disruptions occur during the equilibria between punctuations, and that
these disruptions can drive the creation of equally viable forms. That said, both
mechanisms are drivers of form emergence (Aldrich & Ruef, 2006).
Speciation and the Emergence of New Forms
Speciation is a critical process for understanding the emergence of new
organizational forms. The analysis of punctuated equilibria seeks to explain what
drives the process of variation. On the other hand, speciation addresses the processes
by which new forms emerge as variations of existing organizational forms. Within
populations, organizations traditionally progress through a process of phyletic
gradualism, whereby variations are selected out over substantial periods of time. On
the other hand, speciation represents a branching as the result of environmental
change, whereby a new population emerges. Considering speciation as a process of
form emergence, Padgett (2001) notes that there is a strong biological analogy to the
process of genetic mutation and variation, but biological studies have largely focused
on speciation as a result of randomized events. Organizational studies, on the other
hand, involve human action, and in turn, agency. From the community perspective,
31
the notion of speciation provides a framework with potential applicability to the
creation of new organizational forms. Expanding on this concept, speciation is the
disruption of within species evolution, leading to growth or separation of a genetic
mutation by a variety of means, although most often through geographic separation
(Coyne & Orr, 2004). From the community ecology perspective, speciation is the
process whereby new forms of organizations emerge as the result of environmental
changes. This can occur as a result of environmental shocks, whereby new variations
emerge in response to changing conditions and resource availability. In addition,
speciation can occur as a result of resource partitioning within the community. Thus,
both speciation and punctuated equilibria seek to explain the process of form
emergence: this process can occur through slow, incremental variation, or through
rapid change resulting from major environmental change. This research focuses on
the latter, utilizing the process of speciation to explain organizational change
resulting from major environmental shocks.
Speciated forms thus represent new branches, but new forms are rarely borne
of nothing; rather, they are based on assemblages of previously successful forms (or
species) (Coyne & Orr, 2004). Thus, new forms may be either strongly speciated or
weakly speciated depending on the degree to which a new form mimics core features
of existing forms (Rao & Singh, 1999). For instance, a form that differs in only one
core feature, such as primary market served, would be weakly speciated from
existing forms. Speciation of organizational forms is the result of a blending and
32
modification of existing features as a result of changes in market or environmental
conditions (Hannan & Freeman, 1989; Rao & Singh, 1999). This is a distinct process
from that of entrepreneurship or imprinting. Speciation pertains to the emergence of
an entirely new form developing in parallel to existing organizations, whereas
entrepreneurs imitate and select elements from previously existing forms and
recombine these elements to form new entities (Hannan & Freeman, 1989; Rao &
Singh, 1999). Likewise, imprinting refers to the process of incorporating features
into an organization at its founding based on successful features in surrounding or
neighboring organizational forms (Johnson, 2007).
Quantum speciation, in particular, is a process whereby the normal
population mechanisms that retard drastic change are restrained, and new forms
branch off (Grant, 1963). Astley (1985) proposed that quantum speciation explains
the failure of existing organizations to react to the rapid emergence of organizations
based on new technology, but this process is in fact a simplification of a much larger
concept. Quantum speciation is in fact a narrow subset of the broader concept, used
to describe speciation that occurs as a rapid, instantaneous mutation in an isolated
habitat, leading to the development of a new species (Rieseberg, 2001). The broader
view of speciation provides a more accurate view of the transformational process
that occurs in community ecology. There are four primary strains of speciation:
sympatric, allopatric, peripatric and parapatric (Coyne & Orr, 2004). Sympatric
speciation is the broad class within which quantum speciation is classified (Bush,
33
1975). Sympatric speciation occurs when a new variation occurs within a given
population and remains isolated within that population, developing over time into a
new species.
This process is akin to a new organizational form emerging within a single
population, and reproducing within that population until a critical mass is reach.
Despite the fact that the new form matures within the given population, it manages
by chance not to develop a unique set of traits not carried by the rest of the
population. This is a rare process, however the development of the modern Web
browser provides an exemplar case. In 1991, five primitive Web browsers existed as
means of surfing the World Wide Web: Erwise, ViolaWWW, MidasWWW, Cello
and NCSA Mosaic. These browsers were a common class of software, competing for
the same pool of potential users. From the start, NCSA Mosaic, developed by the
National Center for Supercomputing Applications, contained a unique mutation in
that it allowed for navigation via graphic buttons. This single mutation coexisted
alongside the other browser for two years, until the end of 1993 when it suddenly
experience rapid growth and quickly emerged as a new form of browser that was
quickly imitated and copied. Thus, much like quantum speciation this new
organizational population matured as a variation within an existing population, but
lead to significant transformation in the overall community as a new type of browser,
and ultimately a new community, evolved.
34
Other forms of speciation further help to define form emergence from the
community perspective. For instance, peripatric and parapatric speciation describe
very similar functions. Analogous to the process of resource partitioning, these
processes of speciation occur when a variation emerges within a resource niche and
develops to maturity over time at the periphery of a resource space. In peripatric
speciation, the variant form is completely isolated and distinct from the parent
population, whereas in parapatric speciation the variant form may commingle with
the parent population and thus share some attributes in common. From a biological
perspective, it is often hard to distinguish between instances or peripatric and
parapatric speciation (Gavrilets, Li, & Vose, 1998). Lastly, allopatric speciation
occurs when there is a natural barrier protecting a variation and allowing it to mature.
This would occur in community ecology when an entirely new resource space gives
rise to a new population of organizations.
Ultimately, this view of speciation is largely unexamined, but represents a
promising mechanism for understanding the different maturation processes through
which new forms emerge. Although Astley (1985) presents this as a potential
mechanism for understanding the emergence of new forms, the four variants
suggested here represent a potential extension of theory. Levinthal (1998) considered
additional exploration of speciation, using it as an explanatory mechanism for
understanding technological innovation, but his analysis did not consider specific
variants of speciation. Speciation is additionally complementary to previously
35
explored concepts of form emergence. In previous research, it is estimated that only
10 percent of evolutionary change in biology has occurred as a result of speciation,
whereas the remaining 90 percent has occurred through anagenesis (phyletic
gradualism) (Simpson, 1944). Yet advancements in research surrounding punctuated
equilibria now suggests that although speciations may drive only a fraction of new
form development, these forms are particularly robust having emerged in periods of
marked disruption and competition (Gould, 2007).
In aggregate, the four models of speciation thus describe cases where
variations emerge from an existing population. The biological view posits that
speciation will occur as the result of a mutation due to a change in environmental or
genetic conditions. This is akin to the emergence of new organizational forms as the
result of environmental changes such as the introduction of new technology and new
resources. Here then, speciation is used as a mechanism for understanding the
process by which the new technologies that led to user-generated processes of news
production generated a new organizational form. In order to understand the
development of new forms of news production it is necessary to understand the
macroscopic development of this community as a whole. From this perspective, it is
then possible to examine the process of form emergence through the speciation of
new forms in open resource spaces.
36
News Media in the Community Ecology Framework
It is within this community framework that the evolution of the news media
industry is examined. In recent years, the news media community has experienced
significant shocks, facing new competitors and shifting resources, largely as the
result of new forms of organizations entering into the competitive space. While there
is a long history of research examining issues pertinent to the journalistic
community, there are few studies that have examined this community from a macro-
level perspective. For instance, Epstein (1974) conducted the first in a series of
scholarly ethnographies in the 1970s, which illustrated the dangers of corporate
influence on story production in television newsrooms .
Subsequently, scholars have examined the power of public interest groups to
shape news creation through interaction with reporters (Goldenberg, 1975), the
creation of the daily news agenda through the structure of the editorial process
(Gans, 1979) and observing the daily routines of journalists to illustrate the daily
manufacturing of news information (Fishman, 1980). This school of research
produced studies examining the manner through which news shapes the way
consumers perceive the world within which they live. Tuchman (1978, p. 12) makes
the compelling argument that the production of news is “the act of constructing
reality itself, rather than a picture of reality.” Each of these studies has taken an intra-
organizational perspective, looking at the history of either a single organization or
newsroom, or a small subset of organizations.
37
More recently, news media scholars have undertaken research examining a
number of macro-level changes. Demers (1996; 1998) and Demers and Merskin
(2000) looked at the effect of changes in ownership structure, and found that the
transition from private to public ownership did not significantly affect the content
and coverage of newspaper. Their study did not, however, look at the impact on
profitability or survival. Similarly, Boczkowski (2004b) has more recently looked at
changing practices in news organizations as the result of the introduction of new
technology. He found evidence that the work practice and day-to-day routines of
individuals within newsrooms has transformed significantly based on changes in
technology. From a different perspective, Dimmick’s (2003) work utilized niche
theory from the evolutionary perspective to explain how media organizations
compete for advertising dollars and simultaneously coexist with one another by
forming into generalist and specialist categories. Dimmick’s work is one of the few
existing media studies to directly adapt ecological studies. Work by Carroll and
Delacroix (1982b; 1983) also used an ecological perspective to examine the founding
and mortality rates of newspapers in Argentina and Ireland.
In addition to the above, media ecology has provides a cultural perspective
for examining media impact in a societal context. Media ecology holds that “media’s
impact on culture is formal and environmental while people’s modes of thinking and
social organization are shaped by the dominant modes of communication they
internalize” (Lum, 2006, p. 10). Media ecology is rooted in the work of McLuhan
38
(McCluhan, 1962) and Postman (Postman, 1979, 1982, 1985), and is focused on the
exploration of a given medium as an environment in which culture is represented and
created. The structure of a particular medium is said to entail characteristics that
structure the representation of culture and society, and in turn, the use of a particular
medium entails biases that shape one’s view of society (Lum, 2006). Media ecology
has traditionally separated the audience from the medium, yet in today’s era of
participatory media many are rethinking traditional tenets of media ecology in order
to better incorporate the full scope of user interaction with media. In addition, media
ecology takes a cultural perspective (Flayhan, 2001); as such, it has limited
applicability when examining macro-level structure. With the exception of early
studies pertaining to framing, media ecology studies and Dimmick and
Boczkowski’s more recent work, news media research as of late is largely void of
strong theoretical underpinnings.
The extant body of literature examining journalistic practices and
organizational structure illustrates the changing nature of news production, yet lacks
comprehensive analysis at the macro level. Klinenberg (2005) observes that research
in journalism has been scant in the past few decades as scholars have focused on
sociological media studies. Community ecology theories thus provide a framework
for examining the transformation of the news media industry during a period of
transition. The community of news production consists of a multitude of populations
associated with the manufacture and distribution of news and information. This
39
includes traditional print newspaper organizations, online-only news weblogs, wire
services, and television and radio news services. While this research is focused on
the interaction of these organizations in the online space, both online and offline
interactions are constitutive elements of the community.
Thus, the previous section outlines community ecology, and the process of
form emergence. This section builds on the theoretical discussion to illustrate the
emergence of new forms within the existing news media community in the United
States. This work is particularly focused on the emergence of online news, social
media and user-generated news as competitive forms to traditional newspaper
production. As noted in the above discussion, new forms emerge in response to
major changes in the environment, as well as through gradual changes over time.
Punctuated change, however, creates significant opportunity by opening up resource
spaces and disrupting previous stable forms. In the case of the news media industry,
therefore, newspaper publishers existed for many decades as a stable population,
within a fairly stable and closed environment. Here then, the recent emergence of
new forms within the news media community is examined stemming from a
punctuated change in the community at large. The history of the print newspaper
industry is first considered, but additional attention is paid to competitive industries
such as television, radio and magazines.
40
A Stable Community: A Brief History of the Modern News Media
Today’s news media community in the United Sates is, in fact, an evolution
of the printed news community that has developed over centuries in the United
States. The news industry has been in existence globally since the Romans first
published Acta Diurna, a handwritten daily news letter that was posted in the Roman
Forum. The incarnation of news in the United States was more directly a variation of
printed new in Britain. The modern news industry originated in Britain in the mid
17
th
century, first with the publication of daily corantos (digests of foreign affairs),
and then diurnals (digest of local affairs) which began appearing after regulatory
changes allowed for the creation of a press with relaxed government restrictions
(Schwarzlose, 1987). As early as 1665, news letters were printed by colonial
governments to disseminate news and information (Shaaber, 1954). News sheets, in
one form or another, actually started circulation in the United States in 1690 with the
short-lived publication of Publick Occurrences Both Foreign and Domestic. This
was the first free-press publication issued in the United States. Although Publick
Occurances began publication in 1690, the daily newspaper industry did not form in
earnest until 1704. The first newspaper to print in continuous publication was the
Boston News-Letter, which started printing in late-1704 (Pickett, 1977).
Early news publications in the colonies took many forms, ranging from single
page flyers to dense booklets chronicling the events of the day and promoting
various ideological positions (Ford, 1954). As colonists continued to construct new
41
systems in the United States, these forms varied based on market conditions and
available resources, including consumer demand. Through the period of the
Revolutionary War, this ad hoc publishing system continued to dominate the news
industry. In 1783, Benjamin Towne converted the Pennsylvania Evening Post from a
triweekly newspaper into a daily newspaper, and launched the first daily news
publication in the United States (Kobre, 1944). In ecological terms, this marked the
emergence of yet another variation of the printed news organization. The daily
newspaper required organizations to develop new routines to cope with the pressures
of daily publication (including producing enough content to publish). In fact, the
early Pennsylvania Evening Post lasted merely a month. A competing paper, the
Pennsylvania Packet, overtook the Evening Post and quickly ran it out of business.
These daily papers bear no resemblance to today’s newspapers; the Evening Post
consisted of half a small page of newsprint printed on both side with two columns of
text (Mott, 1950).
Changes in regulatory regimes further bolstered the growth of daily
newspapers. The ratification of the U.S. Constitution in 1791 guaranteed press
freedom with the First Amendment, yet in the early days of the republic this press
remained highly partisan: “post-Revolutionary press freedom guarantees were
intended to protect a partisan press from its political enemies in government and to
protect the mutual dependence of press and government as a central mechanism to
ensure representative government and an informed electorate” (Schwarzlose, 1987,
42
p. xx). Early press freedoms were thus intended to protect the newly formed
government, and reinforce that institution, as much as it was intended to protect a
free press. This partisan system remained strongly entrenched through the 1830s.
Newspapers in this era generally consisted of broad sheets of newsprint and ran from
four to eight pages in length. Newspapers were sold on a subscription basis,
averaging about six cents an issue. The largest paper of the day, the New York
Courier and Enquirer, had a daily circulation of roughly 4,500 (Mott, 1950). By the
late 1820s, major newspapers began to move from a traditional Gutenberg press to a
more modern steam-driven cylindrical press; this enable newspapers to begin
printing upwards of 2,000 copies an hour. Mass printing, paired with an increase in
middle-class skilled labor, drove the emergence of the penny press through the 1830s
and 1840s (Shudson, 1978).
From the 1830s through to the 1860s, the penny press began to develop as a
notable force in news media. The New York Sun, founded in 1830, was the first
penny press newspaper to appear and quickly grew to a massive circulation of
20,000 within two years. The affordable nature of the penny press (usually no more
than a few pennies per copy) spurred the popularity of newspapers; this paired with a
growing middle class demanding affordable and reliable information, lead to an
erosion of the entrenched partisan press that had allowed papers such as the
Philadelphia Enquirer, Providence Journal and the New York Enquirer to maintain a
stranglehold on their regional markets. A key instrument in this movement was the
43
banding together of seven New York City newspapers (including four penny press
papers) to form the New York City Associated Press (AP) between 1846 and 1851
(Schwarzlose, 1987). By the time of the Civil War in 1861, penny papers began to
fully challenge the political influence of the traditional partisan press, in large part
due to the type of mass production facilitated by the development of the AP.
The Civil War changed the nature of American media and drove the
emergence of the modern press (Smythe, 2003). During the Civil War, the American
press corps emerged as a primary source of information for citizens on both sides.
Increased competition from telegraphic news services also pushed newspaper to
begin shifting from evening distribution to a system of morning publication with
late-afternoon update editions. In addition, the need for updated information during
the Civil War drove newspapers to begin publishing Sunday editions (Baldasty,
1992). In the post-war era, from 1865 to 1900, newspapers began to organize into
more formal corporate structures. Newspapers began to rely on advertising revenue
and a regular flow of information from wire services. In addition, the candor and
openness of the penny press had launched a counter-movement to the partisan press,
and by 1900 a powerful and free press had emerged as a dominant engine of news
production (Baldasty, 1992; Smythe, 2003). This period was marked by leaders such
as Joseph Pulitzer, who developed “new journalism.” This was the movement
towards a formal process of local-news gathering, an organized sales staff to
44
generate advertising dollars, and a mindset of selling news through innovative
reporting, editorials and investigative journalism (Schwarzlose, 1987).
Technological innovation further rocked the newspaper industry around the
turn of the century. In 1886 the Merganthaler Linotype, a machine for automatically
setting letters for printing on paper, was introduced to the publishing industry. This
enabled automatic page layout and rapidly increased the speed with which a page
could be produced (Schwarzlose, 1987). In the 1890s, the stabilization of community
resources further enabled the growth of the newspaper industry. Newsprint, in
particular, was a dominating factor in determining newspaper availability. From
1860 through 1868, newsprint prices had risen to more than 25 cents per pound, but
by 1890 the price had normalized back to roughly 3 to 5 cents per pound (A. M. Lee,
1937). Ink and printing equipment prices similarly continued to decrease, enabling
the further growth of a mass press. By the turn of the century, today’s modern
journalism had more or less been established as the accepted form of news
production. By 1910, 2,433 daily newspapers existed in the United States with a
cumulative daily circulation of roughly 24.2 million. By the beginning of World War
I, this number peaked at more than 2,600 daily newspapers. With so many daily
newspapers publishing, this amounted to an average daily circulation of roughly
9,000, and was not tenable in the long run (Schwarzlose, 1987). The market at this
point was oversaturated: from an ecological point of view, the environmental space
had become overcrowded and there were not enough resources to sustain the
45
continuation of the market in its current size. From 1916 to 1943, the number of total
daily newspapers declined dramatically to a stable level of roughly 1,700
newspapers. This was driven by over saturation, increased production costs due in
part to large-scale unionization, increased competition for advertising dollars, and
the continued decline of evening newspapers (Schwarzlose, 1987).
The overall pattern of industry stabilization is illustrated in Figure 2, which
shows the number of daily newspapers in publication from 1783 through 2005. The
dotted lines in the diagram from 1930 through 1990 result from the fact that during
this period reliable census data for US newspapers was released every five years.
From 1940 through to the early 1980s, the number of newspapers in circulation in
the United States remained relatively constant.
46
Figure 2: Number of Daily Newspapers From 1783 – 2005
0
500
1000
1500
2000
2500
3000
1783
1830
1870
1875
1880
1885
1890
1895
1900
1905
1910
1915
1920
1925
1930
1935
1940
1945
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
Year
Number of Daily Newspapers
Source: Newspaper Association of America
During the period following World War I, the news media community in the United
States had essentially reached closure. Relatively few new entrants launched
newspapers during this time, and there were relatively few exits. A few exceptions
include the launching of Long Island Newsday in 1940, the Chicago Sun-Times in
1941 and USA Today in 1982. All in all, however, the size of the overall community
did not grow significantly, but during this time period individual organizations
within the industry experienced healthy growth. The industry entered into a slow
decline beginning around 1980. Circulation, which measures the average number of
subscribers receiving a newspaper daily, grew 53 percent industry-wide from 1940 to
1985, as illustrated in Figure 3.
47
Figure 3: Newspaper Circulation with Breakouts for Morning and Evening
Newspapers
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
1940
1946
1948
1950
1952
1954
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
Year
Circulation (000)
Morning Evening Total
Source: Newspaper Association of America, 2009
In perspective, however, this growth is not all that remarkable given the
demographic changes of the time. Notably, according to Census data the population
of the United States grew 81 percent from 1940 to 1980, while the number of
newspapers grew a mere 5 percent. While the total number of newspapers remained
steady from 1940 through the 1980s, the population exploded. In 1945, 36.5 percent
of the population subscribed to a newspaper but by 1984 only 26.8 percent
subscribed. All in all the news media industry remained fairly stagnant with most
growth rendered insignificant due to the explosion in potential audience during this
48
period. Newspapers created the institution of daily news production, and had grown
and thrived in the United States for over a century, yet from 1940 onwards the
industry stalled:
Commercial press reporting had gradually slipped into the rut of safe
factuality or objectivity by the 1930s, and, when the American
Newspaper Guild, the reporters’ union, appeared in 1933, reporters
and editors were already concluding their first decade of
inconclusive debate on whether journalism was, or ought to be, a
profession, a debate still raging in the 1980s. (Schwarzlose, 1987, p.
xxxii)
Through the 1970s, however, the newspaper population remained largely
entrenched within the new media industry as a primary provider of daily
information. Competition from magazines and radios did challenge
newspapers, but overall newspaper remained the dominant form; each of
these forms is addressed in following section. The news media community as
a whole moved into a period of relative community closure during this
period. There were a few notable events that occurred during this period that
affected the overall shape of the news media community. First, newspaper
organizations continued to shift from evening production to morning
production. As illustrated in Figure 2, the decline of evening news began in
earnest around 1970s, and by the 1980s morning newspaper had overtaken
the industry as the dominant form of news. In addition, the weekly newspaper
continued to lose its relevance as a major news product. From roughly 1913
onwards the number of weekly newspapers in circulation in the United States
49
trended downwards, falling from more than 16,000 in 1913 to roughly 7,000
by 1980. Weekly newspapers had previously been a reliable source of in-
depth and investigative news, but this task shifted to the dailies in the era of
“new journalism.” Most of the remaining weeklies have minor circulation
levels. Daily and weekly newspapers were not the only populations within
the news media community during this time period. Magazines have been a
mainstay of news, although they often focus on in-depth reporting as opposed
to day-to-day news. Print magazines have circulated in the United States
since the Saturday Evening Post began printing in 1821, and by 1900 there
were more than 5,000 magazines in circulation. Newsmagazines, however,
emerged as a major source of information in 1923 with the formation of Time
magazine (Barnhurst & Neron, 2001). U.S. News and Newsweek both
entered into competition a decade later. Radio, which had emerged during the
1920s, became a formal institution in 1927 with the formation of the Federal
Radio Commission (FRC). The first news reports had aired on radio in 1920,
but with the formation of the FRC the National Broadcasting Corporation
(NBC) and the Columbia Broadcasting System (CBS) became legitimate
institutions of news (Marquis, 1984). In addition to the major networks,
affiliate stations were launched across the country to serve local markets. The
major commercial networks served as the backbone of the radio industry; as
of 1945, 95 percent of all radio stations were affiliated with a network and
50
delivered content produced by that network. At the end of WWII, however,
the Federal Communications Commission (FCC) relaxed a number of radio
restrictions and opened up FM bands for commercial use. Independent
stations began to emerge in regional markets, supported in part by the growth
of a local advertising market (Leblebici, Salancik, Copay, & King, 1991). By
the late 1950s, more than 2,000 AM stations and nearly 600 FM stations were
in operation around the United States producing a vibrant wealth of local
news and information.
Radio broadcasting had, however, largely peaked in popularity in the period
from the 1920s through the late 1930s. By World War II, radio news was losing its
dominant power; the growth in stations resulted in a fragmented audience and
consumers began to turn to other sources for news (Lewis, 1992). Television, which
had emerged as a medium in the 1930s, quickly established itself as the next great
source of news and information. 1952 was the first year that many considered
television to be a mass medium. In that year, the number of television sets in homes
reach 50 million. The radio industry suffered as many talented producers and
broadcasters moved from radio to the rapidly emerging television broadcast business
(Leblebici et al., 1991). By the 1960s the three major networks, ABC, CBS and
NBC, had established nightly newscasts. Then, in 1967, the assassination of John F.
Kennedy proved to highlight the power of television as a source of news and
information when 93 percent of television homes tuned in to the broadcast of the
51
funeral. In the 1970s television began to expand its news programming and started to
transmit via satellite. New technology increased the quality of the picture, and also
made television sets more affordable; during this time, television news continued to
develop as a mainstream medium. Broadcasters such as Walter Cronkite and David
Brinkley became household names; a 1972 survey found that Cronkite was the most
trusted person in America (Baym, 2010). Cable television, meanwhile, continued to
grow as a dominant force. Cable television first developed in the 1960s, but cable
news became mainstream in 1980 with the launch of the Cable News Network
(CNN). CNN was the first true 24/7 television news service, followed by the launch
of the Financial News Network (FNN) in 1981 and ABC’s Satellite News Channel
(SNC) in 1982. Cable television continued to grow as a news source, but exploded in
the 1990s, particularly after providing nearly continuous coverage of the first Gulf
War in 1991(Parsons, 2008). Today, the media audience of cable news for the top
three networks – Fox News, MSNBC and CNN – is 3.64 million adults (Edmonds et
al., 2010). Remarkably, despite the entry of radio and television populations into the
news media community, the newspaper industry remained stable up through the
early-1980s.
This history illustrates the evolution of the news media industry in the United
States. In the course of the 20th century, the media industry as a whole faced
significant upheaval. Entire industries emerged at a remarkable pace, including radio,
television and cable television, and more recently online news and social media.
52
Despite continued turmoil in the news media industry, the print newspaper industry
remained relatively stable for a good part of the century. From 1940 onwards, the
newspaper industry remained stagnant despite technological innovations and new
competitors for the news market. Early on, some had foreseen the challenges facing
the newspaper industry. As noted in 1947, “The speed, quantity, and variety of mass
communication will continue to increase. Long since, the volume and variety of
words and images have exceeded the capacity of any individual consumer to
assimilate them,” and yet consumers continued to rely on newspapers as a primary
source of daily news and information (Hutchins et al., 1947, p. 35). As this history
illustrates, newspapers had established a dominant organizational form for the
production of news. Newspaper organizations traditionally consisted of two primary
divisions: business operations and news production. The business division handles
the day-to-day financial operations, advertising, human resources and production.
The news production division is responsible for the production of news. This
organizational split is often referred to as the church and state separation of
newspaper organizations. The news production division thus is free to focus on news
production through the ‘beat system.’ Individual reporters, or small sub-teams, are
assigned to cover a given topic or region for a newspaper. These reporters cultivate
sources, develop story ideas and produce daily news (Tuchman, 1978). Newspapers
generally cover a particular region, although larger newspapers, such as the New
York Times, are considered to be national newspapers, and arguably could be
53
classified as international newspapers. Regional newspapers focus on a particular
city or region of the country, but they may have bureaus in other major cities to
supplement the news. Lastly, news is also supplemented by ‘wire services,’ which
aggregate news from a large network of reporters and feed it to subscribing
newspaper organizations. The dominant form in the business was focused on
morning news, with reporters filing stories in the late-afternoon and evening and the
newspaper printing overnight.
From the 1940s on, this was the form of daily newspapers: a single
organization producing daily newspaper through a beat system of reporters, with a
separate division handling business operations. Stability in the industry has been
driven, in part, by standardization of form across different markets, particularly in
the end product. Regardless of the language, a consumer can travel from country to
country, pick up a morning newspaper, and recognize the same general features.
Internally, this standardization is echoed in technical, organizational and
communication practices (Boczkowski, 2004a). By 1985, the population of
newspaper in the United States consisted of roughly 1,700 daily newspapers
operating on a basic variation of this organizational form. As this brief history has
illustrated, the news media community consisted of other populations besides
newspapers: radio news stations, broadcast news, cable news, weekly newspapers
and news magazines all coexisted within the same resource space. The broad
community also includes the thousands of reporters, broadcasters, technicians and
54
executives who worked within this space, as well as tangential populations that
supplied raw materials for production. As the previous examinations of the number
of organizations and the overall circulation has demonstrated, through the early
1980s, the news media community functioned in near-closure: the predominant beat
system of news production and the daily news cycle had been reproduced across
radio and television news.
One of the most notable risks of community closure is the lack of incentive to
innovate, and a tendency towards complacency (Aldrich & Ruef, 2006; Morgan,
1986). In the news media community, the population of newspapers reached an
approximate state of closure, as evidenced by the relative lack of transformation or
realignment in response to the introduction of competitive populations such as radio
and television news organizations. As a result, for many years the form of newspaper
organizations was unchanged. In the ecological sense, community closure sets a
community up for the possibility of collapse. In a community near closure,
populations have formed extensive ties between one another, and in doing so have
created a buffer between themselves and the environment (Astley, 1985). In the case
of newspapers, this resulted in an inadequate response to environmental changes.
Numerous studies have shown that although newspapers sought to evolve during the
1980s and 1990s, attempts at change focused on fitting new innovations into existing
systems, rather than adapting existing systems to new methods of production
(Boczkowski, 2004a, 2004b; Patterson, 2007; Sylvie & Witherspoon, 2002). In turn,
55
few notable innovations emerged within the newspaper publishing industry that
directly impacted the form of printed news. The invention of the linotype printing
press in the late 19
th
century was the last major innovation to lead to a reorganization
of newspaper organizations.
The Emergence of Online and Social Media as Organizational Forms
The heyday of the newspaper industry lasted into the early 1980s, with year-
over-year growth in advertising revenue, relatively stable circulation, and profit
margins that continuously remained in the double digits. Even though circulation for
evening newspapers began to decline as early as 1968, total newspaper circulation
across evening, morning and Sunday newspapers continued to grow slowly through
the early 1970s and remained relatively stable into the 1990s. Since then, however,
circulation has marched steadily downwards, heralding a similar decline in
advertising revenue, which was stagnant as early as the mid-1990s. Readership
statistics, which measure the number of people actually reading and consuming
newspapers, has also started to decline, although not as rapidly. According to the
Newspaper Association of America ("Trends & Numbers: Total Paid Circulation,"
2007) the average weekday readership for all newspapers in 2006 was 124 million
consumers, equivalent to a little more than half of the population of the United
States. While still a significant percentage of the population, readership dropped 11
percent from 2000 to 2007, further illustrating the decline that is emerging on the
print side of the industry. As readership and circulation has decreased, so too has the
56
number of newspapers in production; the total number of newspapers has dropped
from more than 1,600 in 1990 to roughly 1,500 in 2006. In all, these statistics
highlight an increasing trend of readers migrating away from print newspapers as a
primary source of news information. According to the Newspaper Association of
America (J. Murray, 2007) the average weekday readership for all newspapers in
2006 was 124 million consumers, equivalent to roughly 40 percent of the United
States at that time. This is clearly a significant portion of the population, but
readership had dropped 11 percent since 2000. In addition, total newspaper
circulation in the three main categories (evening, morning and Sunday) has dropped
steadily since the 1990s after decades of relatively stable circulation. In fact,
morning newspaper circulation continued to go up through 2002 before starting to
decline. As general readership has trended downward, so too has the number of
newspapers in production; the total number of newspapers in the United States has
dropped from more than 1,600 in 1990 to roughly 1,500 in 2006 (World Press
Trends, 2007).
Numerous causes have been posited as the trigger that led to the decline of
the newspaper industry: industry stagnation, a lack of innovation, entrenched
leadership and archaic organizational structures to name a few. But ultimately the
newspaper population has lost resources due in large part to the emergence of new
organizational forms such as online news, news blogs and social networking sites. At
first, many newspapers saw Internet technology as a complement to the printed
57
newspaper population. Internet technology, it was predicted, would complement the
printed press and allow reporters to enhance their articles with information queried
from databases of archived knowledge (Barnhurt, 2002). By 1995, 57 percent of
reporters used online information weekly in their reporting. In the same survey,
however, only 17 percent of newspapers reported that they had developed an online
edition (Ross, 1998). In 1997, it was noted that “not a single major paper has even
put email addresses at the end of stories so that readers can communicate easily with
reporters, a simple addition most papers have had the technological capacity to do
for years.” (Katz, 1997, p. 58).
The introduction of new technology delivered a shock to the newspaper
population, but the industry responded with measured innovation rather than any
significant restructuring (Boczkowski, 2004a, 2004b; Patterson, 2007; Sylvie &
Witherspoon, 2002). In 1993, the first graphic Web browser, Mosaic, was launched,
and by 1999 more than 4900 newspapers globally had launched online version of
their newspapers. The Internet itself has roots stretching back to the launching of
Arpanet in 1969; this early network was primarily used by government entities and
academic researchers as a tool for communication and collaboration (Abbate, 1999).
Researchers continued to develop new protocols and standards for transferring
information on the Internet through the 1970s and 1980s. In 1991, however, Tim
Berners-Lee introduced a new language that allowed users to more readily share
information in a visual format. This protocol, known as hypertext markup language
58
(HTML) was freely distributed and launched as the WorldWideWeb (WWW)
(Stovall, 2004). The launching of Mosaic, however, promoted the rapid public
adoptions of the WWW as a standard for sharing information online.
The newspaper industry, in general, viewed online technology as a new
medium for distributing an existing product, and for nearly a decade newspaper
organizations experimented with online forms of news distribution (Boczkowski,
2004a). In 1997 and 1998, the first variations of a new organizational form emerged
online in the form of weblogs, more broadly referred to as blogs. In the 1980s,
newspaper companies had experimented with new forms of news production such as
teletext, newsgroups and customized e-editions (Boczkowski, 2004a), but weblogs
were one of the first forms to dramatically reinvent the form of daily news. But the
Internet, and subsequently the World Wide Web, were not tools born of the
newspaper industry. Rather, these technological disruptions originated as tools in
government and research communities. By 1994, early innovators were using Web
pages as tools for online diaries and personal commentaries. In 1997, Jorn Barger
launched Robot Wisdom, which featured a listing of links that Barger liked to visit,
as well as daily updates from Barger’s daily life (Rettberg, 2008). Similar types of
sites began to pop up en masse, but generally failed to attract large audiences. The
term “blog” was first used to describe these sites by Peter Merholz in 1999; the term
was intended to denote a site as a “wee - blog.” A high barrier to entry plagued these
early sites; a user seeking to build a blog was required to have a sufficient amount of
59
technological expertise in order to build and maintain the site. In October 1998,
however, Open Diary was founded to offer users with free hosting and easy-to-use
online publishing options. Within four months, the site had 25,000 hosted online
diaries. Pitas launched in 1999 offering free blogging tools, followed by the launch
of Blogger. Traditional newspapers continued to adhere to the strict routines of
printed newspapers, but blogs allow writers and reporters to share opinions and
publish relatively raw content outside the bounds of journalistic hierarchy. Early
blogs were relatively simple hypertext documents updated on a relatively frequent
basis, with content ranging from a few roughly assembled sentences to complete
magazine-length features (Matheson, 2004).
Blogs were a relatively early form of interactive media; social networking
represents a second disruption that emerged later on in the life of the news media
community. Social networking is a specific type of online community. General
online communities have been in existence since the late 1970s, but because of many
of the same usage problems that plagued blogs, online communities did not gain an
strong following early on. The earliest online communities, text-based forums and
multi-user domains (MUDs) actually predate the World Wide Web; USENET groups
were text-based discussion forums that existed as early as the late-1970s, although
these early forums lack many of the characteristics of modern online communities
(Turkle, 1995). More recently, MUDs are text-based communities oriented towards a
particular topic or theme. Participants connected to each other through early Internet
60
connections and phone lines, but participation was relatively low. MUDs represented
a more interactive version of USENET, and allowed for synchronous interaction
(Kollock & Smith, 1999). Starting with the introduction of graphic-based chat
communities, online communities began to grow in popularity; the subsequent
adoption of the World Wide Web as a standard platform for online access further
accelerated the development of online communities.
In 1997, the Web site SixDegrees.com launched, allowing users to create
profiles and connect with other friends on the site. SixDegrees is generally credited
with being the first social networking site (SNS) (boyd & Ellison, 2008).
Subsequently, numerous imitators emerged, and many were successful in improving
the social networking platform. In 2002, Friendster.com launched and quickly
gathered a following of more than 300,000 users. From 2003 onward, social
networking sites began to emerge as mainstream media platforms, due largely to the
development of Web 2.0 technology. Web 2.0
1
technologies are a class of platforms
that enable consumer participation and interaction in online environments, including
discussion and creation of the news.
Today, social networking sites (SNS) “connect and present people based on
information gathered about them, as stored in their user profiles” (O’Murchu,
Breslin, & Decker, 2004). Boyd and Ellison (boyd & Ellison, 2008) distinguish
social network sites as Web sites that allows users to (1) create a public-facing
1
Web 2.0 is commonly defined as the “network of interconnected devices and applications that enable the
production, consumption and remixing of technologies at both the individual and group level, ultimately leading
to an architecture of participation” (O'Reilly, 2005).
61
profile, (2) construct a list of users to whom they are connected, and (3) navigate lists
of connections for individuals and their connections. More broadly, social
networking sites are online resources that allow users to create ‘maps’ of their social
networks, and to share information through these networks. The above definitions
devolve to two key elements: interactivity (connect people, construct lists of
contacts, navigate lists) and public presentation (present people based on information
gathered, create public-facing profiles). Thus, social networking sites are defined
here as Web sites that provide users with a space for interaction, with three primary
attributes: public profiles, tools for linking to other users, and tools for maintaining
relationships.
The Disruption of the News Media Community as a Speciated Event
The majority of this chapter has provided a detailed accounting of the
development of the newspaper industry in the United States. Newspapers have a
storied history in the United States, and despite numerous technological innovations
in the 20
th
century – the mass adoption of radio and television, and the introduction
of cable television – the newspaper industry has remained robust. Each of these
individual events are essentially punctuated events, driving the creation of new forms
of communication. Returning to Eldredge and Gould’s (1972) interpretation of the
concept, the rapid introduction of a new innovation (disruption) occurs at a marked
point in time, leading to changes that ripple through the community. When cable
television was introduced for example, a new form of news dissemination was
62
introduced in the form of cable news. This type of disruption simultaneously expands
the community resource space; for example, competition between cable news and
print news reduces the total amount of time that consumers have to spend with any
given medium. As a result, the availability of time as a resource is reduced for the
community as a whole. The disruption that has ensued through the introduction of
and increased competition from online news has, however, punctuated the
community’s resource space so divisively that the entire community has been
disrupted. From the community perspective, punctuated events lead to the emergence
of new forms in open resource spaces, and drive change through shifts in resource
availability (Aldrich & Ruef, 2006).
The punctuated equilibrium mechanism is helpful for explaining the initial
interaction between an existing community and a new organizational form. In the
case of the newspaper industry, the immediate and rapid growth of participatory
media as a form of news was a shock to the industry. Yet blogs and social
networking sites emerged over a period that spans nearly a decade. They did not
directly impact the news industry until the late 1990s, in part because as an
organizational form they did not compete for the same resources during this
incubation period. Furthermore, during this time period there were organizations
within the industry that were aware of the potential competition posed by new
innovations. As demonstrated, newspapers worked to develop new modes of
production and distribution in the 1990s, experimenting with e-paper formats and
63
electronic delivery. Boczkowski (2004a) points to New Jersey Online’s Community
Connection, a joint venture of the Newark Star-Ledger, Trenton Times, Jersey
Journal and News12 New Jersey, as one of the earliest online community ventures
driven by a newspaper company (Advance Publications). Community Connection
launched in 1998, and by 1999 it had a full-time staff of more than 30 employees.
In general, however, participatory media such as blogs was not commonly
recognized until the early 2000s, and even then many were unclear of what impact
they would have in the long run. According to one commentator, writing in 2002,
“[blogging] will drive a powerful new form of amateur journalism… It won’t happen
overnight, and we’re now seeing only version 1.0” (Lasica, 2002). Existing theories
of punctuated events fail to fully explain the mechanisms behind this type of
disruption. The introduction of participatory media created a punctuated equilibrium
in the news media community, yet the effects of that disruption were not
immediately felt. In this context, the development of participatory media as a form of
news is a speciated event. Returning to the earlier discussion, speciation is the
process through which new organizational forms develop as variations of existing
organizations forms; a disruption leads to a branching from existing forms, and as a
result new forms emerge outside the boundaries of existing communities. As the new
speciated form develops, it matures to a critical tipping point at which it may again
come into competition with the existing community (Coyne & Orr, 2004; Mayr,
1982).
64
Participatory forms of news production thus developed outside the
boundaries of the traditional news media community as a speciated organizational
form. The development of these forms was sparked by technological disruption that
created an open resource space for form development. This occurred when new
Internet technology allowed users to create their own Web sites online, and
affordable technology and access created a vast group of users seeking content. In
terms of organizational ecology, disruptions include new means of manufacturing
products, as well as routines for generating output. Levinthal (1998) showed that
critical events occur when an existing technology is applied to a new domain. Within
the new domain, the reapplied technology is able to develop into a new technological
form. This process of technological branching is akin to speciation, whereby the
“application domain” shifts such that the new technological form develops in an
open resource space absent of competition from existing organizations in the original
domain (Levinthal, 1998, p. 223). When the new form reenters into competition with
an existing population a process of creative destruction occurs whereby older
technologies become obsolete in the battle for resources. In the case of online news
production, user-generated forms of news production emerged as the result of new
technology seen in Web 2.0 communication tools. In addition, existing news
organizations were firmly entrenched in existing routines of production, and thus not
attuned to new routines, or the availability of new resources in the form of users-as-
producers.
65
Thus, it can by hypothesized that development of participatory media as a
competitive population in the news media community was the result of a speciated
event. This is in line with the previous discussion of technological branching,
following Levinthal (Levinthal, 1998). Looking further at the biological analogy,
Coyne and Orr (2004) write that parapatric speciation is the specific process by
which new variation emerges and develops in an adjacent but separate resource
space. Figure 4 illustrates the process of parapatric speciation. In the first stage,
variations emerge in a new niche creating a branch from the existing community. In
the second stage of development, the variation evolves and the successful variation
grows in numbers, until the third stage when the variation reenters into competition
with the existing community. Competition between the speciated form and the
existing population occurs once a tipping point has been reached; that is, the new
form to a point where competition with the existing community is inevitable.
66
Figure 4: Parapatric Speciation in an Established Community (adapted from Coyne
& Orr, 2004)
original
population
initial
speciation
growth of
speciated form
renewed competition
leads to equilibrium
The speciated form remains in competition with the existing community until a new
equilibrium is reached. It can be hypothesized then that, following the introduction
of a new speciated form, the new form is likely to have relatively little contact with
the existing community. In the case of online news media, one would expect that
during early development blogs had relatively few organizational links to existing
news organizations. In other words:
H
1
: During the emergence phase, new organizational forms are more likely to
link to other organizations of the same type than to different types.
New organizations developing as a speciated form will thus link to other forms of the
same type, but will not link to previously existing organizational types. More to the
point, the speciated form will be less likely to link to the dominant organizational
form during formational periods. It is not sufficient, however, to simply state that
this process of form creation is driven by the speciation of existing organizational
forms.
67
Speciation and Available Resources
Expanding on Levinthal’s (1998) original work, speciation is developed here
as a mechanism to further explain the processes that lead to the creation of new
organizational forms on a longitudinal scale. The resource landscape is central to the
speciation mechanism, and in turn, is a critical component of form emergence.
Through the process of speciation, new organizational forms are founded where
there is perceived to be a niche of open and available resources (Hodgson, 2010).
From the perspective of organizational ecology, the mechanism of speciation is thus
entwined with form development through the perspective of resource partitioning
theory. Levinthal does not directly reference resource partitioning theory, yet his
focus on the emergence of new forms in open resource spaces is directly analogous
to resource partitioning (Carroll, Dobrev, & Swaminathan, 2002).
Resource partitioning theory posits that the development and emergence of
new forms is largely driven by resource availability; subsequently, the competition
between existing organizational forms and generally resource availability further
explains the development over time of generalist and specialist organizations
(Swaminathan, 2001). Given a finite amount of resources, certain firms will choose a
broad target of market segments, whereas others will choose to focus competitive
efforts in more narrow bands of resources. Those firms that choose broad resource
spaces will compete directly with one another, and the fittest firms will survive as
68
generalist organizations, occupying a wide resource space and generally offering a
variety of products. If firms targeting a narrow resource space overlap with generalist
firms, they either face direct competition for resources or are forced to further
concentrate on a specific niche for which there is less competition. Firms that target
resource spaces towards the fringe of the market will generally have a greater
likelihood of survival and will ultimately develop as specialist organizations (Carroll
et al., 2002; Carroll & Hannan, 2000). This echoes Hawley’s (1986) isomorphism
argument, which states that organizations within a given field will tend to adopt the
same form as other organizations within close proximity, primarily as a result of
institutional pressures and the need for legitimacy.
It follows then that the form of a population is driven to a large extent by
environmental factors, such as competitive pressures and resource availability
(DiMaggio & Powell, 1983). Furthermore, when conditions in the surrounding
environment are unstable, organizations will tend to develop a generalist structure in
order to draw on a wide swath of available resources, whereas in situations of intense
competition organizations are more likely to develop niche strategies and compete
for narrow pockets of resources (Hannan, Carroll, & Polos, 2003). Over time, a
competitive landscape emerges in which a core set of organizations occupy a central
position competing for resources as generalists. At the fringes of the community,
specialist organizations compete. New forms must develop where open niches exist
69
due to institutional constraints, market forces and openings created by existing
competition (Brittain, 1994).
From the perspective of speciation, when new forms emerge they will likely
develop in open niches. Except in rare cases of speciation, however, this process
does not necessarily occur in isolation.
2
Returning to the earlier discussion,
speciation is thus more likely to occur through processes of peripatric or parapatric
speciation, whereby the new organizational form develops in relative isolation, but
will still have intermittent or infrequent ties back to the original population. These
ties occur because total isolation rarely occurs; rather, in the course of natural
development interaction and commingling is expected, even if rarely. In this way, the
emergent form will have new characteristics, but may carry certain traits forward
from the original population (Coyne & Orr, 2004; Gavrilets et al., 1998). In fact,
recent evidence suggests that when the process of speciation occurs rapidly it is
driven more by resource availability than isolation or founding events (Gavrilets et
al., 1998). It would follow that speciated organizations that develop in open resource
spaces would be tied to the adjacent community through connections to specialist
organizations. As discussed above, specialist organizations occupy fringe resource
spaces within a community; in a rapidly changing environment a speciated form
seeking resources will thus be likely to establish ties with these fringe organizations.
2
Allopatric speciation is a variation of speciation that occurs when a new population is fully separated from its
parent population. Examples of this type of speciation are seen in animal population where new breeds develop
after being separated by a geographic barrier (rivers, lakes, oceans, mountains), but generally this is very rare
(Mayr, 1982).
70
This process of tie formation between specialist organizations and emergent
organizational forms is predicted in the following statement:
H
2
: During the emergence phase, there will be a small, but significant,
number of links between a new organizational form and existing
specialist organizations compared to links in the community as a
whole.
The process of speciation typically culminates with the new form entering into full
competition with the existing community; the new form will generally become
numerous enough that competition for resources is inevitable. Over time, therefore,
one would expect that a new speciated form would grow in density prior to
establishing a high proportion of linkages to the existing community (both specialists
and generalists). In other word, a new organizational population must reach a critical
mass before it is able to successfully form relationships with existing populations.
This is summarized in the following prediction, where the critical mass of a new
form is postulated to be a matter of growth in density:
H
3
: Links between a new organizational form and the existing community
will increase after the density of the speciated form begins to increase
following emergence.
The creation of links between the existing community and the new speciated form
would indicate that the process of form development is nearing completion. It is
expected that the two forms would then compete until a new equilibrium is reach
71
within the community. The three outlined hypotheses thus test the mechanism of
speciation; the formation of a relatively disconnected emergent organizational form
in an open resource space; the formation of initially scarce links to the existing
community; and the eventually reintroduction of full competition.
Speciation occurs because open resource spaces become opportunities for
innovation and entrepreneurship. Existing populations within a community develop a
degree of organizational inertia. In other words, organizations within a longstanding
population develop routines for work process. The longer a population is in
existence, the more routinized an organization within that population becomes. Over
time, organizations begin to develop set routines for changing and innovating, and as
a result these organizations are less likely to detect or adapt to new competitive
forces that differ from what has previously been encountered (Amburgey, Kelly, &
Barnett, 1993; Hannan & Freeman, 1984). The newspaper industry in the United
States was stable for most of the 20
th
century. As a result, newspaper organizations
developed static routines for producing news, and for adapting to new
communication technology. With the advent of the World Wide Web, and the
subsequent introduction of Web 2.0, a new form of news production began to
develop. Speciation is an explanatory mechanism that details the process by which
online news developed as a new form and eventually sparked an industry-wide
reinvention.
72
CHAPTER 3: LEGITIMIZATION OF NEW FORMS – THE RISE OF ONLINE
NEWS
New Media and New Forms
The total number of blogs grew from zero in 2002 to 15.5 million active
blogs in 2009, and far more that are inactive (Green, 2009). It is unclear how many
of these blogs pertained to the news or newspapers, or how many of these blogs had
a significant following. It’s additionally unclear how many blogs are able to generate
revenue. A 2007 report estimates that the bloggers with more than 100,000 readers
earn an average of $75,000 a year. The bloggers behind Overheard in New York earn
roughly $97,000 a month, and celebrity gossip blogger Perez Hilton, on the high end,
earns an estimated $1.3 million a year (Agger, 2008). Despite a high degree of
variance, in a period of less than seven years blogs have clearly developed as a
legitimate and significant organizational form. The scale, rapid growth and variance
of the population of bloggers, and online news in general, have called into question
the boundaries of news in the online space:
Publishing used to require access to a printing press, and as a result the act of
publishing something was limited to a tiny fraction of the population… Now,
once a user connects to the internet, he has access to a platform that is at once
global and free. (Shirky, 2008, p. 77)
As Shirky observes, the rapid reduction in communication costs in the publishing
industry has created confusion about who is and is not a legitimate journalist. In the
same way, the rapid rise of organizations founded on blogging, social networking or
73
online news has further created confusion about who is and is not a member of the
organizational community of news.
The news media industry has expanded and realigned in an ecologically rapid
period of barely thirty years, and at the heart of this rapid evolution lies the
emergence of online news production and participatory media. For those functioning
in established populations it is therefore difficult to distinguish a legitimate
organizational form from a passing fad. Previous studies of organizational founding
and organization death have examined the evolution of organizations over a period
of decades; indeed, Carroll and Delacroix (1982a; Delacroix & Carroll, 1983)
studied the birth and death of Argentinean newspapers over a period of 150 years. In
the online news environment, however, change happens at a pace that far exceeds
what has generally been considered in evolutionary studies. For instance, a 2003
study of 150 million Web pages found that over an 11 week period 62 percent of the
pages in the study had significant changes made to one or more attribute such as text,
images, and links (Fetterly, Manasse, Najrok, & Wiener, 2004). Online social
networking sites such as Facebook.com, Myspace.com and Orkut.com are now
among the most popular online sources for online information: 60 percent of adults
use online sources as their primary news source, and of those adults nearly half turn
to either blogs or social networking sites first (Edmonds et al., 2010).
3
3
Study measured all online news users 18-years-old and older.
74
With the rise of online-based organizations, traditional companies such as
media organizations rushed to the Internet. In the late-1990s, companies quickly
launched new ventures and signed into partnerships with emergent organizations in
order to avoid being left behind by the growing tide of online information and
commerce (Boczkowski, 2004a; S. Murray, 2003; Pentland, Fletcher, & Hasson,
2004; Rheingold, 2003). The introduction of online technology created a significant
disruption in the existing business ecosystem, spurring the formation of thousands of
new enterprises. Yet the lack of standards, short time period in which new
organizations emerged, and lack of knowledge about new technology created
industry-wide confusion about the legitimacy of new organizations (Phan, Schmidt,
& Chen, 2010). For instance, Javalgi, Todd and Scherer (2005) found that following
a rapid increase in the registration of e-commerce organizations, online business
experienced a realignment from 2001 to 2003 when more than 580 e-commerce
organizations declared bankruptcy.
Legitimation as an Ecological Process
In a period of roughly ten years the online organizational landscape has thus
experienced significant upheaval, and organizations continue to struggle with
processes of legitimation. The previous chapter examined how new forms emerge,
but as evidenced in the process of variation-selection-retention, emergence does not
indicate longevity. Over time, new forms compete for limited resources and are
replicated as they are deemed successful. Thus, successful forms grow in size and
75
number, and over time new organizational forms become legitimate within a
community. Legitimacy is critical to the long-term survival of an organization; the
conferral of legitimation provides recognition for an organization and establishes it
as part of an existing and generally stable population. From the perspective of
organizational ecology, legitimacy is defined by Suchman (1995, p. 574) as “a
general perception or assumption that the actions of an entity are desirable, proper, or
appropriate within some socially constructed system of norms, values, beliefs and
definitions.” Carroll and Hannan (2000, p. 8) further add the legitimacy is “a social
process by which organizational forms become institutionalized or socially taken for
granted.” This definition is further refined by Aldrich and Ruef (2006, p. 230)
expanded on Suchman’s definition, proposing that legitimacy can be classified as
either cognitive or sociopolitical. Cognitive legitimacy is the acceptance of a new
organizational form as “a taken for granted feature of the environment.” In other
words, cognitive legitimacy occurs when the community at large generally accepts a
new organizational form as part of the community. Sociopolitical legitimacy, on the
other hand, is acceptance by stakeholders within the community including the public,
opinion leaders and political stakeholders. Aldrich further divides sociopolitical
legitimacy into two categories: moral and regulatory. Moral acceptance is seen
through the action of society and public figures, whereas regulatory acceptance is
indicated in government policies and actions. Cognitive legitimacy is generally a
process of organizational learning; new entrants into a community seek to increase
76
awareness about new business practices, technology and products by educating
existing firms and sharing knowledge. Sociopolitical legitimacy, on the other hand,
occurs through strategic actions taken to integrate a new form into the community
both legally and socially (Aldrich & Ruef, 2006).
Traditional organizational ecology has focused on the process of density
dependence as the key factor in establishing legitimacy of an emergent
organizational form (Carroll & Hannan, 1989). The basic tenet of density
dependence is that a new organizational form gains legitimacy as the population
grows in density within a given resource space. As a new organizational form
becomes more successful, other imitators will follow suit and enter into competition,
thus driving an increase in the organizational birth rate of a new form. As more and
more organizations enter into competition with one another, resources become scare
and ultimately the mortality rate of organizations will increase. Thus, the density
dependence argument predicts that as organizational populations grow in density,
there will be a U-shaped relationship between the density and the failure rate of the
population. This implies that there will be an inverted U-shaped relationship between
the population’s density and the founding rate (Hannan & Freeman, 1986).
Traditional ecological studies of populations have thus found that institutions
such as railroads, banks and newspapers have tended, historically, to emerge in small
clusters, develop rapidly into larger populations and then stabilize or decline slightly
(Carroll, 1984). Organizations tend to emerge in a density dependent pattern for a
77
number of reasons: limited resources may facilitate a rapid period of initial growth,
only to lead to a decline as resources diminish; changing social conditions enable
rapid periods of growth as technology and policies change; lastly, populations grow
based on density-dependent process of legitimation and competition (Carroll &
Hannan, 1989). When new organizational forms emerge, the theory of density
dependence posits that as the density of a form increases, competition will increase
and do so at an accelerating rate. Simultaneously, as density increases legitimation
of that form will increase at a decreasing rate (Carroll & Hannan, 2000). The density
dependence mechanism, as originally conceived, functions as a population-level
mechanism. From the community perspective, there is some evidence that the
existence of an analogous form in an adjacent population will have a symbiotic effect
on form emergence of a similar type in another population (Baum & Oliver, 1996;
Ruef, 2000). In other words, if adjacent populations have experienced similar
disruptions, or the emergence of similar organization types, then the focal population
will be more likely to accept the new form’s emergence.
Studies of legitimation at both the community and population levels have
thus focused extensively on the density dependence process of form legitimization.
Density dependence, however, is an empirically retrospective framework for
understanding the process of form legitimization. In other words, it is only possible
to recognize the existence of a legitimate density after that density has been
established and it can be counted. Ultimately, this poses a critical flaw when
78
applying density dependence theory to the study of emergent organizations. Ideally,
it would be possible to set a threshold density at which point a form is said to be
legitimate, however because characteristics of any given population are unique to
that population it is nearly impossible to set a uniform threshold. For instance, in a
study of emergent organizational forms, McKendrick and Carroll (McKendrick &
Carroll, 2001) found that the determination of legitimacy based on density often lags
other measures of legitimacy, in terms of both cognitive and sociopolitical
legitimacy. Their study examined the development of disk array manufacturers in the
computer industry over a twelve-year period. Based on media coverage and
recognition from professional organizations, it was clear that a new form of
computer technology had begun to take shape, yet from a density perspective it was
still unclear as to whether a legitimate form existed. Their study highlighted many of
the issues associated with the organizational lifecycle of technology- and online-
based organizational forms. Birth and death rates for these types of organizations
fluctuate significantly, and as a result the development of a stable density can often
lag significantly.
Consider, for example, web-based e-commerce organizations: given the low
barriers to entry, and minimal requirements for organizational entry, Internet-based
organizational forms can emerge in dense numbers in very short periods of time.
However, Internet-based forms also evolve rapidly, which poses a significant
challenge to sustaining a stable and established form even in a densifying population.
79
Brown and Eisenhardt (1997) point to the rapid changes in the computer industry in
the 1990s, with firms such as Hewlett Packard changing from instrument producers
to computer firms through process of continuous and rapid change. New technology
enables rapid development of innovations on both small and large scales (Eisenhardt
& Tabrizi, 1995). In populations marked by rapid innovation or rapid product
development, a form can become established before it is recognized by outside
institutions. This is exactly the type of development pattern that exists with regards
to online news and participatory media. From 2002 to 2007, for example, the number
of blogs grew rapidly and quickly developed into a robust population of online
information sources.
Other mechanisms can therefore be used to detect the emergence of a
legitimate organizational form prior to the establishment of a legitimacy based on
density. McKendrick and Carroll (2001) theorize that two distinct processes confer
legitimacy onto emerging organizational forms. First, formal organizations, such as
associations and regulatory bodies, can assist in the establishment of new forms.
Second, organizational density can lead to the legitimacy of emergent organizational
forms. The focus here is on the first mechanism. Associations and regulatory bodies
can confer legitimacy, provide resources differentially to different forms, and/or
mandate particular forms. Press coverage (Pollock & Rindova, 2003), recognition
from professional organizations (Rao, 1998), and legal acceptance (Ruef, 2000) can
each provide a form of sociopolitical or cognitive legitimacy prior to a population
80
gaining legitimacy through densification. More importantly, these types of
recognition can help to identify emergent organizational forms and drive the process
of selection.
Media Coverage
Press coverage of firms within a given population serves to increase broad
public awareness of a given industry. Utilizing the press as a communication
channel, organizations within an emergent population are able to disseminate a
coherent story about the function of a new organizational form and its benefits for
the community at large. Previous research has established that activities or
organizations that receive significant media coverage are more likely to be perceived
as legitimate, even in instances when media coverage is fleeting. Furthermore, a
study of biotechnology startups found that startup firms that received increased press
coverage in a given year were more likely to receive higher amounts of capital
through initial public offerings (IPOs) in the following year (Deeds, Mang, &
Frandsen, 2004). Ultimately, media coverage of a new organizational form has a
legitimating affect because it directs public attention. In this way, media coverage is
likely to be a precursor of densification by filtering new market entrants and guiding
public and industry attention to those that are perceived to have the most potential.
Sustained press coverage will further increase familiarity with a new organizational
form, and educate relevant audiences about its functionality. This will in turn
81
continue to drive a growth in the organizational birth rate of a new form (Pollock &
Rindova, 2003).
Press coverage of blogs and social networking sites illustrate one way in
which media coverage can be indicative of the development of a new organizational
form. The blogosphere emerged en masse as a space for online information
production between 2001 and 2002 (Rettberg, 2008). Web sites that are driven by
user-generated content, such as blogs, represented the largest online growth sector
from 2006 through 2009. For those in the news media community, however, it is
unclear when blogs were considered a legitimate part of the community. Given the
previously discussed trends, it is clear that the bulk of growth in terms of density
occurred after 2005. From 2006 onwards there was clearly a rapid explosion in the
density of blogs within the online media community. Yet media coverage of
blogging did begin to increase as early as 2003 and 2004, as illustrated in Figure 5.
82
Figure 5: Press mentions of blogging and social networking (1998 – 2009)
0
20000
40000
60000
80000
100000
120000
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Year
Number of Articles
Blogs Social Networking Sites
Source: Lexis-Nexis, 2010
4
Between 2004 and 2005, the number of US newspaper articles mentioning blogging
exploded, growing threefold. Exponential growth continued through 2007, when the
explosion in media coverage began to slow. Online news Web sites such as
TalkingPointsMemo.com and HuffingtonPost.com have since emerged as opinion
leaders in the online news space. Furthermore, the growth in new coverage of social
networking sites lagged when coverage began to increase in earnest in 2005, marking
the year that MySpace.com was acquired by News Corp for $580 million (boyd &
Ellison, 2008).
4
Articles pertaining to blogs contain “blog”, “weblog”, or “blogging” in the headline or main body. Articles
pertaining to social networking site contain “social networking site” or “social networking sites” in the headline
or main body.
83
Media coverage of an emerging organizational form is thus expected to be an
early indicator of legitimacy of a new organizational form. As the above discussion
suggests, increased press coverage confers legitimacy onto a new organizational
form leading to an increase in the birth rate of a new organizational form. In the case
of the online news media community, this is postulated as follows:
H
4a
: Media coverage of blogs and social networking sites will be positively
related to the birth rate of online-based news forms.
Professional Associations
Industry associations and professional organizations can serve as mechanisms
that detect and identify the emergence of new organizational forms. Professional
organizations in a given population have generally been established as authorities on
the norms and structures of a population. Thus, recognition of new forms by a given
professional organization serves as a direct conveyance of legitimacy onto a new
organizational form. Trade organizations, in particular, oversee the scope of
membership for a given community. Members of a trade organization can thus
decide to expand the boundaries of a given population by voting to include or
exclude new organizational forms (Rao, 1998). In addition to professional
organizations, suppliers and regulatory agencies can also impart legitimacy.
Suppliers serve as gatekeepers, essentially controlling access to resources.
Regulatory agencies, on the other hand, can set population boundaries by governing
who may or may not compete.
84
Trade and professional associations are particularly important as early
indicators of legitimacy because they have relatively low overhead and can be
quickly established when a new form begins to grow in density. In Carroll and
McKendrick’s (2001) examination of the disk array industry, for instance, the
formation of the Storage Networking Industry Association (SNIA) and the RAID
Advisory Board (RAB) established two professional organizations which served as
governing bodies for the emergent industry. Professional organizations can create
awareness for a new organizational form, and can also provide coherence in creating
an organizational identity. In the case of the online news media community, for
example, the Online News Association (ONA) was founded in 1999, and today it has
more than 1600 members. The online news media community, however, is a large
community, and as new forms have developed a host of other professional
associations have developed. The Media Bloggers Association was established in
2004, The Professional Bloggers Association formed in 2005, and more recently the
World Professional Blogger-Journalist Association was established in 2007.
Following in the work of Aldrich and Fiol (2007), among others, it follows
that the formation of professional associations will lead to an increase in legitimacy
for the associated organizational form. This is hypothesized as follows:
H
4b
: The formation of professional associations will be positively related to
the birth rate of online-based news forms.
85
Legal Rulings
In addition to media coverage and the formation of professional associations,
legal rulings and legal recognition can serve to establish the cognitive and
sociopolitical legitimacy of a developing organizational form. For instance, Ruef
sites government influence as one of the key factors in determining legitimacy of
new organizational forms. According to Ruef (2000, p. 671), “given the legal-
rational authority of the state in modern society, its recognition of an organizational
form as a legitimate (or illegitimate) class of collective actors is often one of the
most significant events in highly institutionalized arenas.” When emergent
organizational forms are shown to adhere to legal norms, the act of adherence serves
a public statement of legitimacy. Organizational populations can demonstrate a
commitment to legal validity by adhering to existing legal rulings, or new
organizations can seek exceptions to existing legal charters in order to gain
legitimacy (Delmar & Shane, 2004; Salancik, 1977).
Consider the development of an active blogger-journalist community. Since
blogs first emerged in 2002, individuals have used blogs as a tool for distributing
citizen-reported news. As blogs have grown in density it was unclear whether or not
citizen-reporters and bloggers-journalists were protected by the same legal measures
as reporters. In California, for example, reporters are protected under California’s
Reporter’s Shield Law. This law stipulates that publishers, editors and reporters
employed by a newspaper, magazine or periodical cannot be held in legal contempt
86
for refusing to disclose the name of a source of information that is subsequently
published. In O’Grady vs. Superior Court, the State of California appeals court ruled
that a news blog that had published articles, which included trade secrets about an
unreleased Apple product, were protected under California’s Shield Law. Apple
Corp. had sued the news blog in order to obtain the identity of the information
source, claiming that the news blog was not protected under California’s Shield Law
(Rushing, 2006).
California is the only state to constitutionally protect a reporter’s privilege,
yet nearly every state and most federal jurisdictions have some form of equivalent
law. O’Grady vs. Superior Court is generally thought to have set a mandate that
bloggers acting in a legal capacity as journalists are similarly protected by privilege
(Alonzo, 2005). Numerous other legal rulings have impacted the status of bloggers
and online news sources. As previous research suggests, when new organizational
forms are able to gain inclusion in existing legal charters it will serve to increase
legitimacy. This is hypothesized as follows:
H
4c
: Legal recognition will be positively related to the birth rate of online-
based news forms.
Media coverage, professional associations and legal acceptance are three
mechanisms that can drive organizational legitimacy for new forms. Recently,
notable research has focused on the emergence and legitimization of entrepreneurial
or innovative organization forms. Previous studies have emphasized the importance
87
of media coverage (Deeds et al., 2004; Pollock & Rindova, 2003), professional
recognition (Aldrich & Fiol, 2007; McKendrick & Carroll, 2001; McKendrick, Jaffe,
Carroll, & Khessina, 2003) and legal acknowledgment (Delmar & Shane, 2004;
Ruef, 2000; Salancik, 1977). For emergent technologies, however, the end goal of
innovation and entrepreneurship is to develop an organizational form such that it
both embodies the rules of the existing society and aligns with societal norms and
expectations (Swidler, 1986). Historically, new organizational forms have been
shown to succeed in gaining legitimacy when they can be classified into preexisting
cultural norms within an existing community. Further, forms are likely to succeed
when their growth functions as a coevolutionary process in accordance with changes
in the density of the existing community (Ruef, 2000). Certain organizational forms
develop as constituent forms or “quasi forms.” In this scenario, new organizational
forms develop as a subset of existing populations within a community (Aldrich &
Ruef, 2006). One such example exists in the U.S. health care community, where
trauma centers function as a quasi form of hospital emergency rooms (Ruef, 2000).
Trauma centers were able to gain acceptance as a new healthcare form by adhering
to existing practices, but also by functioning as a quasi form in that many attributes
of trauma centers were borrowed directly from emergency rooms and hospitals.
Speciation and Legitimization
In the case of Internet-based forms, emergence has occurred rapidly from
outside the existing population. As explicated in the previous discussion of
88
speciation, it is posited that such forms developed in open resources spaces where
they were relatively isolated from existing organizations (at least initially). Because
speciation occurs as a result of variations influenced by the environment, the course
of evolution is not dictated by previous acts of variation and selection (Carroll &
Hannan, 2000). Rather it is previous actions in combination with influences from the
environment that guide development of organizational forms; thus there is a blending
of path dependency and environmental influence. A process is said to be path
dependent when its development depends upon a previous occurring series of unique
events (Aldrich & Ruef, 2006, p. 238).
Through the establishment of linkages and formal ties to existing legitimate
organizational forms, new organizational forms can actively seek to gain legitimacy.
In a study of new investment firms specializing in socially responsible investment,
for example, researchers found that firms that established links to mainstream
investment firms were more likely to succeed than those that did not establish such
relationships (Dejean, Gond, & Leca, 2004). Conversely, adjacent populations can
confer legitimacy on new organizational forms. User-produced news sites, for
example, can obtain legitimacy by forming alliances and relationships with
traditional news organizations. Following McKendrick and Carroll’s (2001)
predication that legitimization can be conferred by outside authorities, it is plausible
that recognition of Internet-based forms may be more likely to occur through
legitimization by adjacent populations.
89
When new forms emerge through the process of speciation, proximity to
other forms is a key determinant of long-term success. If a new form is too close in
proximity to other similar forms, then the emergent form will lack the necessary
resources to grow. Yet ties to existing forms are critical for gaining access to
resources, and ultimately for reaching a critical mass to gain legitimate recognition
(Barrett, 1989). New forms are isolated during emergence, but as forms begin to gain
recognition, ties to established populations become a key mechanism for gaining
legitimacy. A new form growing in density will rapidly deplete resources within its
niche; ties provide a mechanism for sharing information, conferring legitimacy and
ultimately gaining access to new resources (Aldrich & Ruef, 2006; Baum & Oliver,
1992; Carroll & Hannan, 1989). Thus, as previously proposed, the speciated form
must reenter into competition with existing populations in the community (Coyne &
Orr, 2004; Gavrilets et al., 1998). Thus, much like innovative organizational forms or
entrepreneurs, ties to the existing community will be a critical measure of legitimacy.
The establishment of ties is thus a critical indicator that a speciated form has
reentered into competition. This is hypothesized as follows:
H
5
: An increase in the number of links between existing organizations and
new organizational forms will be positively related to the birth rate of online-
based news forms.
The establishment of ties is a critical mechanism for the establishment of legitimacy.
The successful growth of a form in the long-run, however, is still related to the
90
process of density dependence. When forms emerge through speciation, variations
emerge in small numbers, and then as they are successful they rapidly grow in count
and stabilize over time. This echoes a density-dependence analysis of evolution in
that the growth and legitimization of a new population is, in the long-run, directly
dependent on the raw count of organizational forms (Carroll & Hannan, 1989;
Delacroix & Rao, 1994).
Density-dependence holds that a form reaches “constitutive legitimation”
when it has reached a critical mass whereby it is able to act as a collective unit
(Carroll & Hannan, 2000). Framing variations as a process of speciation, density-
dependence accounts for the impact that organizational interactions such as the
recruitment of capital and the search for new employees and formal endorsements
have on survival of the form. Furthermore, as density of organizational forms grows
over time, competition for resources will increase. This invokes the notion of
carrying capacity, in that a given population’s resource niche can support a maximal
number of organizations (Carroll & Hannan, 2000). Density dependence can thus be
viewed as complementary to the speciation framework. As a mechanism for
understanding form emergence, speciation predicts that forms will develop in a
manner such that density dependence alone is not a sufficient mechanism for
understanding the process of legitimization.
91
Online News: Moving Towards Critical Mass
The previous discussion integrates the process of speciation with the density
dependent process of legitimization. Traditional mechanisms including media
coverage, legal recognition, the formation of press associations, and the subsequent
passage of time, will lead to an increase in legitimacy of new organizational forms.
In addition, however, mechanisms of speciation, including the formation of links
between populations, are key to establishing legitimacy, particularly in rapidly
emerging populations. In online news, for instance, blogging is now considered a
ubiquitous medium for online information. Web sites and online news sources have
challenged the newspapers for consumers attention since the mid-1990s (X. Li,
2006). In 2001 and 2002, however, blogging entered a period of rapid growth fueled
by easy-to-use graphic interfaces, high data transfer speeds and a continued decrease
in the cost of computers (Rettberg, 2008). This has enabled the rapid growth of user-
communities and user-driven content on a scale rarely seen. Web sites driven by
user-generated content represented the largest online growth sector from 2006
through 2009. The total number of blogs grew from zero in 2002 to 15.5 million
active blogs in 2009, and far more that are inactive (Green, 2009). More specifically,
Flickr.com, an online photo sharing site, grew 201 percent in 2006, up to 6.4 million
visitors, and Wikipedia, an online knowledge sharing site, grew 181 percent up to 29
million users (C. Li, 2007). The growth of social media sites suggests that a new
form of production emerged in the form of community news media production.
92
While the notion of user participation in the media is not new, the rapid explosion of
user-driven Web sites marks a new trend in news production.
Early blogs were not a news-specific medium, but the rapid growth and
continued experimentation by users ultimately led to the development of blogging as
a tool for news reporting. Today bloggers often provide a channel for first-hand
accounts from those who cannot witness an event in person. Blogs allow individuals
to tell stories that are not covered by the mainstream press, and blogs also allow
individuals to filter news and provide commentary on current events (Rettberg,
2008). Within newsrooms, blogs are also used as a replacement for previous types of
news production, and as a complement to existing online news forms (Reese,
Ruigliano, Hyun, & Jeong, 2007). Blogs were formally acknowledged within the
news media community until roughly 2000 when the Guardian launched a blog on its
own Web site (Matheson, 2004).
Change in the industry has further been aided by a shift in consumer
consumption of media. Today, roughly 65 percent of adults age 35 or older read a
daily newspaper, but only 30 percent of adult age 34 or younger read a daily
newspaper. Across demographics, these statistics have declined steadily since 1999.
When asked about primary news sources, 34 percent of Americans report that
newspapers are their first choice, while 39 percent of Americans report that Web
sites are their first choice (Mitchell & Rosenstiel, 2009). Consumers are shifting
93
news consumption to online sources, and spending a decreasing amount of time with
printed products.
For the newspaper industry, however, existing organizations failed to
accurately anticipate the emergence of a new competitive organizational form.
Traditional newspapers experimented with online forms of news as early as the mid-
1980s, but innovation has tended to be focused on short-term profit rather than long-
term change (Mitchelstein & Boczkowski, 2009). Traditional news organizations,
however, were slow to fully embrace radically new forms of online news in general
because they were not perceived to be a significant competitive threat. In describing
Chicago Tribune’s early approach to blogging, one editor noted:
We were really reluctant to engage with the blogosphere early on. By
2004, 2005, we knew that would have to change, but even then blogs
were not a primary part of our strategy. We were just foggy on them
and what they could do for us. (Anonymous, 2009a)
At The San Diego Union Tribune, newsroom staff took a similar approach to blogs.
“We had blogs fairly early on, starting in 2002” explained one editor, “But we
weren’t sure what to do with them. It just wasn’t part of the news business. In recent
years, however, it’s been completely different.”
The slow process of change is reflected in the employment levels in the
newspaper publishing industry. Figure 6 compares the number of employees in
traditional news to the number of employees working in positions that were
classified as online news. Through to present day, employment levels in online news
remain a fraction of employment in the traditional news sector.
94
Figure 6: Number of employees in traditional and online news
0
50
100
150
200
250
300
350
400
450
500
1947
1950
1953
1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007
Year
Number of Employees (000)
Newspaper Publishing Online Newspaper Publishing
Source: Bureau of Labor Statistics, 2010
There was however, a significant increase in the number of employees working in
online news from 1999 through 2001. The increase corresponding with an
acceleration in the decline of employment in traditional news. The increase in online
news employees indicates that as early as 2000, there were strong indications of an
emergent, legitimate online news industry.
Therefore, this chapter concludes by returning to the issue of legitimacy. As
was noted, density dependence is problematic as a pragmatic measure of legitimacy
because it is retrospective. In addition to the proposed hypotheses, this research will
additionally examine the timing of legitimacy measures. The following research
95
question thus explores which measures of legitimacy are most likely to detect the
early emergence of a stable, legitimate organizational form:
RQ
2
: Are there certain legitimacy measures that are more likely than others to
indicate the legitimacy of organizational forms?
Early detection of new organizational forms is particularly critical in the context of
online technology. Given the ease with which online organizations can emerge and
fade away, early indicators of legitimacy are critical for established organizations
seeking to respond in response to new competition.
96
CHAPTER 4: TRANSFORMATION IN RESPONSE – THE EVOLUTION OF
TRADITIONAL NEWS
Chapters 3 and 4 explored the theoretical processes of form emergence and
legitimacy, and illustrated the development of social networks and participatory
media as a new population within the online news media community. When new
forms emerge and establish legitimacy, existing organizations face the challenge of
transforming in response. This chapter examines the process by which established
populations respond when a community enters into a period of transformation.
Transformation in Response to the Emergence of New Forms
Parallel to the process of form emergence, existing populations must adapt to
new competition as emergent forms reach sufficient density such that they are
established as legitimate organizational forms. When a new population emerges with
sufficient proximity to existing populations, there is a clear overlapping of resources
and increased competition (Aldrich & Ruef, 2006; Barnett & Carroll, 1987; Brittain
& Wholey, 1988). In the case of speciation, this process can occur rapidly when a
new form has reached a sufficiently critical mass such that the new population
reenters into competition with other populations. New populations, and new
organizational forms, do not necessarily supplant existing forms. Rather a period of
competition is likely to ensue during which the community as a whole undergoes a
period of transformation before a new equilibrium is reached (Aldrich & Ruef, 2006;
Bryant & Monge, 2008).
97
Print newspapers anticipated the growth of online news long before Web 2.0
technology began to dominate the online landscape. Newspapers had experimented
with videotex technology in the 1980s as potential electronic delivery system for
news, but most videotex programs were miserable failures. In an extensive study of
print newspapers online struggles, Boczkowski (2004a, p. 71) observed,
“[newspapers] have tried to transform a delivery vehicle that has remained unaltered
for centuries, and whose permanence has anchored a complex ecology of
information symbols, artifacts, and practices, while simultaneously aiming to leave
the core of what they do, and are, untouched.” Early on, Internet technology was
viewed as a new innovation that would simply assist reporters in their day-to-day
work. A Nieman Reports article published in 1993 describes the Internet as an
invaluable innovation for information gathering, noting:
No doubt there were journalists who complained bitterly when they were
asked to stop using pens and start using typewriters. It’s not hard to
remember when modern-day journalists complained about the change from
typewriters to computers. Yet in each case, after a few months on the new
system, most reporters were unwilling to do without the new technology. I
predict the same thing will happen with the Internet. (Regan, 1993a, p. 27)
In this way, many journalists saw the Internet as a new form of the encyclopedia. It
was viewed as a tool that would aid in reporting, but the existing traditions of
reporting were expected to remain the same.
Print newspapers thus responded over a period of years with measured
innovation and experimentation. While Internet technology provided consumers with
nearly ubiquitous access to a world wide network of news and information,
98
newspaper companies continued to function on a daily cycle of news production
(Pavlik, 1999). Despite a rapid growth in the number of online news sites, the print
newspaper population remained largely paralyzed through the 1990s; the form of
news did not change dramatically. Journalists realized that technology was allowing
for new forms of news production, but remained unclear about how new technology
would change the distribution and form of news stories (Barnhurt, 2002). As a
medium for communication, the Web has inherent properties that distinguish it from
television or print newspapers. Web sites allow for instant interactivity, a much
shorter lifecycle for updating information and also allow for a rich mix of media
including text, photos, videos and interactive features. In addition, the Web has a
capacity for a wide range of data and flexibility to be scaled on an as-needed basis
(Stovall, 2004).
Newspaper companies were aware that new technology was important in the
media landscapes, and industry leaders did not turn a blind eye. A Christian Science
Monitor reporter wrote in 1993 that, “Weblogs have risen to near the top of the
media’s collective consciousness. There are several reasons for this awakening:
More well-known journalists have started blogs in the past year, and Web sites of
well-known media brands have created blogs to help cover news, politics and other
issues (Regan, 1993b, p. 63).” Print newspapers, however, were not convinced that
traditional newspapers needed to revolutionize, or even evolve. A trade article in
2000 observed, “There is a very important fact that all journalists must bear in mind
99
– our future does not lie on the Web. In fact, if you believe some people, we should
just forget about the Web altogether because its time as a distribution method is
almost past… the Web will be just one of many ways that we will get the news to
those who want it (Regan, 2000, p. 7).” As Boczkowski explains, the 1990s were a
period of significant online development for newspapers, but that development was
focused on the development of products and services that would protect the market
share of business areas that were considered at risk (Boczkowski, Forthcoming).
From 2000 onward, the news media community began to change more
dramatically than in the previous decade. From 2000 to 2010, participatory media
exploded, blogging became a mainstream form of information production, and in the
latter half of the decade social networking established itself as a dominant platform
for online communication. Early in the decade blogs and participatory media began
to develop as competing forms for news media, but they did not immediately emerge
as a clear replacement for traditional newspapers. In the face of increased
competition, however, print newspaper companies redoubled efforts to grow their
online news presence. Print newspapers have competed heavily against blogs,
online-only news Web sites and services, and social networking sites for a share of
audiences’ time, and for an increased share of online advertising revenue
(particularly local advertising) (Althaus & Tewksbury, 2000; d'Haenens, Jankowski,
& Heuvelman, 2004; Dimmick, Chen, & Li, 2004; Wise, Bolls, & Schaefer, 2008).
Users’ time spent online is a critical success factor for news Web sites; time spent is
100
an indicator of how engaged a user is with the content on a given Web site, and also
helps to increase the price that an organization is able to charge its online advertisers
(Arends, 2009). In 2000 it was clear that online news was becoming an important
medium, but by 2004 it was evident that print newspapers were going to need to
make significant changes in order to be able to compete against online newspapers
(Dimmick et al., 2004).
Hand-in-hand with changing competition, there was a clear paradigm shift in
the consumption of media. In 1995, less than 5 percent of the US population were
using online news, but by 2002 that number had grown to 35 percent (Greer &
Mensing, 2006). As consumers shifted consumption patterns to include Internet
based news, an increasing amount of time was being spent with digital information.
Conversely, since 2000 the time a typical adult spends with a newspaper fell from
201 hours per year to 175 hours in 2007. At the same time, the average hours a
typical adult spends with the Internet increased from 104 hours in 2000 to 195 hours
in 2007 (Pew Research, 2007). This suggests a clear shift in resources, and in
response, there is strong evidence that news media organizations have undertaken a
wide array of initiatives in order to transform their core organizational forms. This
includes initiatives to introduce user-generated content into traditional newsrooms,
alliances and partnerships with blogs and user-generated content sites, and a
complete realignment of organizational goals and imperatives (Boczkowski, 2004a).
In a 2008 survey, 72 percent of Americans reported having gone online for news,
101
while 40 percent responded that they had gone online for news within the past
twenty-four hours (Internet Overtakes Newspapers as News Outlet, 2008).
In the midst of industry wide change, one particular incident in 2002
highlighted the growing importance of online news vehicles. At a birthday party for
Sen. Strom Thurmond (R-SC), Sen. Trent Lott (R-MS) gave a speech to friends
gathered in the Senate offices in Washington, DC. During the speech, which was
televised and covered in the press, Lott said, “When Strom Thurmond ran for
president, we voted for him. We’re proud of it. And if the rest of the country had
followed our lead, we wouldn’t have had all these problems over all these years
either.” The quote was overlooked by many, but within a few days the story was
reported on political blogs Atrios.blogspot, Instapundit and Talkingpointmemo. Soon
after, dozens of other blogs retransmitted the story and it quickly began to emerge as
a major political story. The news was first reported in the blogosphere on December
6, 2002; it was December 10, five days after they birthday party, before a newspaper,
The Washington Post, covered the outcry surrounding the event. Following a sharp
increased in press coverage, and a rebuke by President George Bush, Lott announced
on December 12 that he would be stepping down from his post as Senate Majority
Leader (Scott, 2004). The story changed the perception of bloggers as journalists for
many in the newspaper industry. For many years, journalists had tended to regard
bloggers “as a sort of mutant breed, viewing them with skepticism and suspicion. In
the eyes of many journalists, blogs are poorly written, self-absorbed, hyper-
102
opinionated, and done by amateurs (Regan, 1993b, p. 63).” The Lott Incident showed
that blogs could be as powerful as printed newspaper articles in driving change and
uncovering scandal. According to J.D. Lasica, senior editor of the Online Journalism
Review in 2003:
Weblogs could usher in a refreshing new openness in newsrooms by
attaching a face and personality to reporters. Blogs could show that
newspapers aren’t monolithic corporations but a collaborative team of
individuals with varying viewpoints and who have more in common with
their readers than they could possibly know from reading their print articles
alone (2003, p. 72).
2003 was a turning point for participatory media, and the newspaper industry as a
whole began to strategically respond to the large-scale emergence of participatory
journalism (Bowman & Willis, 2003).
For those working at traditional newspapers, the push to integrate
participatory and social media into existing organizational structures proved a
complicated task. “There was a recognition that we couldn’t do it all ourselves, and
that distribution of our content needed to go beyond our own site,” explained a
manager at The Washington Post, “We knew in 2002 that it was going to be hard as
standalone newspaper site to get to a large scale unless we could join with larger
networks” (Anonymous, 2009b). Blogs, social networking sites and online-only
news sites during this period were growing at a scale that far exceeded the growth of
most newspaper Web sites. Newspapers, in turn, began to form partnerships with
blogs and other online content producers. Early on, one of the key differences
between blogs and Web sites was the amount of content that each could aggregate.
103
Newspapers were generally conservative in terms of linking to other Web sites and
aggregating content. Blogs, on the other hand, aggregated content from hundreds of
other Web sites, and readily created links to other content producers. As one editor at
The Las Vegas Sun observed:
When you’re a newspaper website, there are kind of two ways you can do it
at your website. You can either be on the Internet or you can be of the
Internet, and we practice very much being of the Internet. If you weren’t a
newspaper, you wouldn’t even think twice about linking to someone else.
Bloggers – The Huffington Post - Gawker – they never even worry about it.
(Anonymous, 2009c)
When newspapers began to rethink online strategies, linking became a central issue.
Newspapers have traditionally functioned as gatekeepers, filtering news and
essentially crafted the public perception of societal events (White, 1949).
Aggregating links, however, shifts more of the gatekeeper responsibility to the
consumer. Newspapers struggled with this transition, but changing environmental
conditions forced many organizations to take dramatic steps to reinvent themselves.
In the face of community-wide upheaval, the print industry as a whole thus
faced an uphill struggle. Forecasts indicate that online newspapers would need to
grow 33 percent a year for the next 10 to 12 years in order to match the revenue of
print counterparts (IBIS, 2008). Thus there was a clear disconnect between the
decline of print revenue and the growth of online revenue, despite the ability of
online newspapers to reach large audiences. During this time, a shift to public
ownership of media companies added increased pressure; in 1945 eighty percent of
American daily newspapers were privately owned, whereas by 2000 about 80
104
percent of newspapers were owned by publicly traded companies (Klinenberg,
2005). The pressures of public ownership place an added burden of responsibility to
owners on the management, which has propelled a search for new sources of
revenue. In response to a host of factors, including increase competition from digital
content, publicly traded newspaper companies stock declined roughly 20 percent in
value in 2005, followed by another 14 percent drop in 2006 (IBIS, 2008). On a large
scale, the industry did experienced a significant shift in employment numbers as
gains in online staffing have offset losses in print staffing. Only 1,000 net employees
lost their jobs industry wide, although at the Dallas Morning News in Dallas, Texas,
for example, 110 newsroom employees were laid off, while 30 online employees
were hired, somewhat obfuscating the full picture. But it was clear that new
initiatives were needed in order to compete in an online news environment.
As previously discussed, one of the key challenges in identifying a new
organizational form is the process of determining the nature of the form itself
(McKendrick & Carroll, 2001). The above discussion, however, has identified a
number of key characteristics as central to the online news form: in particular, online
news provides immediate information, is easily updated, is interactive and allows for
user participation (Abdulla, Garrison, Salwen, Driscoll, & Casey, 2002; Greer &
Mensing, 2006; Mitchelstein & Boczkowski, 2009; Wise et al., 2008). Table 1
compares and contrasts the online news form that has been discussed in this chapter
to the traditional print newspaper, highlighting the ways in which this new
105
organizational form represents a new form separate from existing modes of news
production.
Table 1: The Online News Form Compared to Print Newspapers
Online News Form Print News Form
Product
Content
News and commentary
(generally more
editorialization)
News and commentary
Publication Cycle
Immediate (often
update hourly)
Generally daily
Media
Online-based (but may
include video, photos,
text)
Print-based (photos
and text)
Cost to User
Generally free
(although some
subscription)
Subscription of per-
use fee
Barriers to Entry
At a basic level, none;
major sites have
higher associated costs
High (printing press,
production staff,
overhead)
Production
(employees)
Size of
Organizations
Minimal (as small as
one)
Major newspapers
employ thousands;
smaller papers employ
as few as 4 or 5
Accreditation
Coverage of major
events requires
accredidation
Coverage of major
events requires
accredidation
Training
None required (college
education is
beneficial)
None required (college
education is
beneficial)
Employee pay Generally low Generally low
As the above table illustrates, online news and print news share certain
characteristics such as the basic nature of content, as well as certain attributes of
106
employees associated with the organizational form. For the most part, however, there
are clear differences between the two organizational forms. Newspapers have much
higher costs of production, publish at a lower frequency, and focus more heavily on
daily news. Online news, however, publishes on a much more frequent cycle,
generally emphasizes commentary, and has a lower cost structure overall.
Drivers of Transformation
Newspapers spent much of the turn of the century seeking new strategies for
producing and distributing news in an online news environment. At its core,
population ecology has treated organizations as inert entities that are unable to adapt
or reorganize organizational goals in the face of new technology and new
competition (Hannan & Freeman, 1984). Organizational ecology has often
overlooked the causes and effects of changes in organizational goals, as well as
changes in authority and technology, on the survival of organizations (Barnett &
Carroll, 1995). McKelvey and Baum urged, in turn, that there is a growing need to
explore multilevel mechanisms whereby micro-level change can be explained within
the context of the macro-level environment (McKelvey & Baum, 1999). From the
community ecology perspective, organizations facing disruptions or new competition
must decide whether or not to transform in an attempt to survive (Astley, 1985). A
key challenge in this process is to distinguish a transformation from a mere structural
adjustment (Carroll & Hannan, 2000). Aldrich, distinguishing this process from
change, defines transformation as “a major change occurring along three possible
107
dimensions: changes in goals, boundaries, and activities. To qualify as
transformations, changes must involve a qualitative break with routines and a shift to
new kinds of competencies that challenge existing organizational knowledge”
(Aldrich, 1999, p. 163). Simply put, an organizational transformation involves a
major shift in an organization between two points in time. This can be decomposed
into the process of change and the content of change (Carroll & Hannan, 2000). In
essence, Carroll and Hannan (2000) prescribe that any study of organizational
transformation must address how the change occurs (process) and the nature of the
change between two points in time (content). Both the frequency with which major
transformations occur and the general impact that transformations have (Carroll &
Hannan, 2000) on the lifespan of a firm remain controversial in the study of
organizations (Barnett & Carroll, 1995).
At the community level, change is driven by the introduction of disruptions
(Tushman & Romanelli, 1985a) and the introduction of speciated forms, as was
discussed in the previous chapter. Change is difficult during periods of stability
because existing routines and structures create resistance both within organizations
and across populations. In a study of the microcomputer industry, for example,
marked changes such as a rapid decrease in available venture capital in 1971, or a
change in chip standards from 16-bit to 32-bit, sparked fundamental transformations
within populations of chip-producers, as well as populations of suppliers and retails
(Romanelli & Tushman, 1994). Thus significant environmental change has
108
ramifications across populations and is a driver of change. Alternatively, in an
examination of the California Savings & Loan industry, policy changes in the 1970s
created new markets, rendering previous population boundaries obsolete almost
overnight. The Financial Institutions Regulatory and Interest Rate Act of 1978, for
example, allowed federal thrift banks to expand into new investment markets.
(Haveman, 1992). Thus, technological change, macro-economic and regulatory
changes can also cause punctuated shifts in the resource environment.
In response to the need for transformation, Levinthal (1998) suggested that
strategic actions taken by individual organizations cannot drive changes to the
organizational core (that is, the core features of the organizational form). But this
perspective is not universally held. From an economic perspective, there is a strong
argument made for the role of agency in guiding the development of an organization
(Zajac & Westphal, 2007). From the perspective of organizational ecology, the
foundational texts suggest that there is a place for agency in the development of
forms. There is an argument for agency even in the writings of Darwin: “The key is
man’s power of accumulative selection: nature gives successive variations: man adds
them up in certain directions useful to him. In this sense he may be said to make for
himself useful breeds” (2003, p. 90). Managers drive adaptive variation in response
to environmental pressure, and to effect internal change, but this often occurs
strategically in an attempt to create value-added innovation (Miner, 1994). This is
supported by Burgelman (1983), who studied the creation of routines in capital
109
venture firms and found that routinization of new innovations occurs as the result of
active strategic decisions on the part of managers. In a study of multiple industries,
researchers looked at Japanese and South Korean enterprises and found that
organizations that diversified into new areas of enterprise could increase
profitability. Likewise, the same study examined mutual funds were able to change
objectives in order to create new funds and drive increases in long term profitability
(Amurgey & Rao, 1996). Such studies, however, have generally looked at actions by
individual organizations without considering the effects the cumulative microlevel
changes can have on populations and communities.
The notion that organizations can adapt has often been framed as an issue of
organizational learning: by strategically adapting to changes in the environment, an
organization improves its long-term likelihood of surviving Darwinian selection
(Bruderer & Singh, 1996). Health care organizations, for instance, have been shown
to evolve through a process of recombination: existing organizational and internal
resources were essentially realigned to develop new organizational forms (Van de
Ven & Garud, 1994) such as health maintenance organizations. Bruderer and Singh
(1996) model this learning process can occur as an explicit recombination of existing
organizational forms. In other words, existing organizations can adapt to changing
environmental landscapes by learning and merging forms with other organizations.
In an examination of the transformation of gas stations and service stations, Usher
and Evans (1996) found strong support for Darwinian processes of evolution: that a
110
primary mechanism of transformation was the replacement of existing organizational
forms by the emergence and success of new organizational forms. Their study also
found strong evidence that Lamarckian evolution played a critical role in the
transformation process. Lamarckian processes occur when new competencies are
adopted by existing organizations as a mechanism of transformation in order to
increase the likelihood of survival. In Usher and Evans’ examination, adaptation
through Lamarckian processes occurred through a recombination of existing
organizational forms with emergent competencies. Within organizations, this type of
transformation occurs through an ongoing process of information exchange and
learning (Bacharach, Bamberger, & Sonnenstuhl, 1996). It is not, however, entirely
clear how this process of recombination and learning occurs between organizations.
One potential explanation is that organizations may mimic behaviour seen in other
organizations as a mechanism for learning (Suchman, 1995).
There is strong evidence, however, that the development of new internal
strategies can lead organizations to more effectively learn and transform in the face
of disruption (Bergelman, 1991). Transformation within the organization thus has the
potential to be considered a strategic process that can be extended beyond the
organizational level. In addition, changes to the core organizational structure,
mission, and technology or marketing strategy create a significant disruption and
essentially reinvents the organization. Indeed, there is some theoretical support that
strategic change can overcome the introduction of new technological competencies
111
(Levinthal, 1998; Suarez & Oliva, 2005). This discussion suggests that organizations
have the potential to use strategic change as a tool for survival. In the case of
newspapers, for instance, established newspaper companies faced a marked increase
in competition in the period from 2000 to 2010. In turn, there was a clear need for
newspaper companies to overcome existing inertia and adapt to changing conditions.
For individual organizations, a conscientious undertaking of inter-organizational
change initiatives can mitigate the risks associated with the emergence of competing
organizational forms. This is hypothesized as follows:
H
6
: During periods of transformation, organizations that actively
undertake strategic change initiatives will have a decreased likelihood
of failure.
This mechanism is predicated on the actions of individual organizations, but
ultimately is a population-level mechanism as it is dependent on the interaction that
occurs between organizations. Following Usher and Evans’ (1996, p. 1435)
suggestion, the above blends Lamarckian and Darwinian processes: an organization
may choose to undertake a succession of strategic initiatives to reduce its risk of
failure (Darwinian evolution for the specific unit that changes), but the overall
success or failure is dependent on a Lamarckian process of adaptation over time.
When a transformational process is undertaken certain barriers must be overcome.
Three moderating factors – the organizational structure, inertia of an organizational
unit, and existing routines – all restrict or moderate the potential for successful
112
transformation. Existing structures represent the preexisting linkages that exist
between populations, as well as organizations. Between populations, symbiotic and
commensalist bonds exist that can potentially create buffers against environmental-
level changes (Astley, 1985). In addition, interorganizational, institutional linkages
may exist between organizations and populations. Linkages have a number of effects
on organizations. By linking to established, institutionalized organizations, new
organizations may gain the benefit of conferred legitimacy. In addition, linkages may
make transformational processes less risky by providing support and resources for
stability during the transformation (Baum & Oliver, 1991; Miner, Amurgey, &
Stearns, 1990). Fundamental, revolutionary shocks to organizational environments
create enough momentum for change that existing structures and systems will evolve
(Romanelli & Tushman, 1994). Romanelli and Tushman’s analysis found little
evidence that gradual changes would overcome existing structures.
Regardless, existing structural inertia must be overcome. Structural inertia
refers to the forces that must be overcome in order for organizations to change.
Inertia is relative to changes in the environment; organizations that are able to
respond quickly to changes in the environment have low inertia, whereas
organizations that respond slowly have high inertia because the momentum and the
ability to learn and adapt to change means that they will respond slowly (Hannan &
Freeman, 1984). Inertia is moderated by a number of factors such as size and age.
Older organizations tend to have greater structural inertia, as do larger organizations,
113
and this corresponds with lower death rates. Transformations disrupt organizational
inertia and lead to increases in the variability of organizational performance. The
population of newspapers in the United States is, in general, plagued by issues of
structural inertia (Carroll & Delacroix, 1982a; Carroll & Hannan, 1989; Dejean et
al., 2004). Furthermore, a number of ecological studies have shown that structural
inertia is particularly problematic for newspaper organizations during periods of
transformation (Carroll & Delacroix, 1982a, 1983). Levinthal (1998) speculated that
organizations struggle to transform against inertia because most change is gradual. It
follows that aggressive strategies of organizational change may lead to a decreased
failure rate in the long run. In the case of newspaper companies, forinstance,
newspapers that adapted to online news during early stages of form emergence were
better positioned to be a dominant online news producer in the long run. The effect
of aggressive organizational strategies is thus hypothesized as follows:
H
7
: During periods of transformation, the adoption of aggressive strategic
initiatives will decrease mortality rates for inertial organizations.
Together, hypotheses six and seven introduce the notion of strategic change as a
process for organizational transformation. Older organizations face an uphill battle in
adapting, but aggressive strategies of transformation can lead to long-term success.
Hannan and Freeman (1984) conclude that changes may reduce the risk of failure in
the long run by aligning organizations with their environments, but in this short run
organizations becomes maladjusted and prone to failure.
114
Hyperlinking as a Strategic Tool
Strategic change can thus lead to transformation, but the mechanism of
change remains unclear. The following section explores the process of change and
adaptation, examining how organizations adapt to the introduction of new
organizations and change in response. The formation of partnerships and
interorganizational linkages is one key mechanism through which an existing
organization, or population of organizations, can facilitate transformation. Tan and
Tan (2005), for instance, found strong evidence indicating that organizations that
change strategies in the face of changing environmental conditions have a higher
likelihood of survival. In addition, the most successful strategies involved strategic
risk-taking through partnerships and alliances (Tan & Tan, 2005). Previous studies
have clearly established that linkages provide economic and reputational benefits, as
well as increased legitimacy (Ruef & Scott, 1998). As Baum and Oliver (1991) note,
organizational ecology has often overlooked the importance of interorganizational
linkages on organizational survival, despite the previously discussed focus on
legitimacy. Their study found that organizations with institutional linkages had an
increased likelihood of long-term survival. Institutional attachments are linkages that
give organizations advantages through increase stability, access to resources and
social support (DiMaggio & Powell, 1983; Meyer & Rowan, 1977). In other words,
organizations that establish ties to established institutions within a population are
likely to have an increased chance of survival. Baum and Oliver’s study found this to
115
hold in Canadian childcare organizations and found evidence from the ecological
perspective that linkages can increase the likelihood of successful transformation.
Interorganizational ties can thus serve two critical purposes. First, ties can confer
legitimacy on new organizational forms, and second, established organizations can
create ties to new organizational forms as a mechanism for learning, with the goal of
increasing survival.
Interorganizational linkages occur in many shapes and forms. Linkages can
be established through formal relationships, established through contracts or formal
agreements, or informal partnerships or acknowledgements. For online
organizations, however, linkages do not necessarily occur through traditional
mechanisms. Thus, it is posited here that hyperlinks are a primary representation of
links between online forms of organizations. A number of prior studies have looked
at hyperlink patterns to understand organizational structure, and have established the
validity of hyperlinks as a measure of organizational linkages. For instance, Halavais
(2008) and Park (2003) propose that hyperlinks are a proxy for organizational
linkages in the online context. Furthermore, for online-based organizations
hyperlinks are the primary tool for formally establishing a public relationship
between two organizations (Halavais, 2008). Other ties may exist (informal,
information-sharing), but the formation of strong, established hyperlinks clearly
denotes that two organizations are connected in an online setting. That said,
however, analysts must be wary of false organizational ties resulting from haphazard
116
linking; this is often resolved by looking for ties that have a strength (or frequency)
beyond what is likely to occur randomly.
In the case of the news media community, printed newspapers have operated
as social institutions for centuries, but online newspapers emerged recently, rapidly,
and rely of large networks of informal information partnerships. Blogs and social
networking sites, in particular, are reliant on large networks of links as a mechanism
for information gathering and distribution. For newspapers, however, this mode of
operation is contrary to decades of strategy. Online newspapers use hyperlinks to
establish relationships with other news sources, content providers and partners.
These links serve as a key strategic source for partnerships and the movement of
traffic online (Halavais, 2008). As print newspapers have moved into the online
space, they have worked to adapt to this online environment, and as a result there is a
significant variance in the strategies the newspapers have adopted. While some
companies have heeded the advice of Jarvis and others, many have continued to
resist transformation. Newspapers have experimented with different patterns of
online linking (Dysart, 2002; Tremayne, 2006). Returning to the concept of
institutional linkages and embeddedness, it follows that organizations that
proactively adopt a strategy of linking will increase their likelihood of survival
(Baum & Oliver, 1991). In other words:
H
8
: The higher the number of institutional hyperlinks maintained by online
newspapers, the lower their mortality rate.
117
Communities and Populations of Embedded Hyperlinks
In addition to functioning as a tool for increasing organizational survival,
hyperlinks also represent structure above and beyond individual organizations.
Hyperlinks serve as a proxy of the general industry structure through the patterns of
links that exist among organizations. In traditional offline spaces this is seen through
strategic partnerships, alliances and mergers among media organizations, whereas in
an online space structures are the patterns created by links among organizations, as
many online newspapers are representations of offline organizations. Others have
examined the general online structure of digital news companies, but this has
primarily been through case studies (Chon, Choi, Barnett, Danowski, & Joo, 2003;
Park, 2003; Park, Barnett, & Nam, 2002; Park & Thelwall, 2008). Hyperlinks as ties
further demonstrate that communities of organizations are tightly interconnected;
hyperlink networks can be used as a map to understand the relationships between
organizations within these communities (Ackland, O'Neil, Bimber, Gibson, & Ward,
2006; Wellman, 1997).
For example, this has been applied in the context of NGO organizations
wherein a number of studies have examined the structure created by hyperlink
relationships. In particular, Shumate and Dewitt (2008) show that online structures
can be used to determine affiliations and alliances between NGOs in the not-for-
profit environment. Furthermore, Shumate and Lipp (2008) explored the motivations
behind linking, and found that more so than preferential attachment, NGOs link
118
based on collective action towards building community around common issues.
Furthermore, while hyperlinks are representative of the structure that exists between
organizations online, there is strategic behavior embedded in linking (2004). Few
studies have extended this research to news organizations. One notable exception is
Tremayne’s (2004) analysis of news organization, which found that over time news
companies are more likely to link more prolifically. This study, however, failed to
control for either the overall growth of the Web or the variation in type and quality
of hyperlinks.
Given that hyperlinks represent online organizational structure, this analysis
therefore shows the effect that symbiotic and commensalist ties have on
organizational survival over time. Revisiting earlier definitions, symbiotic
relationships describe situations where “individuals complement one another in the
performance of their respective assignments; they enter into mutual dependences
based on their functional differences” (Hawley, 1986). Commensalist relationships,
on the other hand, occur when populations of similar forms engage in “coactions,”
and potentially face a situation of competition for common resources (Rao, 2007).
There is ultimately a broad spectrum of linking strategies that can be chosen by
organizations; these range from prolific linking and sharing of content through to a
complete refusal to link and distribute content beyond the boundaries of a single
organization’s domain. This is akin to the previously discussed concept of symbiosis
and commensalism. As organizations face increased competition, organizations with
119
a strong network of symbiotic relationships will have a greater likelihood of adapting
to new market conditions as a result of shared resources and strategies. Symbiotic
ties are particularly critical when new forms emerge, as these ties establish
legitimacy and allow existing organizations to learn from successful entrants (Rao,
2007). For instance, as previously noted blogs represent a particularly successful
type of online news source; blogs have typically linked prolifically to a diverse array
of organizations and draw on a wide variety of partners and information sources
(Park & Jankowski, 2008; Rainie & Horrigan, 2005; Reese et al., 2007). Traditional
news organizations that adapt and link to new entrants such as blogs will in turn
access a diverse network of information resources. As such, by establishing
hyperlink ties early on to other emerging populations an organization will position
itself, via a strategy of symbiotic ties, to take advantage of new resources and new
technology; over time this should result in continued growth in partnerships:
H
9A
: Established organizations that have a symbiotic strategy with a high
ratio of linkages to new populations during periods of emergence will have a
high proportion of outlinks to new populations over time.
On the other hand, commensalist ties insulate an organization within its own
population. As noted by Rao (2007), organizations with primarily commensalist ties
will ultimately compete against similar types of organizations for resources. In terms
of hyperlinking, a strategy of commensalist ties is seen in narrow hyperlink
120
portfolios; this type of behavior will translate to a decrease in prominence over time
and the formation of fewer and fewer ties.
H
9B
: Established organizations with a low ratio of linkages to new
populations during periods of emergence will have a low proportion of
outlinks to new populations over time.
Thus, organizations that establish symbiotic patterns of relationships with new
populations continue to form partnerships over time. On the other hand, this makes a
given organization a more attractive partner for new entrants. This echoes the notion
of preferential attachment theory in online networks (Halavais, 2008), Web sites that
link out over time are more likely to receive reciprocal links as well (Arms, Aya et
al., 2006; Arms, Huttenlocher, Kleinberg, Macy, & Strang, 2006; Murphy, Hashim,
& O'Connor, 2008; Vaughn & Thelwall, 2003a), Thus it follows organizations with
symbiotic patterns of linking will be more attractive partners to other organizations :
H
10A
: Established organizations that have a symbiotic strategy with a high
ratio of linkages to new populations during periods of emergence will have a
high proportion of inlinks to new populations over time.
Alternatively, those Web sites that have a narrow link portfolio are again likely to
have a narrow range of growth over time. Returning to organizational theory,
Romanelli and Tushman’s (1994) analysis found little evidence that gradual changes
would overcome existing structures. Therefore, those organizations with a narrow
link portfolio will likely continue to have a narrow range of partnerships over time:
121
H
10B
: Established organizations with a low ratio of linkages to new
populations during periods of emergence will have a low proportion of
inlinks to new populations over time.
In aggregate, these hypothesis test the notion of link economy, examining the ability
of different types of organizational hyperlinks on prominence over time. These
hypotheses are tested in the following sections against data tracking the development
of the online news industry in order to understand the effect that commensalist and
symbiotic hyperlinking relationships have on organizational structure and position
over time.
The Transformation of News: An Ecological Perspective
The previous discussion presents a new view of organizational strategy. In
the case of newspapers, hyperlinking can be viewed as critical strategic mechanisms
for aligning existing organizations in competition or cooperation within emergent
organizational forms. In aggregate, the previous three chapters have thus outlined a
framework for examining the transformation of the news media community in the
digital era. First, Chapter 2 explained the process by which new organizational forms
emerged rapidly over a ten-year period, developing into new populations of online
news production. Previous theories postulated that organizational forms developed as
the result of punctuated equilibria; speciation was explicated as a alternative
explanation for the rapid emergence of new organization forms. As new
organizational forms emerged, processes of variation, selection and retention drove
122
suboptimal forms to obsolescence, while successful forms replicated. Over time,
emergent organizational forms will gain legitimacy through ties to established
populations, and through traditional mechanisms of media, legal and professional
recognition, as explained in Chapter 3. From the community perspective, the
mechanisms explained in these chapters do not occur in isolation. Instead, these
mechanisms interact with one another and collectively contribute to the ecological
process of community transformation. In other words, organizational-, population-
and community-level processes coevolve as the community transforms over time.
This intertwined process is illustrated in Figure 7:
The transformation process is initiated by changes in environmental level
resources. The first research question (RQ
1
) tests the applicability of the staged
evolution model (emergence – maintenance – self-sufficiency – transformation) as a
framework for examining the transformation process of the news media community.
Following the introduction of new technology into the news media community, new
organizational forms began to develop in relative isolation. These speciated
organizational forms developed within the broader community, but within niches of
available resources. As a result, during emergent stages, new organizational forms
tended to primarily connect with their own type (H
1
). Speciation, however, is not an
isolated process. Although organizational forms tend to develop in relative isolation,
there is a minimal amount of contact with existing populations (H
2
). Once the new
organizational form has begun to replicate towards a critical density, the new
123
Figure 7: Community Transformation as a Multilevel Process
1
2
3
4
5
6
Environment
Community
Populations competing in the online news media
environment (print newspapers, blogs, social
networking sites, television states, radio stations)
Population
Key Variables: number of organizations in each
population, birth and death rate of organizations, links
to other organizations, links to other populations,
density
Organization
Key Variables: Age, ownership, leadership,
organizational strategy, outlinks, inlinks, centrality
Transformation Process
1. Environmental level disruption: technological innovation
2. Speciation: new organizational form emerges with limited contact
3. Legitimation: form grows in density, acknowledgement by others
4. Transformation: existing organizations develop new strategies
5. Transformation: commensalist and symbiotic ties enable change
6. Reinvention: community attempts to return to equilibrium
124
population will reenter into competition with the rest of the community (H
3
). In this
way, speciation helps to explain the rapid disruption that occurred in the online news
media community when blogs and participatory media entered into the community as
a significant source of online news and information.
Following the process of form emergence, the community as a whole moves
towards a period of maintenance as existing populations compete for resources
against new entrants. In order to survive, however, new organizational forms must
achieve legitimacy within the community – achieving recognition and legitimacy
through media coverage (H
4A
), professional organizations (H
4B
) or legal rulings
(H
4C
). In addition, the creation of linkages between existing populations and new
organizations can additionally confer legitimacy on new forms (H
5
). The population
as a whole will move towards stability, and new forms will move towards
legitimacy, as the emergent population increases in age. Thus, as the new
organizations’ population ages, the death rate of new organizations will decrease
(H
6
). Organizational studies to date generally take a retrospective perspective on
studies of legitimacy. In order to better understand the process of form emergence,
and better measure the emergence of new organizational populations, the second
research question examines whether some measures indicate legitimacy earlier than
others (RQ
2
). Online organizations, in particular, can emerge rapidly and itis
increasingly critical for established organizations to better understand when a new
form represents a legitimate competitive threat.
125
In the face of new technology and increased competition from the
development of new populations of organizations, existing organizations must either
adapt and transform, or fade away into obsolescence. For older organizations,
however, change is particularly troublesome due to existing organizational inertia.
Older organizations are, in general, less likely to undertake change initiatives (H
7A
),
whereas newer organizations generally have less inertia and as a result will be more
likely to attempt change (H
7B
). In general, interorganizational linkages are key
mechanisms for learning and gaining access to new resources. Interorganizational
linkages will substantially reduce the effect of new competition, lowering mortality
rates (H
8
). More important, in an online environment existing organizations can use
hyperlinks as tools to create relationships and establish interorganizational linkages.
Organizations that adapt an aggressive linking strategy during emergent periods, will
be more likely to establish links over time (H
9A
). Those organizations that do not link
early on will struggle over time and form fewer links (H
9B
). In turn, organizations
that actively link to others will be more likely to be linked to (H
10A
); organizations
that do not link will tend to remain isolated and will generally not receive links
(H
10B
). In aggregate, these two hypotheses test the functioning of a link economy in
online environments. Hyperlinks are a critical communicative tool for online
organizations, and as the news media community moves towards a self-sustaining
period hyperlinks serve as critical interorganizational linkages in the community.
126
Table 2 provides a summary of the outlined hypotheses, as well as the
general level of analysis and a brief description of how these hypotheses will be
analyzed. Each of these hypotheses and methods will be considered in more detail in
the next chapter. In aggregate, the above hypotheses present a new model for
understanding the transformation of organizational communities, and the general
process of reinvention currently underway in the US news media community.
127
Table 2: Summary of Hypotheses
Level Hypothesis Type of Analysis
Community RQ
1
: To what extent does the evolution of the online news
community adhere to the stage model of community development?
Univariate Network
Analysis, Longitudinal
Network Model
Organization –
Population
H
1
: During the emergence phase, new organizational forms are more
likely to link to other organizations of the same type than to different
types.
Longitudinal Network
Model
Organization –
Population
H
2
: During the emergence phase, there will be a proportionally
small, but significant, number of links between a new organizational
form and existing specialist organizations.
Longitudinal Network
Model
Population –
Community
H
3
: Links between a new organizational form and the existing
community will increase after the density of the speciated form
begins to increase.
Longitudinal Network
Model
Population H
4
: Media coverage of blogs and social networking sites (A), the
formation of professional associations (B) and legal recognition (C)
will be positively related to the birth rate of online-based news
forms.
Generalized Linear
Model (Poisson
Regression)
Population H
5
: An increase in the number of links between existing
organizations and new organizational forms will be positively
related to the birth rate of online-based news forms.
Generalized Linear
Model (Poisson
Regression)
127
128
Table 2: Continued
Population RQ
2
: Are there certain legitimacy measures that are more likely than
others to indicate the legitimacy of organizational forms?
Generalized Linear
Model (Poisson
Regression)
Population H
6
: During periods of transformation, organizations that actively
undertake strategic change initiatives will have a decreased
likelihood of failure
Generalized Linear
Model (Poisson
Regression)
Organization –
Population
H
7
: During periods of transformation, the adoption of aggressive
strategic initiatives will decrease mortality rates for inertial
organizations
Generalized Linear
Model (Poisson
Regression)
Organization –
Population
H
8
: The higher the number of institutional hyperlinks maintained by
online newspapers, the lower their mortality rate.
Generalized Linear
Model (Poisson
Regression)
Organization –
Population
H
9A
: Established organizations that have a symbiotic strategy with a
high ratio of linkages to new populations during periods of
emergence will have a high proportion of outlinks to new
populations over time.
Longitudinal Network
Model
Organization –
Population
H
9B
: Established organizations that have a symbiotic strategy with a
high ratio of linkages to new populations during periods of
emergence will have a high proportion of inlinks to new populations
over time.
Longitudinal Network
Model
129
Table 2: Continued
Organization –
Population
H
10A
: Established organizations with a high ratio of linkages to new
populations during periods of emergence will have a high proportion
of inlinks to new populations over time.
Longitudinal Network
Model
Organization –
Population
H
10B
: Established organizations with a low ratio of linkages to new
populations during periods of emergence will have a low proportion
of inlinks to new populations over time.
Longitudinal Network
Model
129
130
CHAPTER 5: DATA AND METHODS
New Approaches to Transformation Research
This chapter discusses the data used for the testing of the hypotheses that
have been formulated in chapters two, three and four. In addition, the methods of
analysis are discussed. First, the method for collecting historical Internet data is
discussed in detail. Second, the additional data sources are discussed. Then, the
methods of analysis are discussed in turn (univariate network analysis, longitudinal
network analysis and event history analysis). Each method is discussed in relation to
a specific hypothesis.
Studies of community ecology require the researcher to examine vast
organizational domains. Bryant (2003) outlines three methodological challenges
associated with community ecology research: (1) the establishment of community
boundaries, (2) the scale of data collection and (3) the determination of appropriate
analytical techniques. Before detailing the data and methods utilized in this study,
each challenge is first addressed.
The rapid rate of growth of online networks, paired with the problems
associated with online archival research, required new approaches and a
recombination of existing techniques. That said, the first challenge in studies of
organizational communities is in the establishment of community boundaries.
Previous studies delineate community boundaries in one of two ways: the use of
geographic boundaries or the demarcation based on functional criteria (Ruef, 2000).
131
The online news media community interacts primarily on the World Wide Web,
rendering geographic or spatial boundaries generally irrelevant. This study, however,
will focus primarily on the community as it exists within the United States. The
boundary of the online news media community is thus broadly defined based on the
function of the community: to provide news and information to consumers via
Internet-based media. Populations that exist within this community were described in
detail in chapter two.
The second challenge is the determination of the appropriate data. Studies of
community evolution require in-depth, longitudinal analysis in order to explain the
mechanisms of evolution. The scope of community-level studies is thus inherently
demanding, and more so in online environments. Scholars have previously
emphasized that such studies require a mixed methods approach in order to
accurately infer causality from an organizational history. Quantitative and qualitative
data must be used hand-in-hand in order to accurately explore community
development (DiMaggio, 1994; Fombrun, 1988). Both types of data are utilized in
this research.
The third challenge is the selection of appropriate analytical techniques. In
the past five years significant improvements have been made in terms of the
computational power available to researchers and the analytical tools developed for
longitudinal analysis. The current project thus employs new tools for network and
statistical analysis, and takes full advantage of advances in computing technology. In
132
response to the second and third challenges, this research is conducted following in
the practices of Web Science research (Berners-Lee, Hall, Hendler, Shadbolt, &
Weitzner, 2006; Lazer et al., 2009). The tools and techniques presented here merge
best practices from computer science with the theoretical advancements of the social
sciences. In combination, the full scope of community evolution is thus able to be
analyzed.
Digital Archive Research Using the Internet Archive
In order to study the evolution of a community of online organizations, it is
first necessary to extract a history of the community on the Internet. Archival records
have proven to be a valuable resource in organizational research, and organizational
texts, documents and databases have often been used in evolutionary studies
(Hannan & Freeman, 1977, 1984; Meyer & Rowan, 1977; Perrow, 1991). Stemming
from historiographic studies, this body of research has sought to recreate the life
histories and evolutionary process of populations and communities of organizations
(Ventresca & Mohr, 2002). In recent years, online archival databases have similarly
proven to be useful as a research tool. Numerous archival databases from Alexa have
been used in conjunction to examine the growth of journal citations over time
(Vaughn & Thelwall, 2003b) and Elsevier and Google Scholar have been utilized to
examine the impact of journals over time based on citation networks and impact
factors (Bakkalbasi, Bauer, Glover, & Wang, 2006; Hall, 2006). Prior studies (Arms,
Aya et al., 2006; Arms, Huttenlocher et al., 2006; Bakkalbasi et al., 2006; Vaughn &
133
Thelwall, 2003b) have thus shown that historical archives can be utilized for
organizational studies, and furthermore that data can successfully be extracted from
online archives for the purposes of historical research. To date, however, there are no
large-scale studies in the social sciences that have examined the long-term evolution
of a cohesive community utilizing online archives. In previous years, this type of
research has been difficult due to a lack of computing power and the lack of
available tools.
This study answers these issues by developing a new tool for research via
online archives. The primary database used for this study was the Internet Archive
(Archive.org). Founded in 1996, the San Francisco-based Internet Archive is a non-
profit that was founded with a mission to preserve the history of the World Wide
Web, and to build an Internet library containing that history. The Internet Archive is
often referred to as the Wayback Machine (WM); the Wayback Machine is the
graphic interface for the Internet Archive databases, and allows users to freely access
the information stored within the archive.
The Internet Archive provided direct
access to its databse, thus permitting large scale crawling. The Internet Archive
works in partnership with University of California, Berkeley’s Digital Library
Project, the Online Computer Library Center, and Alexa Internet to track Web pages
and archive digital copies. Initially the Internet Archive stored only Web pages, but it
has now expanded its mission to include the archival of texts, audio, moving images
and software. As of 2006, the Internet Archive had recorded, parsed and archived 40
134
billion Web pages, and with the rapid growth of the Internet the number of Web
pages has expanded to 85 billion Web pages as of 2009. While this constitutes only
twelve complete crawls of the full Internet to date, it provides a substantive sample
of Web pages. In addition, it is the single most extensive archive of the Internet. The
archive is also an actively expanding database, and continues to grow at a rate of 2
terabytes of data a day. There is, however, a 6- to 14-month lag between the time
when a site is crawled and the time when a site appears as a searchable file in the
Internet Archive.
A number of previous studies have examined the Internet Archive as a viable
research tool, and proposed potential research agendas (Arms, Aya et al., 2006;
Arms, Huttenlocher et al., 2006; Murphy et al., 2008; Vaughn & Thelwall, 2003a).
Despite this work, few studies have actually extracted, analyzed and applied data
from the Internet Archive in a research context, due in part to the numerous
challenges associated with this type of archival research (Arms, Aya et al., 2006).
Murphy, Hashim and O’Connor (2008) tested the validity of the Internet Archive’s
Wayback Machine as a research tool. They demonstrated the validity of the Internet
Archive as a tool for studying Web site use based on three measures of validity: face,
predictive and nomological validity. Their research additionally demonstrated that
the Internet Archive is a reliable tool for measuring the age of a Web site, and for
viewing revisions to a Web site’s content over time.
135
From a global perspective, data from the Internet Archive provides a setting
for the examination of relationships between organizations at a macroscopic level, as
well as on an organization-by-organization basis. This does not, however, provide
information on the strategies, alliances and partnerships that individual organizations
or groups of organizations implemented. Thus, in addition to the archival, macro-
level data, additional information about individual organizations was collected from
a combination of archives and interviews.
History Crawl: A Tool for Crawling the Internet Archive
Despite the wealth of information available to researchers through the
Internet Archive, there are a number of challenges that must be overcome in order to
extract data from the repository for research purposes. The Internet Archive is
structured implicitly; the nature and pattern of associations in the archive are based
on the relationships in the data as opposed to an explicit structuring based on
specified fields and relations (as in database programming) (Arms, Aya et al., 2006).
This structure is the result of a series of independent actions aggregated over time; as
a result, the burden of aggregating the data into a single, cohesive dataset is placed
on the researcher. Furthermore, the sheer scale of the data adds additional
complexity in the storage and archival of extracted data. In order to extract viable
data, intelligent search and manipulation tools must be implemented. Crawlers and
filters must be developed in order to extract viable data for study. Thus, the primary
136
challenges are the actual structure of the database, and secondly, the extraction of the
data into a format that can then be analyzed.
First, with regards to structure, the web site archives the majority of
accessible Web sites, with a number of notable exceptions: (1) pages with robots.txt
file
5
(2), pages that are coded entirely in the Java programming language, (3) web
pages that require the originating server to function (containing server side image
maps), (4) unknown Web sites that are not listed with Alexa, and (5) orphan pages
that do not link to other Web sites. Alexa Internet is a Web services company owned
by Amazon Inc. Alexa is a search engine and Web directory; the Web sites contained
within this search engine serve as a seed list for the Internet Archive’s Web crawler.
The Internet Archive additionally adheres to the Oakland Archive Policy for
Managing Removal Request and Preserving Archival Integrity. This is a formal set
of guidelines from the University of California Berkeley’s School of Information
Management and Systems, and outlines a framework for managing digital archives
and repositories.
6
Beyond those restrictions, the Internet Archive’s crawlers capture
the majority of pages accessible from US based servers. In order to maximize
crawler speed and storage space, images and videos greater than 10 megabytes in
size are not included in the archive. Data in the Internet Archive’s database is stored
in the WARC (Web ARChive) file format, which compresses portions of Internet
5
A robots.txt file is a standard exclusion file. It is implemented by Web designers to indicate to other Web sites
that the information on the given site is not to be crawled. Standard Web ethics indicates that such requests
should be honored.
6
http://www2.sims.berkeley.edu/research/conferences/aps/removal-policy.html
137
Archives crawls and specifies an architecture for aggregating multiple compressed
files to form a cohesive archive.
The actual crawling of the data presents structural challenges. Because the
data are extracted in a bitwise fashion, data are not in a cohesive order. When data
are crawled for a given year, for example 1999, data are returned across a multiple
year time span. If the Internet Archive crawls the New York Times in 1999 and finds
a link to the Los Angeles Times, it will not crawl the Los Angeles Times
immediately. Rather, it will queue the request and the linked-to page will be crawled
later on. In this way, an aggregate history is assembled, but in a bitwise fashion. This
places the burden of parsing data into a workable form onto the researcher.
In response to these challenges, a customer Web crawler, HistoryCrawl, was
coded and compiled in order to facilitate data crawling and extraction. The
development of HistoryCrawl was enabled, in part, by research being conducted by
Cornell University (Arms, Huttenlocher et al., 2006). Researchers at Cornell
University, working at the Cornell Theory Center, have worked for the past five
years examining large subsets of the Internet Archive. Their work provided a
foundation for the development of HistoryCrawl. HistoryCrawl accesses the Internet
Archive through direct access to the Internet Archive’s servers in San Francisco, CA.
Through the Internet Archive’s database, the crawler is able to extract a primary
record of the Universal Resource Locators (URLs), IP addresses, data and time of
capture for each site. The primary record also points to a ‘Page Store’ account, which
138
includes a single copy of each Web page, and is indexed to remove duplicate pages
resulting in an increase in search speed (Arms et. al., 2006).
HistoryCrawl is written in the Python (v2.6) programming language
(Rossum, 2010). HistoryCrawl is a multistage crawler that allows for a significant
amount of flexibility. In step one, the user inputs preliminary requirements into the
configuration file (config.ini): the preliminary requirement for the crawler is a seed
list of URLs reduced to the root URL (i.e. www.usc.edu/sports is reduced to
www.usc.edu). In addition, in step two the user must specify the date range and
depth for the crawl in the start file (start.py). HistoryCrawl is capable of crawling
any date from 1996 through to six months prior to the crawl date. The user specifies
years to crawl, but the crawler will examine every date within that year that is
available for a given URL. In addition to the dates for the crawl, the user must
specify a depth for the crawl. A depth of zero will crawl only the initial set of URLs;
a depth of two will crawl the initial set of URLs and one step outwards; a depth of
three will crawl three steps outward. In addition, a user can specify the depth within
a single domain, although this is generally best left unrestricted.
Once the configuration and start files have been specified, the user can
initiate the crawl. The crawler examines each URL in a chronological order, and
accesses the WARC files stored at the Internet Archive. As illustrated in Figure 8,
the crawler will select a given URL and examine it for one year. It will then examine
the pages that URL points to and store those links. As previously explained, because
139
of the bitwise fashion in which the crawler operates, a Web page in 1998 may point
to Web pages in 1997, 1998 and 1999. All pages are accessed and links are retrieved;
the chronological order is filtered after the crawl is complete. Because the crawler is
accessing compressed archive data, the time to complete a crawl generally takes a
minimum of an hour for a dozen Web sites, and scales up exponentially. Once the
crawl has been completed, two post processing scripts run. Post-processing Script A
reduces the crawled data to a link list that includes the date of the archived Web site,
the pages the Web site links to, and the frequency of links for that date. Post-
processing Script B filters the link list to remove spam and advertising Web sites.
The the resultant data are stored in a MySQL database. Final output includes a
master list of sites crawled, and a tab for each year containing the link list from the
output. In addition, a summary audit is generated in hypertext markup language
(HTML). The summary audit lists each Web site crawled and specifies if any Web
sites were unavailable.
140
Figure 8: HistoryCrawl operating schema
Input URL List
Time Period
Internet Archive
Database
Site 1 | Site 2 | Site 3 | Site 4 | Site 5 É
1996 | 1997 | 1998 | 1999 | 2000 | 2001 É
1996 | 1997 | 1998 | 1999 | 2000 | 2001 É
Post-Processing Script A reduces data to a link list
containing the date and number of times Site X links to Site Y
Post-Processing Script B removes spam and advertising Web sites
Record ID | Site 3 | Links to -> Site X | Frequency | 1997
Record ID | Site 3 | Links to -> Site Y | Frequency | 1998
Record ID | Site 3 | Links to -> Site Z | Frequency | 1999
The HistoryCrawl program has a text-based interface that allows users to
input a number of parameters, including sample set, year and scope. Sample set
refers to the input URLs, which serve as the starting point for the crawler. For this
research project, the sample consisted of the top 100 primarily US-based news Web
sites as rated by Alexa.com; after removing invalid sites excluded by robots.txt
exclusions, the sample list included 76 Web sites. As discussed previously, sites with
141
a robots.txt file are excluded from the Internet Archive, and thus could not be
included in this study. The 76 seed organizations are listed in Appendix A. This
provides a substantive starting set; with this sample, the crawler was run for a single
year at a time, from 1997 through 2007. The crawl was run at a scope of three (scope
= 3), which captures the Web sites three steps out from the starting sample. A depth
of three was chosen in an attempt to maximize the number of sites collected by the
Web crawler; it is potentially possible to crawl beyond a depth of three, but to date
the crawler has not been shown to be stable beyond this point.
In all, with a starting set of 76 Web sites, a network of 25,628 Web sites was
extracted from the Internet Archive using HistoryCrawl. After filtering, the number
of Web sites was reduced to 2,977. This is a 90 percent exclusion rate, but this is not
surprising given the large amount of advertising that exists on a most commercial
Web sites. A review of nytimes.com, for instance, reveals that in 2007 only 3 of 25
links on the Web site’s home page were directed to domains outside of nytimes.com.
Furthermore, the requirement of three hyperlinks in a year to establish a tie between
two organizations set a stringent requirement. This study erred on the side of caution
in order to insure validity of the study results. Each record extracted contains four
pieces of information: starting URL, linked-to URL, count and date. The count field
indicates the number of times that a link occurred for the given date. The crawl was
executed between June and December 2009 utilizing a dual processor configuration
to increasing processing and crawl speed. Once crawling was completed, and prior to
142
any analysis, it was necessary to validate the accuracy of the data returned by the
crawler. In order to assess the validity of this sample pool of data two tests were
conducted: (1) a subset of data was analyzed to verify the accuracy of the crawler in
accurately extracting links from the Internet Archive and (2) the data were analyzed
to assess the impact of restrictions imposed by the depth of the database.
In order to assess the validity of the crawler’s accuracy in extracting links
from the Internet Archive, 10 Web sites were selected randomly and the results from
the HistoryCrawl search were assessed against a crawl of the Internet Archive
conducted by hand.
7
Random selection should avoid any major bias resulting from
Web site size or type, but its possible that these biases may exist. Links were
checked in 2000, 2002 and 2004, in January of each year. The results of the hand
verification are compared against the results from HistoryCrawl in Table 2.
Table 3: Accuracy of HistoryCrawl versus hand crawl
HistoryCrawl Hand Crawl Accuracy
January, 2000 955 1028 93%
January, 2002 1102 1295 85%
January, 2004 902 968 93%
As the results indicate, taking the hand crawl to be an accurate accounting of links,
HistoryCrawl is shown to be 85 to 93 percent accurate in this analysis. This
represents a limited sample, yet it demonstrates generally strong validity in detecting
a substantive population of links.
7
The 10 test sites were www.boston.com, www.statesman.com, www.sun-sentinel.com, www.freep.com,
www.latimes.com, www.omaha.com, www.jsonline.com, www.pe.com, www.oregonlive.com,
www.ocregister.com, www.sfgate.com
143
Second, it was necessary to analyze the impact of restrictions imposed by the
depth of the database on the results of the search. HistoryCrawl is designed to search
through records in the Internet Archive and extract links to other Web pages. In
many cases, however, the Internet Archive has not captured the full depth of a Web
site. For example, on the Los Angeles Times web site, latimes.com, it is possible for
a user seeking information on entertainment to click seven times before getting to the
end of a story. In other words, latimes.com has a Web site depth of seven, and there
may be links on each of these pages. But HistoryCrawl only accesses what is
available in the Internet Archive, and very often the Internet Archive will crawl and
store only a portion of a given Web site. In order to assess the impact of this
discrepancy, the 10 Web sites used in the previous test were again utilized for this
analysis. For each Web site, the number of links in the current Web site was
compared against the number of links in the most recent archived version. Links
were counted using DiagnosticWeb’s page rank tool, which counts the number of
links on a given website (www.diagnosticweb.com). The average lag between the
test sites and the most recent archived version was 7.2 months.
Table 4: Page Depth and Link Validity Analysis
Live Site Archived Site
Depth (avg) 7.2 5.4
Number of Links (total) 850 768
144
As shown in Table 4, the results show that HistoryCrawl captured 90 percent of the
links that actually existed, even though the archived versions contained only 80
percent of the depth of a given Web site.
Thus, HistoryCrawl was utilized to collect data for this research, and shown
to be a reliable research tool for extracting archival network data from a body of
Web sites.
Interview Data
In-depth interviews were conducted in order to gain additional information
about the strategies employed by organizations included in this study. The
exploration of strategic imperatives at the organization level suggests an approach
similar to Burgelman’s (1983) implementation of Glaser and Strauss’s (1967)
grounded theory. Prior studies of organizational relationships and strategic decisions
in the organizational setting have approached research from one of three
perspectives: functionalist, interpretive and postmodern (Gioia, 1998). The
functionalist approach advocates an objectivist approach oriented towards the formal
testing of hypotheses and postmodern scholars focus on the interpretation of what
they view as fragmented and indeterminate texts. Interview data can be applied and
analyzed from an interpretive approach, looking at the implementation of strategy
and the use of alliances as perceived by the actors involved in this system. Further,
when paired with network studies, Wasserman and Faust (1994) note that interviews
are particularly useful in providing additional information about the context and
145
strength of interorganizational relationships. This study follows in the functionalist
approach, using the raw text of the interviews as a tool for supporting and
complementing what is revealed in the analytical results of this study. The functional
approach treats the observations as given, and does not seek to interpret added
meaning beyond the context of the given conversation. Furthermore, the functionalist
approach is complementary to existing hypotheses, and complements formal
statistical tests. Both the interpretive and postmodern approaches rely more heavily
on interpretation based on context, and extend the qualitative analysis beyond the
context of traditional statistical testing (Gioia, 1998).
In order to capture comprehensive data on the strategic use of
interorganizational linkages and relationships, open-ended interviews were
conducted at a number of media organizations. The objective in selecting interview
sites was to select subjects from a wide-array of organizations, in both management
and the production side of news media. Subjects were selected based both on
availability and history with a given organization. The minimal tenure of a
participant was two years; the maximum tenure at a given organization was 25 years.
Both traditional news organizations and new media news sites were included. In
total, 48 interviews were conducted over a period of four months, with each
interview lasting roughly sixty minutes. The primary goal of the interviews was to
learn about the organizational strategies employed with regards to hyperlinking. In
many cases it was possible to ascertain what the strategy was during formational
146
periods, but in certain cases turnover made this difficult. In these cases, strategy was
extrapolated based on the history that was available. In total, 42 hours of interviews
were collected. The organizational role of interview subjects, the names of the
organizations, and the interview lengths are listed in Appendix B.
Interview subjects were initially contacted via email, with follow-up
correspondence via telephone. Subjects were not offered any compensation, however
participating organizations were provided with an overview of key findings as a
result of this research study. The interviews were semi-structured and open-ended,
with 10 preset questions. The interview script used is provided in Appendix C. Due
to the sensitive nature of interview questions (with regards to organizational
strategy), interview subjects were offered the option of anonymity in providing
information. Identifying information is provided if a participant waived his right to
anonymity. Scripted questions addressed a number of critical organizational issues
pertaining to strategic partnerships, policies regarding blogging and the use of other
new media, as well as decisions to form coalitions and content sharing partnerships.
Questions further pertained to individual reactions to organizational policy, as well
as the perceived effectiveness of these policies. In this way, interview questions
addressed organizational issues at multiple levels, and from multiple perspectives.
Structured questions were designed to take no more than thirty minutes to
complete, leaving significant time for discussion of other issues. Participants were
told that interviews would last for sixty minutes, however many interviews were
147
longer. Additional information was collected, and for many interviews organization-
specific events were examined.
Additional Data Sources
In addition to the Internet Archive data, additional archival records were used
to collect information about the population of print newspapers that competed within
the online news media community. The source of data for this research was the
Editor and Publisher International Year Book, published by Editor & Publisher. It is
one of the most extensive directories of newspapers in the United States and also
provides comprehensive coverage of international newspapers. The Editor and
Publisher Yearbook is based on a yearly survey of newspapers in the United States,
and has been published every year since 1920. The yearbook contains information
about the circulation, ownership, and function of nearly every daily newspaper in the
country. Data were coded into an electronic SQL database by hand. Data were
collected to correspond with the newspaper organizations that were obtained from
the HistoryCrawl results, as described above.
In addition to the yearbook, additional data were collected from Alexa.com
for the Web sites extracted from the crawl. Alexa’s databases contain a wealth of
information about millions of Web sites, including details of traffic activity,
popularity, founding date, location and ownership. Alexa was used to collected
attribute data for each of the Web sites extracted from HistoryCrawl.
Variables
148
In addition to the historical network data and the interview data, a number of
variables were collected using the previously described data sources. These variables
were used for visualization, network analysis and the event history analysis. For each
organization extracted from the Internet Archive, the following attributes were
included as variables in the database: the URL, the location of the company to which
the URL is registered and the Web site name.
In addition to the above, a number of additional variables were either
calculated or extracted. For each Web site, the Alexa traffic rank value was recorded
as a variable. The Alexa traffic rank is a composite measure that averages the daily
visitors and the total pages views for the past three months. This measure is a proxy
for the overall popularity of a given Web site, and can be used to compare a Web site
against others in the set.
Newspaper Organization Variables
The development of the newspaper population was a central focus in this
research. Thus, for newspaper organizations a number of additional variables were
collected. Utilizing the Editor & Publisher International Yearbook, the following
variables were collected: founding date (of the print newspapers), average weekly
print circulation in each year, number of employees in each year, type of paper
(morning or evening), publisher, editor, online editor and parent company. In
addition, for each newspaper a variable was created to account for the potential
audience. This was measured as the Audit Bureau of Circulation’s retail trade zone,
149
which is defined as the extended metropolitan area of a city or town within which
business for the city or town is normally conducted.
Event Variables
For the event history analysis, a number of additional variables were
collected. Population level event data were collected for the amount of media
coverage, professional associations and legal events. A keyword search was
performed on the Lexis-Nexis Database to collect information about media coverage
and the development of new organizational forms. Media coverage data were
collected for the full time period, searching every year from 1998 to 2007. Media
searches were conducted for two types of organizations: blogs and social networking
sites. For blogs, the following keywords were searched: blog, weblog and blogging.
For social networking sites, the following keywords were searched: social
networking, social network and social networking site. Keyword searchers were
restricted to return results only when search terms appeared in the headline of the
main body of a given article. The search was further restricted to US newspapers;
this feature restricted the search to 557 major newspapers and wire services in the
US.
Professional associations were coded through a combination of extensive
document review and interview research. Interviewees were asked to list the
professional organization to which they belonged. In addition, the International
Directory of Professional Association’s online database
150
(www.associationsdirectory.org) and Google’s organization’s directory were used to
research additional organizations. Each organization’s Web page was then accessed
to obtain information about the founding date of the organization, the area of focus
and approximate membership. Where information was not available through the
organization’s Web site, the organization was contacted directly. The primary
organizations are listed in Table 5:
151
Table 5: Professional Organizations in the Online News Media Community
Organization Founding Date Membership Focus
Interactive
Advertising Bureau
1996 461
organizations
Online and
interactive
advertising
The Web Marketing
Association
1997 Online marketing
and media
Online News
Association
1999 1600 News professionals
focused on digital
media
Online Publishers
Association
2001 52 organizations Online publishing
International
Blogging and New
Media Association
2002 2000 Bloggers
Media Bloggers
Association
2004 5000 Bloggers
Internet Satirical
Newspaper
Association
2005 250
organizations
(estimated)
Satire-focused
online news sites
International
Association of
Online
Communicators
2006 800 Broad online
community:
subgroup for
bloggers
Social Networking
and Media
Association
2007 1100 Social networking
and participatory
media
World Professional
Blogger Journalist
Association
2008 52000 Bloggers focused
on news media
152
Legal rulings were researched utilizing the Global Legal Information
Network and the Internet Law Library (www.internetlibrary.com), which is an online
repository of legal rulings pertaining to online information. The directories were
searched for findings pertaining to online news media and First Amendment rights.
In addition, rulings pertaining to online media and Shield Laws were also included.
Seven key rulings were found, summarized in Table 6:
153
Table 6: Key Court Rulings Affecting Online Media – 1998 – 2009
Case Key Finding Population
Apple Computer v Doe 1, et al
Case No. 1-04-CV-032178
Superior Court, CA
March 11, 2005
Shield laws protect bloggers, but does not
protect information that provides evidence of a
crime. The same is true for journalists.
Journalist-
bloggers
BidZirk, LLC, et al. v Philip J. Smith
Case No. 6:06-109-HMH
District Court, SC
October 22, 2007
Statements of opinion posted on a blog are
protected under freedom of speech protection.
Bloggers
In re Does 1-10
Case No. 06-07-00123-CV
Texas Court of Appeals
December 12, 2007
Shield laws protect bloggers, but does not
protect information that provides evidence of a
crime. The same is true for journalists.
Journalist-
bloggers
O’Grady v. Superior Courtney Schultz
Case No. WL1452685
California Appeals Court, 4
th
District
May 26, 2006
Bloggers are extended the same privilege as
journalists under California’s Shield Law. This
ruling extends to the publishers and editors of
blogs acting as news sources, and includes
protection against the disclosure of sources.
Journalist-
bloggers
Grand Jury Proceedings (Jason Wolf)
Northern District of California
Case No. CR 06-90064 WHA
2006 – 2007
A blogger, Jason Wolf, was held in contempt
in prison for 169 days. Wolf refused to testify
and turn over video he had taken of a protest.
The charges were dropped when Wolf posted
the video directly to his Web site.
Journalist-
bloggers
154
Table 6: Continued
Too Much Media, LLC vs. Shellee
Hale
Superior Court of New Jersey
Case No. MON-L-2736-08
July, 2009
A New Jersey court ruling found that the
state’s shield law does not apply to
independent bloggers. New Jersey’s shield law
specifically applies to journalists affiliated
with a news media organization.
Bloggers
Google Inc. vs. Joan A. Madden
Supreme Court of New York
Case No. 100012/09
August, 2009
A bloggers identity is not protected in
instances where the comments posted by a
blogger are shown to be defamatory. Matters
of opinion are not protected as privilege.
Bloggers
154
155
Data Coding and Transformation
Interview Data
Interviews were all transcribed and verified for transcription accuracy.
Interview data were coded and analyzed using the Atlat.TI qualitative analysis
software package. Atlas.TI is a program for storing, coding, structuring and
analyzing interview data. Each interview was loaded into the system, and then coded
according to key themes that emerged in the data. The advantage of thematic
analysis in Atlas.TI is the quantifiability of the data. Once the full body of interviews
was coded, it was possible to analyze the intensity with which different themes
emerged, based largely on the frequency of mentions and the prominence of their
discussion. In this way, the findings from interviews could then be layered into the
network analysis for hypothesis testing.
After the interviews were transcribed, two coders analyzed the data. The first
coder was the researcher for this project, and the second coder was a graduate
student from a major Midwestern university. The second coder was familiar with the
framework used in this research, but had not previously worked with any of the
organizations participating in this study. Given the relatively small sample of
interviews, each coder reviewed all interviews. Each coder reviewed the material and
coded each interview for evidence of a particular strategic alignment with regards to
the creation of hyperlinks between the organization of interest and other
organizations. The strategic orientation of an organization was coded with regards to
156
the degree to which an organization used hyperlinks to prolifically connect with
other organizations. With no preconceived strategies, organizations were coded on a
Likert-type scale, with strategies being coded as (1) isolating, (2) somewhat-
isolating, (3) somewhat-prolific, and (4) prolific. Reliability was tested for the
agreement of coders on the type of strategy employed by an organization using
Cohen’s kappa. Cohen’s (1960; 1968) kappa is a measure used to assess the
agreement between two coders assessing the same body of information. Values for
Cohen’s kappa range from 0 to 1; a value of 1 indicates perfect agreement between
codes, whereas a value of 0 indicates no agreement. Any value greater than 0.81
indicates excellent agreement between coders, and values between 0.40 and 0.80
indicate moderate to substantial agreement (Landis & Koch, 1977). Intercoder
reliability for the coded interviews was excellent (kappa = 0.89).
Network Data
In order to analyze the results from HistoryCrawl as network data, it was
necessary to transform the data from a database format to a network format. For each
year, a link list was extracted from the database. A link list is a text file that lists the
web site searched and the web site linked to, as well as the number of times that a
given link occurred. There is a line for each link. The link list file is flexible and
requires minimal memory. In addition, all major network analytics programs can
read it. Within the network data, each actor was identified with a unique
157
identification number. A separate master list provides the organization names that
correspond with the actor identification numbers.
It was also necessary to delineate between random hyperlinking and strategic
hyperlinking. In other words, it was necessary to determine how many hyperlinks in
a year were needed to establish an interorganizational link. Hyperlinks generally
have no monetary cost, are easily maintained and are readily deleted (Halavais,
2008). As a result, it is generally very easy for an organization to haphazardly create
and delete such linkages. In order to account for this, a network tie was not recorded
unless there were three or more hyperlinks between two organizations in a given
year.
Newspaper Strategies
The interview data coding determined the baseline strategies employed by the
42 newspaper organizations for which interview data were available. The
HistoryCrawl results yielded 487 newspaper companies owned by 123 parent
companies. In order to fully assess the strategies employed by members of the
newspaper population, it was necessary to extrapolate strategies based on the
combined network and interview data.
Four strategies were found to be predominant: blogification (blink),
diversification (dlink), experimentation (elink) and isolation (ilink). These strategies
were determined based on the results of the previously described coding of the data.
It is possible that a more nuanced breakdown of link strategies exists, however the
158
above categories represent the predominant hyperlinking strategies as found in this
analysis.
Table 7: Organizational Linking Strategies and Density Ranges
Strategy OutDegree Range No Overlap
Number of
Organizations
Isolation 0.0000 – 0.0013 0.0000 – 0.0004 162
Experimentation 0.0004 – 0.0062 0.0006 – 0.0018 176
Diversification 0.0016 – 0.0078 0.0019 – 0.0048 72
Blogification 0.0028+ 0.0049+ 77
The first column shows the normalized out degree centrality for the 42 newspaper
organizations that were hand coded. There is clearly overlap in the density ranges;
thus, in order to create distinct categories for the densitites corresponding to the
above strategies, the difference between two categories was split based on the mean
For example, the upper bound density of Isolation was 0.0013 and the lower bound
for Experimentation was 0.0004. Examining the distribution of these strategies, it
was found that the upper bound for Isolation was 0.0004 (two outliers existing at
0.0010 and 0.0013) and the lower bound for Experimentation was 0.0006. The
resulting ranges are given in the second column. Lastly, the number of organizations
utilizing each strategy (based on the density ranges) are given. The distribution
clearly skews towards isolating strategies, a fact that is further reinforced by the
histogram in Figure 9, demonstrating the overall distribution of strategies.
159
Figure 9: Histogram of Hyperlinking Strategies Based on Density Ranges
!
"!
#!
$!
%!
&!
'!
(!
)!
*!
"!!
!+!!!#
!+!!!$
!+!!!%
!+!!!'
!+!!!)
!+!!"
!+!!"#
!+!!"%
!+!!"'
!+!!")
!+!!#
!+!!##
!+!!#%
!+!!#'
!+!!#)
!+!!$
!+!!$#
!+!!$%
!+!!$'
!+!!$)
!+!!%
!+!!%#
!+!!%%
!+!!%'
!+!!%)
!+!!&
!+!!
!+!!&%
!+!!&'
!+!!&)
!+!!'
!+!!'#
!+!!'%
!+!!''
!+!!')
!+!!(
!+!!(#
!+!!(%
!+!!('
!+!!()
!+!!)
!+!!)#
!+!!)%
!+!!)'
The effects of this distribution, and of organizational choice in hyperlinking strategy,
are the subject of subsequent hypotheses.
Specialists and Generalists
Each of the 487 newspapers was coded as a generalist or a specialist. A
generalist is an organization that focuses on a broad resource pool, whereas specialist
organizations compete in niche locations or resource pools. Organizations were thus
coded as generalists if they were a major daily newspaper; niche newspapers were
coded as those occupying either topical niches (business, sports, etc.) or metro niches
(i.e. community newspapers). Both coders reviewed each of the 487 Web sites.
Intercoder reliability for the coded interviews was good (kappa = 0.79). 120
organizations were found to be generalist organizations, and 367 were determined to
be specialists.
160
Environmental Level Events
This study examines organizational transformation at all levels of community
ecology; thus, in addition to the network and strategic data, it was also necessary to
recreate a history of major environmental level events that impacted the development
of the community. During interviews, the researcher asked participants to identify
major events that had helped to shape the development of the online news
community at large. Instead of using theory driven language, the interviewer told
participants to describe significant developments in politics, technology, society or
economics that had impacted the development of the community. This aligns with
previous concepts of what constitutes an organizational community (Aldrich & Ruef,
2006; Ruef, 2000). Fifteen different events were mentioned throughout the course of
the interviews; only three events, however, were mentioned three times or more.
Those events are listed in Table 8:
Table 8: Environmental events affecting the online news media community
Environmental Event Time Period
Number of Participants
Mentioning Event
Dot.Com Bubble Burst 2000 – 2001 45
Growth of Blogs 2002 – 2003 42
Emergence of Social
Networking Sites
2005 – 2007 44
161
All community level events mentioned by interview participants were focused on
technological developments. Given the nature of change in the industry, this is not
altogether surprising. The rapid growth of blogs and social networking sites could
potentially be viewed as a population level event, but the technology that was
introduced during both periods drove significant community-wide change. In this
way, both events introduced new technology that disrupted environmental level
resource availability. As such, they are treated here as community level events.
Analysis Overview
Two familiar analytic techniques were used to analyze the proposed
hypotheses. Before discussing the specific analyses, an overview of both methods is
provided. The network analysis techniques, in particular, represent recent
advancements in communication research, and as such a brief introduction is
beneficial.
Network Analysis
Social network analysis has been widely utilized for the study of
organizational relationships that result from processes of organizational
communication (Monge & Contractor, 2003), and in recent years network analysis
has been shown to be particularly useful in the study of evolutionary processes at all
levels (Monge et al., 2008; Monge et al., 2009 in press). Exponential random graph
(ERG) analysis, also known as p* analysis, is a particular class of network analysis
that allows for the analysis of a variety of structural patterns such as transitivity,
162
homophily, and preferential attachment, among others (Robins, Pattison, Kalish, &
Lusher, 2007). ERG modeling takes a starting set of parameters, simulates a
distribution of random graphs against the observed inputs and repeats the process
until a stable set of parameters is returned. Markov chain Monte Carlo maximum
likelihood estimation was used to predict the probability of certain structures while
accounting for interdependencies among network nodes (Robins et al., 2007;
Snijders, Pattison, Robins, & Handcock, 2006). The advantage of the method of
analysis is that parameter values predict the probability that certain structural
characteristics will occur: the probabilities can then be compared over time to predict
the probability of increases or decreases in certain relationships. Pairing parameters
with network attributes further enhances this analysis (Snijders et al., 2006). ERG
modeling also allows the researcher to examine the actions of an organization or a
population within the context of the larger community. The object of study is the
network itself, and as a result the analysis inherently takes into account the linkages
between the object of study and the community at large.
Statnet and RSiena were used for analysis: both Statnet and RSiena are social
network applications based on the R open-source statistics framework. R is a
programming language and programming environment designed for statistical
computing ("R: A language and environment for statistical computing," 2010).
Statnet (Handcock, Hunter, Butts, Goodreau, & Morris, 2003, 2008) is designed for
analysis of single instances of a social network. Statnet allows users to calculate a
163
wide array of social network metrics including density, centrality (degree to which a
user is centrally positioned within a network of alters), reciprocity and homophily.
Statnet is not a standalone program; it is interdependent on a number of other
packages that support its operation: network (Butts, 2008) is a package that contains
routines for calculating basic network measures and helping to prepare data for
analysis, and ERG models (Hunter, Handcock, Butts, Goodreau, & Morris, 2008)
This software suite allows users to execute a wide host of network analysis routines.
One primary advantage of the statnet suite of programs, compared to other network
packages, is that the program is optimized to run in the R framework. R is a
computationally lean package, which means that it is optimally suited for the
analysis of large data sets. As a result, this package is particularly well suited for
large-scale ERG models, which in the past have been analytically difficult due to
computational requirements (Hunter et al., 2008).
ERG models were thus used to calculate parameters for the network data
collected in this research. Parameters are exponential probabilities that represent the
likelihood that a particular network configuration appears within a given dataset.
Positive and significant parameters indicate that a hypothesized graph configuration
is present in a given network and statistically significant compared to random
change. A negative and significant parameter, on the other hand, indicates that a
hypothesized graph configuration is present in a given network, and that fewer
164
configuration are present than would be expected by random chance (Robins et al.,
2007).
Likewise, RSiena utilizes the computational power of R to allow for the
longitudinal analysis of social network data. Unlike statnet, RSiena (Snijders,
Steglich, Schweinberger, & Huisman, 2007a) is not capable of estimating ERG
parameters. There are longitudinal network analysis packages capable of estimating
change using ERG routines, including Siena 3.1 (Snijders, Steglich, Schweinberger,
& Huisman, 2007b) and LPNet (Lusher & Ackland, 2008). Siena 3.1 is capable of
calculating longitudinal change for basic ERG parameters across multiple points in
time. LPNet is capable of calculating longitudinal change for higher-order ERG
parameters, but only between two points in time. Both programs are limited in their
scalability; the maximum matrix that either can calculate results for is limited to
roughly 300 x 300. RSiena utilizes the R framework, and the established statnet and
network packages, to allow for the analysis of much larger matrices. The current
version of the program is capable of handling matrices up to 3000 x 3000. Thus, in
this research RSiena is utilized for longitudinal analysis, but does not allow for ERG
analysis. Rather, RSiena calculates parameters specifically designed to capture
longitudinal behavior. RSiena calculates parameters that predict the probability that
organization A will send a link to organization B; parameters are calculated on a
logit scale and are interpreted as unstandardized effects in a logistic regression.
165
RSiena is based on a statistical process known as Markov-based models.
Markov analysis is utilized because it is a process suitable for time-dependent
stochastic processes. In the process applied here, time, t, is taken to be continuous,
but the network has been observed at a subset of intervals. RSiena calculates a
probability matrix that predicts the likelihood of a link between organizations
transitioning from its original state at time t, into another state at time t + !. Markov
analysis calculates the transition rates rather than the probabilities (Snijders, 2009;
Snijders et al., 2006; Snijders et al., 2007a; Snijders, Steglich, & Van de Bunt, 2010).
In other words, this analysis calculates the rate at which links will change states
(from link to no link, for instance). RSiena, however, is an actor-oriented estimation
and simulation package. Actor-oriented social network models treats the
development of a social network over time as the result of relational actions that
occur between individual actors. The structure of the network, distribution of actors
and attributes are used to determine restrictions on the network. Beyond those
restrictions, however, a key assumption of actor-based models is that each actor will
seek to maximize its individual utility (Snijders, 2009; Snijders et al., 2010; Van de
Bunt & Groenewegen, 2007).
The resulting models merge continuous Markov models and simulation, but
optimize the utility of individual actors. Using multiple samples of a network from
various points in time, the RSiena package estimates the utility maximization
preferences for each actor in the network and simulates the actions of a continuous
166
time scale. This model has been used extensively to study social interaction and
social networks (van Duijn, Zeggelink, Huisman, Stokman, & Wasseur, 2003). More
recently, a number of studies have shown that actor-based models can similarly be
applied to examine organizational networks that represent the formation of
partnerships, alliances and collaborative networks between organizations (Andrew,
2009; Gonzalez-Bailon, 2009; Van de Bunt & Groenewegen, 2007). As the
theoretical arguments of this paper have outlined, organizations within the news
community network are predicted to act as rational actors, either seeking out
partnerships or choosing to remain in isolation. The actor-oriented approach of
RSiena will capture this network behavior through a longitudinal examination and
modeling of each actor’s over-time behavior.
Thus, statnet and RSiena were used in conjunction to analyze the network
data and assess the predictions made in the research hypotheses. Statnet was used to
establish a baseline model for the network. Although Statnet cannot predict change
between periods, the baseline ERG model shows the most probable configurations in
each time period. Subsequently, RSiena was used for the longitudinal analysis to
calculate rate change parameters and to estimate the influence of modeled attributes.
The structural model of the network serves two purposes. First, the modeling of
density and inlink and outlink creation captures the overall behavior of the network
over time. Second, the structural model controls for the general structure of the
network; when modeled in conjunction with attribute data the analysis thus controls
167
for the structure of the networks and allows for a more accurate assessment of the
exogenous attributes that affect the network. Furthermore, by modeling the observed
network against a range of randomly generate networks that share the same structural
features, the role that exogenous variables have in predicting link creation is more
accurately captured (Gonzalez-Bailon, 2009; Snijders et al., 2010).
UCINET (Stephen P. Borgatti, Everett, & Freeman, 2002) was used for
standard network analysis of sociomatrix data. In addition, Netdraw v1.0 (Steven P.
Borgatti, 2002) was used to visualize the data.
Poisson Process: Estimating Organizational Birth and Death
In addition to the network analysis, the organizational birth and death rates
within the community where modeled. A Poisson regression was used to estimate the
covariates of the birth and death rates in order to assess the validity of the
hypothesized relationships. Poisson regression has been well established as an
appropriate strategy for the measurement of organizational count data (Hannan &
Freeman, 1988, 1989; Haveman, 1995; Kelly & Amburgey, 1991; M. D. Shumate,
2003). The baseline model is as follows:
PR( Y = y
it
) = e
-
!
!
it
y
it
/ y
it
! (Haveman, 1995, p. 600)
where y
it
represents the number of new online news-based organizational form births
or deaths in a given year, t. The ! is further specified as follows:
168
! = e
"
’x
it
(Haveman, 1995, p. 601)
Where exogenous variables are incorporated in the vector x
it
. Thus the baseline
model is extended by the inclusion of historical data and explanatory factors. The
exponential function further insures that the founding rate will be nonnegative. In
addition to the above, McFadden’s pseudo r
2
was calculated as an estimated r
2
in
order to compare the fit of the different models.
The use of a Poisson distribution assumes that the variance is less than or
equal to the mean, but ecological count data often violates this assumption (ver Hoef
& Boveng, 2007). As a result, standard errors are artificially small, and significance
levels are inflated (Cameron & Trivedi, 1986; Haveman, 1995; King, 1988). In order
to account for this, a quasi-Poisson model can be utilized. The advantage of utilizing
a quasi-Poisson approach, as compared to other approaches such as negative
binomial models, is that the parameters are left in their natural state and standard
model diagnostics can still be employed. The traditional Poisson model captures
variance in the term #V(µ) where # is a constant and V(µ) specifies the variance as a
function of the mean. In the Poisson model, # =1; the quasi-Poisson process allows
for estimation of the model parameters without an error distribution for the response
variable (McCullagh & Nelder, 2003). Essentially, this allows for an estimation of
the # parameter to correct for overdispersion (Everitt & Hothorn, 2006). In cases
where overdispersion is a possibility, both the Poisson regression and quasi-Poisson
regression should be presented to determine the optimal fit (Minkoff, 1993).
169
Following Shumate (M. D. Shumate, 2003), the quasi-Poisson model is given for
comparison in the case of the fully saturated model. In addition, Dean’s test was
utilized as an additional check for overdispersion. Dean’s test compares the observed
variance against the theoretically expected variance of a normal Poisson model – the
model calculates an approximated Pearson’s test comparing the variance of the fitted
model against the expected variance. Significance of the resulting parameter, D,
indicates that variance is greater than expected and overdispersion is an issue.
In addition to the above restrictions, historical data for the news media
community in the online context is left censored due to the restrictions imposed by
the Internet Archive (data collection started with 1996). Additionally, data is right-
censored to a certain degree. Archive data extends through 2007, although a
significant portion of Web sites continued to function today; data beyond 2007 has
not been sufficiently verified for analysis. Further, in order to capture substantive
relationships in the online news media community, relationships were analyzed on a
year-by-year basis. Thus, organizational births and deaths are modeled with a
waiting between events censored to one-year intervals.
Hypotheses and Analysis
The first research question sought to apply Bryant and Monge’s (2008)
community stage evolution model to this analysis of the news media community.
Bryant and Monge established that a stage model can be evaluated by analyzing the
densities, $, for the full networks for each time period. In their study, the minimal
170
density, at which the community moved from emergence to maintenance, was 0.50,
and the critical mass, at which point the community moved from maintenance to
self-sufficiency, was 0.75. It’s likely that in the case of the online news community
the densities levels will be significantly lower, given that online networks tend to be
sparsely populated (Barabasi, 2002). Densities were thus calculated for each of the
ten years (1998 – 2007) for which data were collected using HistoryCrawl. The text-
based edge lists generated by HistoryCrawl were input into UCINET 6.0. Data were
then assigned time periods based on the key inflection points shown in the data.
Densities were compared using the UCINET “Compare Densities > Paired” routine;
this compares two sets of densities uses bootstrapping techniques to estimate the
standard errors and t-tests. Alpha was set at 0.05 for these tests.
Prior to further hypothesis testing, descriptive statistics were calculated to
demonstrate the general development of the online news media community. In
addition, using Gephi 0.7’s Atlas algorithm, visualizations were generated to
illustrate the network parameters in each time period. The Atlas method is useful for
large networks because it reduces data to a general core-periphery structure to
generate visualizations. This algorithm is advantageous for large sets of data because
the core-periphery structure provides an clear visualization of the macro-level
attributes of the network.
Form Emergence
171
Hypotheses one through three test the emergence of new organizational
forms through a process of speciation. These hypotheses were tested using the
network data to examine the emergence of populations of blogs and online
communities. The generalized process, as discussed above, involves two steps: (1) a
baseline modeled is calculated utilizing Statnet and the ERGM package, and (2) a
longitudinal model is calculated to determine the rate of change utilizing RSiena.
Parameters are visually specified below in Table 8.
The baseline ergm model serves two purposes. First, density is modeled with
the edge parameter. Second, the baseline model provides a preliminary estimation of
the parameters that are expected to be significant at any given point in time.
Following Lusher and Ackland (2008), the baseline ergm model provides an anchor
against which findings from the longitudinal model can be compared. Lusher and
Ackland utilized the LPNet package to determine ergm parameters for both the
baseline and longitudinal model. Given the size of the dataset analyzed in this
research that is not possible. Thus, the ergm baseline model provides a directional
guideline for general behavior, but direct comparison is generally not possible unless
there is clear evidence that the models are equivalent.
Baseline models were thus calculated using the statnet ergm package in R.
The edge list data sets for each year were imported and converted to network format
using the network package. In order to compare year to year, the individual networks
were time stacked. This procedure merges together time series network data and
172
includes structure zeros for nonpresent nodes. For instance, if organization z is
present from 2000 onwards, but not present in 1998 and 1999, the matrices for 1998
and 1999 would include organization z, but it would be indicated with structural
zeros. One side effect of this procedure is that density will become overly low due to
the presence of the structural zeros. The side effects are relatively few, as most
longitudinal analyses are conducted by comparing effects, but it will be important to
note that densities are superficially low.
Each model was first estimated with the core parameters for density (edges),
and degree centrality (idegree – in degree centerality, odegree – out degree
centrality). In-degree and out-degree were modeled, as opposed to degree, because
the networks are directed. In addition, parameters were included for reciprocity
(mutual) and asymmetry (asymmetric). In certain cases where degree centrality
skewed significantly in a particular direction, a more robust measure of degree
centrality was included (gwidegree, gwodegree). Lastly, two attributes were included
in the ergm models: link strategy and organization type, as previously discussed.
Both were modeled as categorical node covariates, and modeled using the nodecov
parameter. This parameter measures the degree to which the difference in nodal
attributes between two nodes affects the creation of a link. A non-significant
parameter indicates no effect. Significance indicates that nodes will seek out nodes
of other categorical types. The nodematch parameter was also included; significance
of this parameter indicates that nodes will seek out other nodes of the same type.
173
To model a given year, parameters were added in a stepwise fashion until all
parameters were included (saturated model). For each step, models were compared
using the Akaike information criteria (AIC), (Arms, Huttenlocher et al., 2006)which
provides a measure for assessing the improvement in fit between two models. In
addition, model convergence for a given simulation is given through the goodness-
of-fit procedure, which provides a p-value comparing estimated models against the
actual data. For fitted parameters, p-values should be below 0.1 to indicate
convergence. Significance of a specific parameter is indicated in a given model by
the p-value generated in the maximum likelihood estimation.
In order to estimate the goodness-of-fit of the ergm models, the
mcmc.diagnostics routine was used: this routine simulates networks based on the
parameters that were calculated in a converged model. The degree distributions
(outdegree and indegree) for the simulations are then graphed and compared back
against the original values. In a good-fitting model, there should be general
alignment of the degree distributions. Table 8 provides a summary of the baseline
parameters, with visual illustrations.
174
Table 9: Visual representation of estimated ERGM parameters for Statnet (adapted
from Goodreau, 2007)
Legend
Network Tie
Node with attribute
Node without attribute
Graph Statistics
Parameter Illustration
Edges
ODegree
IDegree
Mutual
Asymmetric
Attribute Parameter Name
Node covariate (nodecov)
Node match (nodematch)
175
Subsequent to the baseline models, a longitudinal model was generated in
RSiena. The RSiena modeling procedure estimates the rate of change between two
periods. In addition to calculating a generic rate of change for every stage in the
model, the following parameters were included: out-degree effect to control for
density; reciprocity, to account for reciprocal ties; transitive triplets, which accounts
for network closure; and balance effect, to measures the preference of organizations
to connect with organizations that have similar network structures. In addition to the
above, attribute (covariate) effects were included for specific hypotheses.
With regards to the first research question, which explored the stage model of
community of evolution, it’s expected that both the baseline ergm model and rate
function of the longitudinal model will support the density analysis. Both parameters
should increase in parallel to changes in the density function.
Hypotheses 1 through 3
The first hypothesis predicted that new organizational forms are likely to link
to other organizations of the same type during emergence periods. In other words,
the hypothesis tests the degree to which a new organizational form emergence as a
relatively isolated cluster. During the period analyzed, both online communities and
blogs emerged as new organizational forms. To test for significance, two
dichotomized attributes, BLOG and SNS, were included where a value of 1 indicated
that a node was either a blog or a social networking site, respectively. Significance
will thus be indicated by a positive and significant value for both SameCov.BLOG
176
and SameCov.SNS. In the RSiena results, the Same Covariate (SameCov) parameter
models the likelihood that actors will form links to other actors with exactly the same
attribute value, and is particularly well suited for testing similarity effects for
categorical variables.
The second hypothesis predicted that during the emergence phase there
would be a proportionally small but significant number of links between a new
organizational form and existing specialist organizations. To test this hypothesis, a
new attribute, Speciate, was created. Organizations in this attribute were coded as a 4
if they were specialist newspaper organization, 1 if they were a blog or social
networking site, and 0 otherwise. A high-degree of difference is used in order to over
exaggerate the parameter. This was necessary is order to accurately detect the effect;
by overemphasizing the degree difference it is in turn easier to clearly detect the
resulting change. The only effect being tested is the creation of links, and thus this
modification will not affect other results or add any new bias. Here then, the
Covariate Alter effect was included. The Covariate Alter affect predicts that other
actors will link to a given actor with a high value for the tested attribute. A
significant and strongly positive parameter for CovAlter.Speciate would suggest that
there is a significant and positive likelihood that links will be created between
specialists and new organizational forms (a likelihood of links between blogs and
other sites – 1 and 0 – would generate a weakly positive parameter value).
177
The third hypothesis predicted that the number of links between a new
organizational form and existing members of the community will increase after the
density of the speciated form begins to increase. Thus, it is expected that as the
community moves back into a period of maintenance, the number of links between
blogs and other organizational forms, and social networking sites and other
organizational forms, will increase. Here again the Covariate Alter effect was used.
During these periods it is thus expected that both CovAlter.BLOG and
CovAlter.SNS will be significant. A positive magnitude for the parameter would
indicate a positive rate of tie formation between blogs and other organizations, and
between social networking sites and other organizations, respectively.
Organizational Birth Rates
Hypothesis four stated that the birth rate of online-based news forms would
be positively influenced by (a) an increase in media coverage of blogs and social
networking sites, (b) the formation of professional associations for online media
professionals and (c) positive legal recognition. The birth rate was based on the
recording of Web site founding dates in the results extracted from HistoryCrawl. The
organizational lifecycle of online organizations is rapid; as a result, given the one-
year time unit there is no need for a lagged birth rate. A positive effect from media
coverage, for instance, would be realized within the same year, given the ease with
which organizations can establish an online presence. The organizational count was
178
based on the total number of organizations detected within the community in a given
year.
Hypothesis five predicted that an increase in the number of organizational
links between existing organizations and new organizational forms would be
positively related to the birth of online-based news forms. In order to assess this
variable an additional variable, SPlinks, was created to record the number of links
that existed between new organizational forms and existing organizations. The
variable consists of a year-by-year count of the number of links between new
entrants and existing entrants. For the purposes of this analysis, an established
organization in time t was an organization that has existed two years prior, t – 2. This
was chosen as an arbitrary threshold in order to distinguish established organizations
from those that had existed one year or less.
Partially nested Poisson regressions were used to test the relative fit of
different levels of parameters. Individual parameter estimates where used to assess
the significance of the research hypotheses. The AIC measure was used to compare
different models and assess the optimal fit. The aggregate models accounting for
organizational birth rate are given in Table 12. Thus, the results of the nested Poisson
regressions were used to assess the results of hypotheses three and four.
The second research question sought to determine if certain legitimacy
effects are more likely than others to indicate the legitimacy of new organizational
forms. This question will be addressed through an examination of the significance
179
and magnitude of the results of the Poisson regression, and comparing those results
qualitatively against interview data regarding the development of new organizational
forms. The qualitative data was examined to see if interview subjects referenced or
emphasized particular legitimization processes more than others, and if those
referenced processes corresponded with what was found in the statistical analysis.
180
Table 10: Poisson Regression Models and Hypothesis Summary for Organizational Foundings
Ecological
Level
Theoretical
Mechanism
Parameter
Model
I
Model
II
Model
III
Model
IV
Model
V
Model
VI
Model
VII
Model
VIII
Environment
Number of
Internet Users
!
! !
Media Coverage
General Online
News (H4a)
!
!
Media Coverage
Blogs(H4a)
! !
!
Media Coverage
Social Networking
Sites (H4a)
! !
!
Professional
Associations
(H4b)
! ! !
! !
Legitimization
Legal Rulings
(H4c)
! ! !
! !
Community
Legitimization
- Speciation
Organizations
Links
(SPLinks) H5
! ! ! ! !
Established Forms ! !
! ! !
Population
Legitimization
– Density
Dependence
Emergent Forms ! !
! ! !
180
181
The Effects of Organizational Strategy on Failure
Hypotheses six through eight predict that organizational strategy will effect
the long-term survival of an organization. In order to assess the validity of these
hypotheses, a second Poisson regression was conducted to determine the effect of
strategy on organizational death. Data for this portion of the analysis were restricted
to newspaper organizations only. These were the only organizations for which a
substantial enough sample of organizational-level event data were available (from
the Editor and Publisher International Yearbook). Strategic failure of an
organizational was modeled as either a change in owner or a change in publisher.
Together, these were counted as events of organizational failure. In making the
previous assumption, it is possible that certain cases of organizational success were
falsely included as failures, but this is expected to have a minimal effect on the
results given the large sample size.
In order to assess hypothesis seven, and following Miner, Amburgey and
Stearns (1990), a variable was introduced to account for the effect that the number of
strategic changes had on failure rates. A strategic change was said to have occurred
when an organizations link strategy changed from one type to another and remained
in the new state for at least two years. This was an arbitrary threshold chosen in
accordance with the one-year time window used in the archival data collection; by
using a two-year period any strategic changes occurring in the first year of a strategy
182
were not included. In accordance with hypothesis seven, the more change initiatives
undertaken the lower the failure rate should be.
Hypothesis seven predicted that the adoption of aggressive strategic change
initiatives will decrease mortality rates in inertial organizations. In order to test this
hypothesis, the previously coded organizational strategies were included in the
Poisson regression. It follows that if aggressive linking strategies (blogification) lead
to decreases in mortality rates, then hypothesis seven would be supported.
Hypothesis eight predicted that the higher the number of institutional hyperlinks
maintained by online newspapers, the lower the mortality rate. To test this
hypothesis, the SPLinks variable was again included in the model, and is expected to
have a negative effect on organizational failure. Partially nested Poisson regressions
were again used to assess the predicted relationships. The aggregate models
accounting for organizational failure are given in Table 11, with the respective
hypotheses denoted.
183
Table 11: Poisson Regression Models and Hypothesis Summary for Organizational Failures
Ecological
Level
Theoretical
Mechanism
Parameter
Model
I
Model
II
Model
III
Model
IV
Model
V
Model
VI
Model
VII
Number of Internet
Users
! ! !
Environment
Environment Level
Event
! !
Interorg.
Linkages
Organizations Links
(SPLinks) (H8)
! ! !
Emergent ! ! !
Community
Legitimization -
Speciation
Established ! ! !
Blogification
(H7)
! ! !
Diversification
(H7)
! ! !
Experimentation
(H7)
! ! !
Organizational
Strategy (H7)
Isolation
(H7)
! ! !
Strategic Change
(H6)
Num. Organizations
Changing Strategy
! ! !
Population
Print-based
Newspapers (Online)
! !
183
184
The Effects of Linking Over Time
Hypotheses nine and ten test the effect that initial linking strategy has on the
formation of organizational linking over time for established organizations. These
hypotheses predict that early strategic action by a new organization will affect
interorganizational linking in later time periods. Both hypotheses were tested using
the longitudinal network models established in RSiena.
Hypothesis nine (a) predicted organizations that establish a high number of
interorganizational linkages to new populations during the emergence stage will
continue to have a high proportion of links out over time. This hypothesis was tested
with the variables CovEgo.blink and CovEgo.dlink. A positive significant value of
CovEgo.blink, would indicate that organizations prefer ties to create ties to other
organizations; likewise for CovEgo.dlink. Hypothesis nine (a) is supported if this
variable is significant over time
Hypothesis nine (b) predicted organizations that establish a low number of
interorganizational linkages to new populations during the emergence stage will
continue to have a low proportion of links out over time. This hypothesis was tested
with the variables CovEgo.elink and CovEgo.ilink. A negative significant value of
CovEgo.elink, would indicate that people prefer ties to create ties to other
organizations; likewise for CovEgo.dlink. Hypothesis nine (b) is supported if this
variable is negative and remains so over time. This would indicate that relatively few
links of this type exist, and even fewer exist over time.
185
Hypothesis ten (a) predicted organizations with a high number of
interorganizational linkages to new populations during periods of emergence will
have a high number of inlinks from new populations over time. This hypothesis was
tested with the variables CovAlter.blink and CovAlter.dlink. A positive significant
value of CovAlter.blink, would indicate that other people create ties to organizations
that have a blogging link strategy; likewise for CovAlter.dlink. Hypothesis nine (a) is
supported if this variable grows in magnitude over time.
Hypothesis ten (b) predicted organizations with a low number of
interorganizationl linkages to new populations during periods of emergence will
have a low number of inlinks from new populations over time. This hypothesis was
tested with the variables CovAlter.elink and CovAlter.ilink. A negative significant
value of NodeCov.elink, would indicate that people create ties to organizations with
experimentation strategies, but such links are few and far between; likewise for
CovAlter.dlink. Hypothesis nine (b) is supported if this variable is negative and
becomes more negative over time. This would indicate that relatively few links of
this type exist, and even fewer exist over time. Table 12 summarizes the RSiena
parameters modeled, matched with their respective hypotheses.
The results of the previously described hypotheses are given in the following
chapter.
186
Table 12: Visual representation of estimated ERGM parameters for RSiena
Legend
Network Tie in t
1
Network Tie in t
2
Node with attribute
Node without attribute
Graph Statistics
Structural Parameter Hypothesis Illustration
Edges
Mutual
Transitive Triplets
Balance Effect
Attribute-Based Parameter
SameCov.SNS
SameCov.Blog
Hypothesis 1
CovAlter.Speciate Hypothesis 2
CovAlter.Blog
CovAlter.SNS
Hypothesis 3
CovEgo.blink
CovEgo.clink
Hypothesis 9a
CovEgo.elink
CovEgo.ilink
Hypothesis 9b
CovAlter.blink
CovAlter.clink
Hypothesis 10a
CovAlter.elink
CovAlter.ilink
Hypothesis 10b
187
CHAPTER 6: RESULTS
In this chapter, the results of the previously outlined hypotheses are examined
in four stages. First, descriptive results are given describing the data extracted from
the Internet Archive with the HistoryCrawl Web crawler. Second, the results from
the longitudinal network analysis are presented, examining the emergence of
organizational forms. Third, the results of the Poisson regression are given, detailing
the effects of legitimation and strategic initiatives are examined. Lastly, the results of
the longitudinal network analysis are described, explaining the subsequent process of
transformation.
Descriptive Results
Figure 10 summarizes the general nature of the online news media
community, based on the data gathered with HistoryCrawl. The graph depicts the
number of organizations in the online news media community, measured as the
number of organizations with Web sites producing content associated with online
news. Each line in the graph represents the growth of a different type of
organization: the seven types of organizations are blogs, online only news sites,
newspaper Web sites, radio stations, television stations, online communities (forums.
online communities and social networking sites), and magazines. 1998 was the first
year for which stable data were available from the Internet Archive. As noted, the
data were both left and right censored based on the information available in the
archive. From the seed sample used, the online news media community in 1998
188
consisted of 326 organizations. That number grew considerably over the next three
years, more than doubling by 2001 to 874 organizations. Between 2001 and 2006,
the total number of organizations fluctuated, peaking again in 2006 at 885
organizations.
In general, television stations, radio stations, magazines and online only
newspapers accounted for fewer than 25 percent of news media organizations in any
given year. Blogs and online communities grew notably during two critical periods:
2000 – 2001 and 2004 – 2006. The first period of growth occurs in the year leading
up to the dot.com bubble burst, and the second period corresponds roughly with the
explosion in growth associated with the emergence of social networking sites and
Web 2.0 technology.
Figure 11 illustrates the birth of organizational Web sites in a given year for
the observed time period. As the diagram illustrates, there are two critical periods
during which organizational births increased significantly. In 1999 and 2000, there
were notable foundings of newspaper, online community and blog Web sites; in
2001, however, there is a clearly a significant increase in the number of blogs
founded (139). Following that year, the number of births was relatively flat until
2005, when there was a significant spike in the number of online community Web
sites (incl. social networking sites) that were founded (81).
189
Figure 10: Number of Organizations by Year
!
"!
#!!
#"!
$!!
$"!
%!!
#&&' #&&& $!!! $!!# $!!$ $!!% $!!( $!!" $!!) $!!*
!"#$
+,-./ 01,213401,54637/ 637/8983:/ ;9<2- =3,3>2/2-1 ?3@4A2B3/ C9./
190
Figure 11: Organizational Births by Year Based on new Web sites
!
"!
#!
$!
%!
&!!
&"!
!
&$!
&''' "!!! "!!& "!!" "!!( "!!# "!!) "!!$ "!!*
!"#$
+,-./ 01,213401,54637/ 637/8983:/ ;9<2- =3,3>2/2-1 ?3@4A2B3/ C9./
191
Figure 12: Changes in Newspaper Ownership, Publisher and Editor
191
192
A substantial sample of each type of organization was collected, although
given foundings and deaths of Web sites, the numbers clearly fluctuated in any given
year. Deaths were relatively few in numbers (92), especially when compared to
foundings across all years (1120). This is due in part to the fact that even if an
organization fails, or ceases to operate actively in an online environment, Web sites
can remain for a substantial period of time. A summary of the organizational
breakdown in the final 2007 sample is given in Table 13. Newspapers were a large
population (487), and this was expected given that the seed sample consisted almost
exclusively of newspaper Web sites. In addition, blogs (269) and online communities
(192) combined accounted for 47 percent of all Web sites in the 2007 sample. The
table also provides the average Alexa pageview rank for each category. This gives a
comparative number for the popularity of each category. As the numbers show,
online only news sources were actually the most popular category, followed closely
by blogs and online communities. Newspapers were somewhat popular, ranking
fourth out of the sample group, but the Web sites for radio stations, televisions
stations and magazines were generally not drivers of audience traffic. Lastly, the
chart provides the mean number of interorganizational links created by each type of
organization (rounded to the nearest whole number). Blogs, online only news sources
and online communities are clearly prolific linkers.
193
Table 13: Organizational Breakdown by Type with Descriptive Attributes (based on
final network – 2007)
Organization Type
Number of
Organizations
Average Alexa
Pageview Rank
Average Number of
Interorganizational Links
(rounded)
Blogs 269
2,480,112
6
Online Only News
Sources
60
2,731,232
4
Newspapers 487
1,929,415
3
Radio Stations 46
1,003,886
2
Television Stations 15
1,023,308
2
Online
Communities
(Forum, SNS, etc.)
192
2,476,646
5
Magazines 60
989,512
3
Drilling further down into the data, Table 14 provides a breakdown of the
attributes of the newspaper population. This table provides a summary for the broad
population of newspaper Web sites that existed within the population for the given
time period. A single organization during this period may have had multiple Web
sites, as many organizations experimented with different online incarnations. Thus,
the numbers given below actually exceed the number of organizations. Within the
newspaper community, there were more specialist Web sites (367) than generalist
Web sites (120). In addition, the majority of newspaper organizations utilized initial
linking strategies that were generally conservative, with 162 employing an isolation
194
strategy and 176 employing a diversification strategy. On the other hand, far fewer
adopted diversification (72) or blogification (77) as initial strategies.
Ownership for the sample of newspaper Web sites was highly skewed
towards conglomerates: 422 newspapers in the sample were subsidiaries of a multi-
newspaper parent company or media conglomerate, while only 65 were independent
newspaper Web sites. In addition, there was notable change in the population for the
time period sampled. From 1998 through 2008, there were 905 changes in editors-in-
chief, 867 changes in publisher and 208 changes in ownership. Despite the changes
in management structure and ownership, the newspaper population itself was clearly
well established offline. The average age of print newspapers in the population was
114 years. On the other hand, the average age of a print newspapers’ Web site was
nine years.
195
Table 14: Breakdown of Newspaper Population (for all newspapers)
Variable Count
Generalist vs. Specialist
Generalist 120
Specialist 367
Initial Linking Strategy
Isolation 162
Experimentation 176
Diversification 72
Blogification 77
Ownership Type
Subsidiary 422
Independent 65
Changes in Editor (all years) 905
Changes in Publisher (all years) 867
Changes in Ownership (all years) 208
Average Age (offline) 114 years
Average Age (online) 9 years
The above data summarizes the overall composition of the network of
organizational forms included in the sample data extracted from the Internet Archive.
These data provide a substantive sample of the population of United States news
media. For instance, the United States Statistical Abstract estimates that there were
1,408 printed newspapers in 2008; this sample captured 15 percent of those
newspapers, assuming that each one had an online presence in that year.
Furthermore, the sample was somewhat restricted by the qualifier that organizations
196
were only included if there were three or more links to another organization within
the crawl. Prefilter, there were 672 newspaper Web sites in the sample, accounting
for 48 percent of newspapers; but including all of these sites would not have
accurately represented the behavior of linking and alliances between these
organizations. Additionally, the sample underepresents the number of radio,
television and magazine Web sites. It is hard to accurately determine the number of
organizations for each of these populations, but according to the United States
Census there are roughly 1600 digital television stations in the United Sates. The low
numbers represented in this sample suggests that these organizations are relatively
unconnected with the larger news media community.
There was significant growth and disruption in the community during the
time period stretching from 1998 to 2008, as illustrated by the data describing the
newspaper population. The first research question thus sought to extend the stage
model of community development to the evolution of the online news media
community. Densities (!) were calculated for each time period, and the results are
given in Table 15. Using the UCINET “Compare Densities > Paired Densities”
routine, t-tests were calculated comparing densities to the prior time period. The
results are also given in the table below. As the results show, calculated densities
were low, but this is in line with what would be expected for an online network.
Online networks are scale-free, and the number of nodes within such a network can
grow at a pace that far exceeds the growth of ties. Furthermore, the number of
197
potential ties is far greater than what is supported by the community itself.
Additionally, as the results show, the densities decrease for the first three years, and
then fluctuate between 0.0016 and 0.0028 thereafter. As the corresponding t-tests
indicate, the change in densities during this time was only significant from 1998 to
1999 (t
1998-1999
= -3.903*). The decrease in density in 2000 corresponds with a period
of community wide growth; the number of nodes grew significantly but the number
of ties did not grow in turn, and the change in density was not significant. The
decrease in network density from 2001 (!
2001
= 0.0023) to 2002 (!
2002
= 0.0015) is
again significant (t
2001-2002
= -1.120*) as is the growth from 2002 (!
2001
= 0.0015) to
2003 (!
2003
= 0.0024, t
2002-2003
= -1.540*). New entrants emerge at the periphery and
thus may establish very few ties to the core of the network. Furthermore, the full
network includes imputed organizations and the densities are thus artificially small.
The results of the initial density analysis do not, therefore, allow for a clear
determination of the stages of network growth.
198
Table 15: Network Densities and Main Component Density 1998 - 2007
Year Density t-value
Main
Component
Density
t-value
!
1998
0.0084 - 0.0422 -
!
1999
0.0034 -3.903* 0.0411 -0.221
!
2000
0.0016 -2.941 0.0314 -0.462
!
2001
0.0023 1.842 0.0486 0.312*
!
2002
0.0015 -1.120* 0.088 2.320*
!
2003
0.0024 1.540* 0.098 0.122*
!
2004
0.0023 0.231 0.095 -0.098
!
2005
0.0022 -0.024 0.106 0.118*
!
2006
0.0026 0.182 0.113 0.109
!
2007
0.0028 0.105 0.112 0.004
*t(5000), p < 0.05 with bootstrap sample
Therefore, in addition to the initial density calculation, densities were also calculated
for the weak main components in each year. The UCINET main component routine
was used to extract the nodes and relations that constitute the largest component of
the network in a given year. This extracts the core of the network, and in turn
densities were calculated for the main component of the network in each year. As the
network grows and becomes more stable, it would be expected that there would be
more competition at the core of the community. The results of the main component
analysis show positive and significant growth in density from 2000 through 2003
(t
2000-2001
= 0.312*, t
2001-2002
= 2.320*, t
2002-2003
= 0.122*).
199
As noted in Bryant and Monge (2008), during the emergence phase the
community is in relative chaos and there is not a clear sense of legitimacy. Looking
at the main component, one would thus expect a relatively weak component during
this period. As a community moves into a period of maintenance it is expected that
there will be an increase in competition for resources. The density of the main
component will thus increase as more organizations compete and legitimacy is
established. Finally, self-sufficiency occurs when a high proportion of potential ties
are realized, and there is a stable and balanced flow of resources.
The original density thresholds calculated by Bryant and Monge (Bryant &
Monge, 2008) were 0.50 for the transition from emergence to maintenance, and 0.75
for the transition from maintenance to self-sufficiency; the growth curve was
previously illustrated in Figure 1. That said, their original study examined the
children’s’ television community, which was a much more established, traditional
community of organizations. Given that the online news community operates in a
scale-free online environment, and that densities were measured against the online
network as a whole, the threshold densities are much lower. The originally
approximated curve should still apply, albeit at lower densities; thus, the density
growth curve is given in Figure 13, utilizing the densities for the main component in
an attempt to accurately measure the primary competition for resources in the
network. As the chart illustrates, there is continual growth in the main component
density through 2003, with an inflection in growth in 2004. From 2005 through
200
2007, growth stabilizes and the community enters into a general period of
maintenance. Ultimately, more data are needed to conclusively show that the
community has indeed moved into a maintenance period, but for the purposes of this
analysis, emergence is said to have occurred from 1998 to 2004, and maintenance is
the period from 2005 to 2008. These periods will be used in the analysis from this
point forward.
As further illustration, network diagrams are shown in Figure 14 – 17,
illustrating the broad network configuration in 1999, 2001, 2003 and 2005. The
diagrams show the interaction between previously existing organizational
populations (newspapers, radio stations, television stations and magazines, shown in
red) and emergent organizational forms (blogs, online only newspapers and online
communities, shown in blue). The full communities are shown. From 1999 to 2001,
there is a dramatic increase in the in density of the community structure, and a
notable increase in the interaction between existing and emergent organizations.
From 2001 to 2003, the influx of new organizational forms continues, but the
community begins to stabilize in terms of density. Finally, Figure 17 shows the
community in 2005: the influx of new organizational forms slowed by this point, and
although the density of the community continued to increase it was not a major
change.
201
Figure 13: Network and Main Component Densities by Year
Theoretical
Form
201
202
Figure 14: Network Diagram – 1999
Note: emergent organizational forms are in blue (blogs, online communities and
online only news organizations); established organizational forms are in red
(newspapers, radio stations, television stations and magazines). Size represents
degree centrality.
203
Figure 15: Network Diagram – 2001
Note: emergent organizational forms are in blue (blogs, online communities and
online only news organizations); established organizational forms are in red
(newspapers, radio stations, television stations and magazines). Size represents
degree centrality.
204
Figure 16: Network Diagram – 2003
Note: emergent organizational forms are in blue (blogs, online communities and
online only news organizations); established organizational forms are in red
(newspapers, radio stations, television stations and magazines). Size represents
degree centrality.
205
Figure 17: Network Diagram – 2005
Note: emergent organizational forms are in blue (blogs, online communities and
online only news organizations); established organizational forms are in red
(newspapers, radio stations, television stations and magazines). Size represents
degree centrality.
206
Longitudinal Network Analysis
The network structure and network dynamics of the online news media
community was examined in a two stage process: the baseline network configuration
was analyzed using the Statnet ergm routine, and the longitudinal change in the
network was analyzed using RSiena. The results of the baseline Statnet ergm models
are shown in Table 16 and Table 17. Table 16 shows the AIC selection criteria for
each model and Table 17 shows the model results of the Statnet baseline model.
Models were estimated for each year of data, from 1998 through 2007. Seven
different models were analyzed, based on the previously discussed parameters. The
first three models are structural baselines: the first modeling edges, the second edges
and reciprocity (mutual), and the third, edges, reciprocity and the likelihood of
unbalanced relationships (asymmetric). Asymmetry was only included in subsequent
models if it improved the fit from Model 2 to Model 3. This was only the case in
2003, 2005, and 2007, and was included to improve the general model fit. Model 4
and Model 5 tested in-degree and out-degree distribution models. Model 4 used the
idegree and odegree parameters, whereas Model 5 used the gwidegree and
gwodegree parameters. The latter accounts for widely distributed degree
distributions more effectively. Finally, Model 6 and Model 7 added in the attribute
parameters, as well as a parameter to account for isolates.
Isolates were significant in almost every model, but this was expected
because the matrices for each year were stacked. Thus, the isolates parameter
207
captures, in part, the structural zeros that were introduced in the stacking procedure.
As the results in Table 16 show, the idegree and odegree parameters were sufficient
to capture degree distribution for the models from 1998 through 2003. This is shown
by the general improvement in model fit shown by the decrease in AIC, as well as
the subsequent model significance shown in Table 17. For example, in 2003 the
included of the idegree and odegree parameter improves the overall model fit
(AIC
MODEL3
= 6481, AIC
MODEL4
= 6463), but the inclusion of gwidegree and
gwodegree actually generates a worse fit (AIC
MODEL5
= 6580). This indicates that the
inclusion of idegree and odegree improves the model fit, as compared to the
gwidegree and gwodegree parameters. From 2003 onward, the degree distributions
were more efficiently modeled using the weighted degree distribution parameters.
Parameters were estimated for each year and each model; Tables 16 and 17
summarize the results for optimized model for each year.
208
Table 16: Statnet Baseline Models 1998 – 2007: AIC Indices
Model 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Model 1: edges 4983 6728 6447 6546 9026 6978 7570 8534 6708 6964
Model 2: edges + mutual 4973 6285 5840 5839 8898 6717 7262 8213 6141 6605
Model 3: edges + mutual + symmetric 4976 6381 4924 5562 8601 6481 7337 7627 6165 6702
Model 4: Model 2 + idegree + odegree 4962 5908 4747 5574 8573 6463 7117 7822 6122 6188
Model 5: Model 2 + gwidegree +
gwodegree
4972 6108 4830 3408 8671 6580 6424 7430 5823 5756
Model 6: Model 4 + nodecov.type +
nodecov.linkstrat + isolates
4979 5815 4561 3200 8136 6250 na na na na
Model 7: Model 5 + nodecov.type +
nodecov.linkstrat + isolates
na na na na na na 6404 7424 5365 5551
208
209
Table 17 provides the baseline parameters for the online news media
community from 1998 through 2002. Because stacked parameters were used, it is
possible to compare across years although it is not possible to determine significance
at this point. As previously discussed, the Statnet model provides a baseline for
comparison, but does not allow for a longitudinal comparison. Rather, the results can
be used to gain directional insight into the network behavior. In line with the density
calculations illustrated in Figure 13, the edge parameter shows that density decreased
in 1999 and 2000, but began to increase in 2001 and 2002. Reciprocity is present and
significant in all years, although the effect of reciprocity becomes more significant as
the community ages. The idegree and odegree parameters are significant in 2002 and
2003, and although they are close to zero in magnitude indicating generally small
degree distributions. Lastly, the effect of that organization type and organizational
link strategy has in each year is given by the nodecov.type and nodecov.linkstrat
parameters. Both parameters are significant for all years, but are generally small in
magnitude meaning that directionally they are difficult to interpret. Nodecov.type is
generally negative, indicating that nodes of different types are less likely to form
links, but the magnitude is too small to be conclusive. Likewise, nodecov.linkstrat is
generally negative but close to zero magnitude, again indicating a generally small
effect.
210
Table 17: Statnet Baseline Models 1998 – 2002: Fitted Parameters
Parameter 1998 1999 2000 2001 2002
edges -5.443***
(0.011)
-7.155***
(0.162)
-7.069***
(0.120)
-6.418***
(0.022)
-6.203***
(0.108)
mutual 1.781*
(0.743)
5.358***
(0.944)
3.321***
(0.592)
4.743***
(0.259)
6.328***
(0.402)
asymmetric - - - - -
idegree 0.251
(0.980)
-0.671***
(0.100)
-0.186
(0.022)
-0.174***
(0.004)
-0.644***
(0.073)
odegree 0.561
(0.510)
-0.459
(0.284)
-0.535
(0.342)
0.045***
(0.009)
-0.917***
(0.080)
gwidegree - - - - -
gwodegree - - - - -
nodecov.type 0.042
(0.011)
-0.122***
(0.018)
-0.102***
(0.011)
-0.083***
(0.001)
-0.142***
(0.012)
nodecov.linkstrat -0.061***
0.001
-0.049***
(0.013)
-0.026**
(0.003)
-0.057***
(-.001)
-0.057***
(0.010)
isolates 1.098***
(0.132)
1.452***
(0.241)
1.311***
(0.112)
1.013***
(0.125)
0.851***
(0.050)
note: 0.001*** 0.01** 0.05*
note: Standard Errors are provided in parentheses
Table 18 provides the baseline parameters for the online news media
community from 2003 through 2007. The density remains low through 2005, with
some fluctuation, and begins to increase again in 2006 and 2007. This indicates that
the number of ties began to increase as the community moved into a period of
maintanence. Directionally, this provides more support for the staged model of
community development. Reciprocity is significant in all years, and positive,
indicating that reciprocal relationships are present throughout the network. In 2003,
2004 and 2005 the asymmetric parameter is present and significant. The parameter
value, however, is generally negative (asymmetric
2003
= -0.223, SE = 0.015;
211
asymmetric
2005
= -2.656, SE = 0.283; asymmetric
2007
= -3.588, SE = 0.435) indicating
that although there are unbalanced relationships, they do not occur frequently. The
presence of negative and significant weight in-degree and out-degree parameters in
2004, 2005, 2006 and 2007 suggests that as the community ages the degree
distributions become less concentrated. The large value of these parameters on the
whole indicates that as the network grows there is a larger number of high-degree
nodes, although they are still relatively few in number. The effect of organization
type on link creation is again small, but significant (nodecov.type
2003
= -0.164, SE =
0.019; nodecov.type
2004
= -0.123, SE = 0.010; nodecov.type
2005
= -0.165, SE = 0.014;
nodecov.type
2006
= -0.074, SE = 0.016; nodecov.type
2007
= -0.092, SE = 0.024). This
would, preliminarily, suggest that organizations prefer to link with other
organizations of the same type, although directionally the parameter becomes less
negative as time passes. Lastly, link strategy has a relatively small effect throughout,
and is not significant in 2006 and 2007 as indicated by the model values.
212
Table 18: Statnet Baseline Models 2003 – 2007: Fitted Parameters
Parameter 2003 2004 2005 2006 2007
edges -6.360***
(0.157)
-4.878***
(0.076)
-5.532***
(0.260)
-4.602***
(0.159)
-4.571***
(0.585)
mutual 5.897***
(0.410)
5.472***
(0.242)
5.120***
(0.119)
6.636***
(0.486)
5.644***
(0.352)
symmetric -0.223***
(0.015)
- -2.656***
(0.283)
- -3.588***
(0.435)
idegree -0.626***
(0.097)
- - - -
odegree -0.795***
(0.092)
- - - -
gwidegree - -0.580***
(0.019)
-0.165***
(0.155)
-0.689***
(0.027)
-5.843***
(0.224)
gwodegree - -5.887***
(0.002)
-2.640***
(0.155)
-4.327***
(0.136)
-0.640***
(0.115)
nodecov.type -0.164***
(0.019)
-0.123***
(0.010)
-0.165***
(0.014)
-0.074***
(0.016)
-0.092***
(0.024)
nodecov.linkstrat -0.047***
(0.013)
-0.036***
(0.008)
-0.022*
(0.010)
-0.014
(0.011)
0.019
(0.019)
isolates 0.980***
(0.023)
0.870***
(0.095)
1.023**
(0.103)
0.801***
(0.043)
0.910***
(0.082)
note: 0.001*** 0.01** 0.05*
note: Standard Errors are provided in parentheses
In aggregate, the baseline model shows that the community as a whole
increases in density throughout the time period analyzed. Furthermore, reciprocity is
an important structural configuration in the network. In-degree and out-degree
distributions become more complicated as time passes. On the other hand it is hard to
infer much meaning from the attribute parameters in this model. Furthermore, statnet
examines each network in turn and the large network size results in an overemphasis
on statistical significance. Here then, the RSiena models provide a more accurate
213
estimation of network change by analyzing the change from one period to another,
thus overcoming any overestimation issues.
Hypotheses 1 to 3
RSiena was used to estimate the longitudinal change in the online news
media community network from 1998 through 2007. The first step in obtaining
RSiena results was to establish a converged baseline model, much like in the case of
statnet. The baseline model was obtained by adding in parameters in a stepwise
fashion, testing fit, and removing parameters that do not converge. The converged
model of the news media community is presented in Table 19. Moderate
convergence is generally indicated by having all t-ratios less than 0.3. Good
convergence is generally indicated by having all t-ratios less than 0.2. Excellent
convergence is indicated when all t-ratios are less than 0.1 (Snijders, 2009). The t-
ratio is a fairly simple measure of convergence, and is calculated by dividing the
average value by the standard deviation. As the results show, the converged model
for the news media community generally has good convergence. The major
exception is the rate parameters for period 2 (1999 – 2000) and period 3 (2001 –
2002). The rate parameters indicate averaged values for each parameter, as
calculated by average of the simulated values generated by the estimated parameters
(as compared to the observed values found in the data). The convergence test is a
check to insure that the resulting model accurately fits against the estimated data, and
in turn provides an accurate representation of the community. The deviation in
214
period 2 and period 3 primarily indicates that there was significant change in the
network during these periods. Although ideally the t-ratios would be closer to
converged, the deviation in rate parameters is acceptable given the generally good fit
of the model overall (Snijders, 2009).
215
Table 19: Convergence Diagnostics for RSiena Model 1998 - 2007
Parameter Average Std. Dev. t-ratio
Rate Parameters
ß
1998 – 1999
1.512 6.102 0.248
ß
1999 – 2000
3.510 8.950 0.392
ß
2000 – 2001
4.930 9.035 0.546
ß
2001 – 2002
1.437 7.356 0.195
ß
2003 – 2004
0.840 8.230 0.102
ß
2004 – 2005
1.282 9.103 0.141
ß
2005 – 2006
1.265 8.015 0.158
ß
2006 – 2007
1.980 9.320 0.212
Structural Parameters
Edges -1.890 9.360 0.202
Mutual 0.901 8.502 0.106
Transitive Triplets 0.687 6.305 0.109
Balance Effect 0.722 5.980 0.121
note: moderate convergence is indicated by t-ratios < 0.3;
good convergence is indicated by t-ratios < 0.2;
excellent convergence is indicated by t-ratios < 0.1
Thus, Table 20 gives the results for the baseline estimation of change in the
community from 1998 through 2007. The results for the model estimation show the
parameter estimate and the standard error. Significance (p < 0.05) is indicated if the
t-value is calculated to be greater than 2.0. The rate parameters represent the
estimated number of changes that a given organization makes between two time
periods. The structural parameters indicate the types of configurations that are likely
to develop between two time periods. The rate parameters, ß, show that the most
notable change is in period 2 (ß
1999 – 2000
= 2.010, SE = 1.142), period 3 (ß
1999 – 2000
=
3.450, SE = 1.235) and period 5 (ß
1999 – 2000
= 2.401, SE = 0.216). Significance is not
given for periodical change; the period parameters are critical to network structure
and thus significance is not relevant (change is a given in RSiena models). More
216
importantly, the edge parameter (edge = -6.231, SE = 0.036) indicates that the
network as a whole is generally low in density, and the mutual parameter (mutual =
4.250, SE = 0.195) indicates that there are a significant and positive number of
reciprocal relationships. The transitive triplets effect (Transitive Triplets = 1.235, SE
= 0.210) suggests that organizations in the network will seek triadic closure over
time, and the balance effect is significant and positive (Balance Effect = 1.366, SE =
0.130) indicating that organizations seek structurally equivalent relationships over
time (seeking out other organizations that are in structurally equivalent positions
within the community).
Table 20: Estimation for RSiena Model 1998 – 2007
Parameter Estimates SE
Rate Parameters
ß
1998 – 1999
1.098 0.189
ß
1999 – 2000
2.010 1.142
ß
2000 – 2001
3.450 1.235
ß
2001 – 2002
1.517 0.127
ß
2003 – 2004
2.401 0.216
ß
2004 – 2005
1.654 0.109
ß
2005 – 2006
1.035 0.836
ß
2006 – 2007
1.011 0.055
Structural Parameters
Edges -6.231* 0.036
Mutual 4.250* 0.195
Transitive Triplets 1.235* 0.210
Balance Effect 1.366* 0.130
* indicates significance at p < 0.05
The structural parameters provided by RSiena provide a baseline model for
examining changes in the network as a result of nodal covariates. In order to model
the effects of attributes, the nodal attributes were estimated in two stages. The first
217
model included the covariates corresponding with the first, second and third
hypotheses. The second model included the covariates associated with hypotheses
nine and ten. Both models were based on the same structural model; the results for
hypotheses one, two and three are presented in Table 21.
Hypothesis one predicted that during emergence, new organizational forms
are likely to link to other organizational forms of the same time. The emergence
period, based on research question one, was defined as the period ranging from 1998
through 2004. During this emergence period, the SameCov.SNS is generally not
significant, while the SameCov.Blog parameter is positive and significant. The
SameCov.SNS parameter measures the likelihood that between two periods links
will be created between two organizations that are online communities or social
networking-oriented Web sites and is significant in two periods during the
emergence phase (SameCov.SNS
98-99
= 1.201, SE = 0.210; SameCov.SNS
99-00
= 1.003,
SE = 0.209). The SameCov.Blog parameter measures the likelihood that blogs will
link to other blogs, and is significant and positive in all time periods
(SameBlog.SNS
98-99
= 1.003, SE = 0.104; SameBlog.SNS
99-00
= 1.103, SE = 0.140;
SameBlog.SNS
00-01
= 0.980, SE = 0.158; SameBlog.SNS
01-02
= 0.880, SE = 0.146;
SameBlog.SNS
02-03
= 1.002, SE = 0.201; SameBlog.SNS
03-04
= 1.022, SE = 0.280).
Combined, these two parameters indicate that social networking sites are likely to
link to one another very early in the emergence period, and blogs are likely to link to
one another through the emergence phase. Both blogs and social networking sites
218
were emergent organizational forms. Support for social networking sites is moderate,
but support is strong for blogs, and thus hypothesis one is largely supported.
Hypothesis two predicted that during emergence, there will be a
proportionally small, but significant, number of links between a new organizational
form and existing specialist organizations. The variable Speciate was created to
measure this hypothesis, and its effect is measured by the variable
CovAlter.Speciate. The variable is positive, but only significant in half of the time
periods: 1999-2000 (CovAlter.Speciate
1999-2000
= 2.336, SE = 1.603), 2001-2002
(CovAlter.Speciate
2001-2002
= 3.403, SE = 1.506) and 2002-2003
(CovAlter.Speciate
2002–2003
= 3.503, SE = 1.608). A positive parameter between 1 and
4 indicates that organizational linkages are created between emergent organizational
forms and specialist newspapers. Particularly from 2001 to 2003, the parameter’s
magnitude indicates that the mechanism of linking to specialists occurs. That the
parameter is not significant in all periods indicates that there is only moderate
support for this mechanism. Thus, hypothesis two is partially supported.
Hypothesis three predicted that links between a new organizational form and
the existing community will increase after the density of the speciated form begins to
increase. This hypothesis was measured by the CovAlter.Blog and CovAlter.SNS
parameters, and it was predicted that these parameters would be significant after the
density of the speciated form began to increase (thus, in the maintenance period from
2004 – 2007). The results of the longitudinal network analysis show strong support
219
for H3. CovAlter.Blog is significant and positive in all three periods: 2004 – 2005
(CovAlter.Blog
2004–2005
= 1.306, SE = 0.549), 2005 – 2006 (CovAlter.Blog
2005–2006
=
1.103, SE = 0.502) and 2006 – 2007 (CovAlter.Blog
2006–2007
= 1.110, SE = 0.512).
CovAlter.SNS is also significant and positive in all three periods: 2004 – 2005
(CovAlter.SNS
2004–2005
= 1.601, SE = 0.720), 2005 – 2006 (CovAlter.SNS
2005–2006
=
1.610, SE = 0.701) and 2006 – 2007 (CovAlter.SNS
2006–2007
= 1.662, SE = 0.709). The
positive and significant parameter indicates a positive rate of tie formation between
blogs and other organizational forms, and social networking sites and other
organizational forms. The CovAlter.Blog parameter is also significant from 2001 to
2004 (CovAlter.Blog
2001-2002
= 1.222, SE = 0.601; CovAlter.Blog
2002–2003
= 1.126, SE
= 0.554; CovAlter.Blog
2003–2004
= 1.306, SE = 0.549). The CovAlter.SNS parameter is
significant from 2002 – 2004 (CovAlter.SNS
2002–2003
= 1.430, SE = 0.701;
CovAlter.SNS
2003–2004
= 1.603, SE = 0.714). This indicates that in addition to the
above results, the rate of tie formation between blogs and other organizational forms
began to increase significantly as early as 2001; for social networking sites, the
growth of ties lagged somewhat, increasing beginning in 2002. These results further
support the proposed hypothetical relationship.
220
Table 21: Estimation for RSiena Model 1998 – 2007 (Attributes for H
1
, H
2
, H
3
)
Attribute-Based
Parameter
98-99 99-00 00-01 01-02 02-03 03-04 04-05 05-06 06-07
SameCov.SNS
1.201*
(0.210)
1.003*
(0.209)
1.023*
(0.450)
1.053*
(0.499)
1.044*
(0.518)
1.020
(0.555)
0.980
(0.546)
0.889
(0.543)
1.003
(0.540)
H
1
SameCov.Blog
1.003*
(0.104)
1.103*
(0.140)
0.980*
(0.158)
0.880*
(0.146)
1.002*
(0.201)
1.022*
(0.280)
1.102*
(0.208)
1.115*
(0.257)
1.002*
(0.239)
H
2
CovAlter.Speciate
2.301
(1.536)
2.336*
(1.603)
2.450
(1.546)
3.403*
(1.506)
3.503*
(1.608)
2.225
(1.607)
2.103
(1.668)
2.210
(1.708)
2.413
(1.700)
CovAlter.Blog
1.002
(0.890)
1.030
(0.808)
1.020
(0.709)
1.222*
(0.601)
1.126*
(0.554)
1.306*
(0.549)
1.060*
(0.480)
1.103*
(0.502)
1.110*
(0.512)
H
3
CovAlter.SNS
1.103
(0.820)
1.302
(0.790)
1.210
(0.774)
1.109
(0.702)
1.430*
(0.701)
1.603*
(0.714)
1.601*
(0.720)
1.610*
(0.701)
1.662*
(0.709)
* indicates significance at p < 0.05
221
Table 22: Poisson Regression Results for News Community Birth Rate
Ecological
Level
Theoretical
Mechanism
Parameter Model
I
Model
II
Model
III
Model
IV
Model
V
Model
VI
Model
VII
Model
VIII
Model
VIIa
Intercept 4.180*
(-
0.001)
2.315*
(0.001)
2.305*
(0.003)
3.369*
(0.004)
3.359*
(0.041)
3.294*
(0.046)
2.163
(0.054)
2.133*
(0.047)
2.163
(0.064)
Environment Number of
Internet Users
-0.001*
(0.029)
0.016*
(0.001)
0.020*
(0.004)
0.016
(0.089)
Media Coverage
General Online
News (H4a)
0.243*
(0.002)
0.375*
(0.005)
Media Coverage
Blogs(H4a)
0.434*
(0.002)
0.384*
(0.004)
0.533*
(0.002)
0.533
(0.143)
Media Coverage
Social
Networking
Sites (H4a)
0.095
(0.172)
0.464
(0.052)
0.454
(0.056)
0.454
(0.171)
Professional
Associations
(H4b)
0.098*
(0.005)
0.078*
(0.006)
0.256*
(0.002)
0.093*
(0.001)
0.107*
(0.000)
0.093*
(0.039)
Legitimization
Legal Rulings
(H4c)
0.123
(0.127)
0.146*
(0.008)
0.148*
(0.002)
0.071*
(0.007)
0.040
(0.127)
0.071
(0.108)
Community
Legitimization
- Speciation
Organizations
Links
(SPLinks) H5
0.007
(0.826)
0.012*
(0.001)
0.003
0.364
0.006*
(0.002)
-0.001
(0.002)
0.006
(0.089)
Emergent
-0.002*
(0.040)
-0.009
(0.060)
0.007*
(0.001)
-0.005
(0.060)
0.004*
(0.047)
-0.005
(0.165)
Population Legitimization
– Density
Dependence Established
0.025*
(0.002)
0.026*
(0.002)
0.011*
(0.001)
0.023*
(0.006)
0.021*
(0.002)
0.024*
(0.030)
222
Table 23: Poisson Regression Results for News Community Birth Rate – Model Fit
Model AIC
Null Deviance
(df)
Residual Deviance (df) Estimated r
2
Model 1: Environment Predictors 1085.1 992.70 (17) 987.95 (16) 0.31
Model 2: Legitimization +
Speciation + Density
296.35 992.70 (17) 189.16 (11)
0.73
Model 3: Media Breakout +
Speciation + Density
281.83 992.70 (17) 172.64 (10)
0.75
Model 4: Legitimization +
Speciation
480.55 992.70 (17) 375.36 (13)
0.53
Model 5: Speciation 889.61 992.70 (17) 891.42 (16) 0.68
Model 6: Density 480.61 992.70 (17) 381.42 (16) 0.56
Model 7: Saturated (Media
Breakout)
169.1 992.70 (17) 57.91 (9)
0.79
Model 8: Saturated (Media) 180.83 992.70 (17) 71.64 (10) 0.77
Model 7a: Quasi-Poisson NA 992.70 (17) 57.91 (9) -
223
Table 24: Poisson Regression Results for News Community Failure Rate
Ecological
Level
Theoretical
Mechanism
Parameter
Model
I
Model
II
Model
III
Model
IV
Model
V
Model
VI
Model
VII
Model
VIIa
Intercept 3.899*
(0.007)
3.843*
(0.009)
1.399*
(0.226)
2.005*
(0.013)
1.876*
(0.154)
3.881*
(0.034)
1.001
(0.121)
1.001
(0.491)
Environment Number of
Internet Users
0.008*
(0.001)
0.008*
(0.001)
0.019*
(0.005)
0.019*
(0.012)
Environment
Level Event
0.229*
(0.004)
0137
(0.107)
0.237
(0.375)
Interorg.
Linkages
Organizations
Links (SPLinks)
(H8)
0.008*
(0.002)
0.013*
(0.001)
0.008*
(0.002)
0.008
(0.184)
Emergent
-0.019*
(0.004)
-0.009*
(0.001)
-0.013*
(0.006)
-0.013
(0.111)
Community
Competition
for Resources
Established
0.005
(0.074)
0.029*
(0.002)
-0.083*
(0.010)
-0.083
(0.099)
Blogification
-0.222*
(0.003)
-0.210*
(0.011)
-0.243*
(0.010)
-0.243*
(0.043)
Densification
-0.004
(0.067)
-0.004*
(0.001)
-0.004*
(0.018)
-0.004
(0.102)
Experimentation
-0.014*
(0.003)
-0.018*
(0.001)
-0.005*
0.007
-0.005
0.154
Org. Strategy
(H8)
Isolation
0.098*
(0.001)
0.109*
(0.010)
0.120*
(0.002)
0.120*
(0.042)
Strategic
Chnge. (H6)
Num. Orgs.
Changing Strategy
0.126*
(0.007)
0.223*
(0.012)
0.210*
(0.019)
0.210
(0.476)
Population
Print Newspapers
(Online)
0.017
(0.122)
0.143*
0.033
0.143
0.092
223
224
Table 25: Poisson Regression Results for News Community Failure Rate – Model Fit
Model AIC Null Deviance (df) Residual Deviance (df) Estimated r
2
Model 1: Environment Predictors 994.18 1704.22 (17) 887.17 (16) 0.45
Model 2: Environment Predictors +
Events
970.67 1704.22 (17) 861.66 (15) 0.47
Model 3: Competition + Strategy 209.65 1704.22 (17) 86.64 (8) 0.90
Model 4: Competition 521.09 1704.22 (17) 410.08 (14) 0.72
Model 5: Strategy 243.36 1704.22 (17) 130.35 (13) 0.87
Model 6: Change in Strategy 992.10 1704.22 (17) 885.10 (16) 0.43
Model 7: Saturated 153.49 1704.22 (17) 20.48 (6) 0.93
Model 7a: Saturated Quasi-Poisson na 1704.22 (17) 20.48 (6) -
224
225
Organizational Transformation and Poisson Regression Results
Hypotheses four and five predicted that various processes would confer
legitimacy on new organizational forms and subsequently drive an increase in the
organization birth rate, whereas hypotheses six, seven and eight predicted the effects
that various strategies would have on the failure rate of newspaper organizations. A
Poisson regression was utilized to assess the aggregate hypotheses: nested models
were used to assess the impact of each independent hypothesis. This was calculated
in order to address potential issues of multicollinearity. Second, the results for the
birth rate model are given in Table 22, and the results for the various fit measures are
given in Table 23. Hypothesis six, seven and eight examined the newspaper
population in specific, and look at the effect that various community, population and
organizational variables had on the effective failure or disruption of a given
organization. The results for the failure are given in Table 24, and the results for the
various fit measures are given in Table 25.
In order to check for issues of multicollinearity, a covariance matrix was
generated and is included in Appendix E. For over-time analysis correlations of 0.70
or higher are potentially indicative of multicollinearity, consequently, variables with
higher correlations were given extra examination. In order to assess the impact of
particularly troublesome parameters, additional models were run without the
identified variable in order to assess the true significance of each (Everitt & Hothorn,
226
2006). Generally, multicollinearity was not an issue for the data used in hypothesis
testing, as illustrated by the covariance matrix.
Hypotheses 4 and 5
Hypotheses four and five predicted the effect that various processes of
recognition would have on establishing a population as legitimate, and driving an
increase in organization birth rates. Several Poisson regressions were modeled in
order to assess the impact of each predicted relationship. The parameter results are
given in Table 22 and the model fit is given in Table 23. Based on the results of the
model fit parameters, the saturated model separating media coverage of blogs from
media coverage of online communities and social networking sites (AIC
Model 7
=
169.10) provides the optimal model, as compared to the model with a composite
media coverage measure (AIC
Model 8
= 180.83). Additionally, a Wald test was used to
compare the saturated model against the null model (intercept), showing the
saturated model to be statistically significant (!
2
= 565.43, df = 8, p<0.001). The
Quasi-Poisson regression is given in Model 8a, accounting for the overdispersion of
residuals and the resulting inflation of standard errors, and is used as an additional
verification of significance of results. Furthermore, the results of Dean’s test for
overdispersion suggest that overdispersion may be an issue in the birth rate data.
Dean’s test compares the observed variance against the theoretically expected
variance of a normall Poisson model; for the saturated model, D = 5.520, p < 0.05,
indicating that the observed variance is greater than the theoretical expectation; thus,
227
overdispersion for the birth rate model is an issue and the quasipoisson model should
be included for analysis.
Hypothesis four (a) predicted that media coverage of blogs and social
networking sites would have a positive relationship to the online news media
community birth rate. The combined effect of media coverage of blogs and online
communities was negative and significant in Model 2 ("
Model 2
= 0.243, p = 0.002)
and Model 8 ("
Model 3
= 0.375, p = 0.005). Based on the model fit measures, however,
the optimal measure of media coverage separated coverage of blogs from coverage
of social networking sites, shown in the AIC and estimated r
2
for Models 7 and 8
(AIC
MODEL7
= 169.7, r
2
MODEL7
= 0.79; AIC
MODEL8
= 180.83, r
2
MODEL8
= 0.77) Media
coverage of blogs had a positive and significant relationship to the organizational
birth rate in multiple models ("
Model 2
= 0.434, p = 0.002; "
Model 3
= 0.384, p = 0.004;
"
Model 7
= 0.533, p = 0.002). Media coverage of social networking sites did not have a
significantly positive relationship with organizational births. In addition, when
correcting for over-dispersion the relationship between blogs and organizational
births was no longer significant ("
Model 7a
= 0.533, p = 0.143), indicating that the
results are not overly robust. Thus, H4a is partially supported.
Hypothesis four (b) predicted that the formation of professional associations
would have a positive relationship to the birth rate of organizations in the online
news media community. The formation of professional associations had a positive
relationship to the rate of organizational foundings in all models ("
Model 2
= 0.098, p =
228
0.005; "
Model 3
= 0.078, p = 0.006; "
Model 4
= 0.256, p = 0.002; "
Model 7
= 0.093, p =
0.001). When correcting for overdispersion the relationship between blogs and the
formation of professional associations remained significant ("
Model 7a
= 0.093, p =
0.039). Thus, H4b was supported.
Hypothesis four (c) proposed that positive legal recognition would have a
positive relationship to the birth rate of organizations in the online news media
community. Positive legal rulings had a positive and significant relationship to the
rate of organizational births in Model 3 ("
Model 3
= 0.146, p = 0.008), Model 4 ("
Model
4
= 0.148, p = 0.002) and Model 7 ("
Model 7
= 0.071, p = 0.007). Correcting for
overdispersion, the relationship between blogs and organizational births was no
longer significant ("
Model 7a
= 0.071, p = 0.108), indicating that the results are not
overly robust. Thus, hypothesis four (c) is partially supported.
Hypotheses five examined the impact of the speciation mechanism,
predicting that an increase between existing organizations and new organization
forms would be positively related to the birth rate of online based news forms. The
results of the Poisson regression show that there is a positive and significant
relationship in model 3 ("
Model 3
= 0.012, p = 0.001) and model 7 ("
Model 7
= 0.006, p
= 0.002). The parameter is not, however, significant in the model correcting for
overdispersion ("
Model 7a
= 0.006, p = 0.089), thus significant drops to between p =
0.05 and p = 0.10. The findings are not particularly strong in terms of absolute value.
Therefore, while the relationship is generally significant, H5 is partially supported.
229
Based on the results of the legitimacy analysis conducted via Poisson
regression, the second research question sought to ask if there were certain measures
that were more likely than others to indicate the emergence and legitimacy of new
organizational forms. Results from the above analysis show that the formation of
professional associations had the strongest relationship to an increase in
organizational birth rates.
Hypotheses 6, 7 and 8
Hypothesis six, seven and eight examined the newspaper population in
specific, and tested the effect that various attributes had on organizational failures in
the newspaper population. Failures were considered to be changes in editors,
publishers or owners. Based on the model fit results given in Table 25, the saturated
model was the optimal fit for the estimated parameters (AIC
Model 7
= 169.10), and a
subsequent Wald test showed that the saturated model was significant as compared
to the null model (intercept) (!
2
= 87.78, df = 11, p<0.001). The quasi-Poisson
regression was again estimated to correct for the over-dispersion of residuals, and the
corrected model is given as Model 7a. Furthermore, the results of Dean’s test for
overdispersion suggest that overdispersion is not a major issue in the data. For the
saturated model, D = 1.41, p > 0.05, indicating that the observed variance is not
greater than the theoretical expectation. Overdispersion is not a major issue.
Hypothesis six predicted that during periods of transformation, organizations
that actively undertook strategic change initiatives would have a decreased
230
likelihood of failure. This hypothesis was tested by measuring the number of changes
in organizational strategy for each time period. Based on the Poisson regression
results given in Table 24, an increase in the number of organizations changing
hyperlinking strategies had the opposite effect, increasing the rate of failure ("
Model 3
= 0.126, p = 0.007; "
Model 6
= 0.223, p = 0.012; "
Model 7
= 0.210, p = 0.019). As the
parameter values indicate, a change in organizational strategy in a given year had a
positive impact on the number of organizational failures. In addition, the parameter
value was not significant in the quasi-Poisson model ("
Model 7a
= 0.210, p = 0.476).
Thus, H6 is not supported.
Hypothesis seven predicted that during periods of transformation, the
adoption of aggressive strategic initiatives would decrease the failure rate for
newspaper organizations in the online news media community. In order to test this
hypothesis, a variable was included for each of the four types of organizational
strategies: blogification, densification, experimentation and isolation. The results
show that blogification has a negative and significant relationship to organizational
failure in all models, including the quasi-Poisson model ("
Model 3
= -0.222, p = 0.003;
"
Model 5
= -0.210, p = 0.011; "
Model 7
= -0.243 p = 0.010; "
Model 7a
= -0.243, p = 0.043).
Densification had a negative and significant relationship to organizational failure in
two models ("
Model 3
= -0.004, p = 0.067; "
Model 5
= -0.004, p = 0.001; "
Model 7
= -
0.004 p = 0.001). The relationship was not significant in the quasi-Poisson model
("
Model 7a
= -0.004, p = 0.102). The small magnitude of the parameter suggests the
231
finding for densification does not represent a particularly strong effect.
Experimentation had a negative and significant relationship in all models ("
Model 3
= -
0.014, p = 0.003; "
Model 5
= -0.018, p = 0.001; "
Model 7
= -0.005, p = 0.007), with the
exception of the quasi-Poisson model ("
Model 7a
= -0.005, p = 0.154). Lastly, the
isolation strategy had a positive and significant relationship in all models, including
the quasi-Poisson model ("
Model 3
= 0.098, p = 0.001; "
Model 5
= 0.109, p = 0.010;
"
Model 7
= 0.120, p = 0.002; "
Model 7a
= 0.120, p = 0.042). The results suggest the
aggressive blogification strategy had a strong negative relationship with
organizational failure rates. On the other hand, isolation strategies had a positive and
significant relationship with organizational failure rates. The hypothesis predicted
that aggressive strategies would reduce the likelihood of failure, thus there is support
for the hypothesis in the effect of the blogification strategy. Hypothesis seven is
therefore supported.
Hypothesis eight predicted that the higher the number of institutional
hyperlinks maintained by online newspapers, the lower the mortality rate. This
relationship was tested with organizational links parameter, representing the number
of links between different populations of organizations. The results indicate a
significant and positive relationship in three models ("
Model 3
= 0.008, p = 0.002;
"
Model 5
= 0.013, p = 0.001; "
Model 7
= 0.008, p = 0.002), but the relationship is not
significant in the quasi-Poisson model ("
Model 7a
= 0.008, p = 0.184). The results
indicate a positive relationship between organizational links and the organizational
232
failure rate. The relationship is significant but relatively small in magnitude, and is
not supported when correcting for overdispersion. As a result, hypothesis 8 is
partially supported.
In aggregate, the Poisson regression models provide insight into the
interactions between the community, populations and organizations. By testing
multiple models, the true significance of each parameter is more efficiently and
accurately assessed. The use of the quasi-Poisson regression was particularly
important in determining parameter significance for organizational birth and failure
rates.
The Role of Strategy in Organizational Communities
Hypotheses 9 and 10
The previous hypotheses tested, in part, the role of strategy as a tool for
organizational transformation. This is further examined in hypotheses nine and ten,
which examine the role that initial organizational strategies have on organizational
position within a network over time. As previously discussed, RSiena was used to
estimate the longitudinal change in the online news media community from 1998
through 2007. A baseline model was previously established, including the estimation
of structural parameters. This baseline model was then used to calculate the effect of
organizational strategy. The organizational strategies in the newspaper population
were generally categorized as blogification, densification, experimentation and
isolation. Each attribute was coded as a dichotomous variable to estimate the effect
233
of the attribute on the network over time (elink - experimentation, blink -
blogification, dlink - densification and ilink – isolation).
Hypotheses nine (a) predicted that organizations with a high ratio of linkages
to new populations during periods of emergence are likely to have a high number of
outlinks over time. This hypothesis was tested with two parameters: CovEgo.blink
(blogification) and CovEgo.dlink, (densification). The results show that a
blogification strategy has a strong impact on link creation in later time periods, from
2000 onwards (CovEgo.blink
2000-2001
= 1.468, SE = 0.500; CovEgo.blink
2001-2002
=
2.010, SE = 0.510; CovEgo.blink
2002-2003
= 1.980, SE = 0.509; CovEgo.blink
2003-2004
=
2.011, SE = 0.512; CovEgo.blink
2004-2005
= 2.001, SE = 0.511; CovEgo.blink
2005-2006
=
1.950, SE = 0.507; CovEgo.blink
2006-2007
= 1.655, SE = 0.697). The densification
strategy had no significant impact on organizational strategy over time. The results
lend mixed support to hypothesis nine (a). The blogification strategy has a strong
impact on link creation whereas densification had no significant impact at all.
Therefore, hypothesis nine (a) is partially supported.
Hypothesis nine (b) predicted that organizations with a small number of
linkages to new populations during periods of emergence are likely to have a low
number of links over time. This hypothesis was tested with two parameters:
CovEgo.elink (experimentation) and CovEgo.ilink (isolation). The results show
strong support for the experimentation strategy having a positive effect on link
creation from 2000 onwards (CovEgo.elink
2000-2001
= 0.099, SE = 0.049;
234
CovEgo.elink
2001-2002
= 0.102, SE = 0.044; CovEgo.elink
2002-2003
= 0.110, SE = 0.048;
CovEgo.elink
2003-2004
= 0.113, SE = 0.052; CovEgo.elink
2004-2005
= 0.108, SE = 0.049;
CovEgo.elink
2005-2006
= 0.106, SE = 0.041; CovEgo.elink
2006-2007
= 0.107, SE = 0.040).
Organizations that had an isolation strategy had a negative and significant parameter
from 2000 onwards (CovEgo.ilink
2000-2001
= -0.102, SE = 0.019; CovEgo.ilink
2001-2002
= -0.130, SE = 0.016; CovEgo.ilink
2002-2003
= -0.502, SE = 0.063; CovEgo.ilink
2003-
2004
= -0.306, SE = 0.059; CovEgo.ilink
2004-2005
= -0.810, SE = 0.054;
CovEgo.ilink
2005-2006
= -0.313, SE = 0.041; CovEgo.ilink
2006-2007
= -0.130, SE =
0.049). The results present conflicting findings. Organizations that chose an
experimentation strategy actually had an increased likelihood of outlinking over
time, despite a relatively conservative strategy. Organizations that remained isolated,
however, were likely to form relatively few links over time. In addition, neither
parameter was significant during the first two time periods, despite general
significance during the emergence period. The conflicting results indicate that
hypothesis nine (b) is not supported.
Hypothesis ten (a) predicted that organizations with a high number of links
during the emergence stage were likely to receive a high number of links over time.
The hypothesis was tested with the above strategies, but with the covariate alter
(CovAlter) parameter, which measures the effect an attribute has on inlinking over
time. The results show that diversification has no effect on inlinking over time, as
shown by the lack of significance of the CovAlter.dlink parameter. There is,
235
however, moderate support for the effect of the blogification strategy. There was a
significant and positive effect in seven time periods: 1999-2000 (CovAlter.blink
1999-
2000
= 0.880, SE = 0.361) ; 2000-2001 (CovAlter.blink
2000-2001
= 0.751, SE = 0.329);
2001 – 2002 (CovAlter.blink
2001-2002
= 0.712, SE = 0.362); 2002 – 2003
(CovAlter.blink
2002-2003
= 0.810, SE = 0.389); 2003 – 2004 (CovAlter.blink
2003-2004
=
0.823, SE = 0.409); and 2006 – 2007 (CovAlter.blink
2004-2005
= 0.860, SE = 0.411).
The results show strong support for the effect of the blogification strategy, while
there is no support for the effect of the densification strategy. Thus, hypothesis ten
(a) is partially supported.
Lastly, hypothesis ten (b) predicted that organizations with a low number of
links during the emergence stage were likely to receive a low number of links over
time. The hypothesis tested the effect that experimentation and isolation strategies
had on organizational inlinking over time. The results show little support for the
effect of isolation: there is a significant and positive effect in 1998-1999
(CovAlter.ilink
1998-1999
= 0.130, SE = 0.086) and 1999-2000 CovAlter.ilink
1999-2000
=
0.023, SE = 0.010. Additionally, organizations that had an experimentation linking
strategy had an increased probability of receiving links over time in six time periods:
1998 – 1999 (CovAlter.elink
1998-1999
= 0.489, SE = 0.119); 1999 – 2000
(CovAlter.elink
1999-2000
= 0.302, SE = 0.150); 2000 – 2001 (CovAlter.elink
2000-2001
=
0.309, SE = 0.142); 2001 – 2002 (CovAlter.elink
2001-2002
= 0.298, SE = 0.149); 2003 –
2004 (CovAlter.elink
2003-20004
= 0.289, SE = 0.122); and 2005 – 2006
236
(CovAlter.elink
2005-2006
= 0.298, SE = 0.124). The results for both experimentation
and isolation strategies show that there is an increased probability of receiving links
over time: this is opposite of the predicted direction. Hypothesis ten (b) is therefore
not supported.
Summary
The hypotheses analyzed in this chapter are summarized in Table 27. The
results show moderate support for the speciation mechanism, and furthermore
highlight the impact that organizational strategy can have on network position over
time. In aggregate, the results provide a complete picture of the transformation of the
news media industry; these results and subsequent implications are discussed in
depth in Chapter 7.
237
Table 26: Estimation for RSiena Model 1998 – 2007 (Attributes for H
9a
, H
9b
, H
10a
, H
10b
)
Attribute-Based
Parameter
98-99 99-00 00-01 01-02 02-03 03-04 04-05 05-06 06-07
CovEgo.blink
1.023
(0.609)
1.095
(0.590)
1.468*
(0.500)
2.010*
(0.510)
1.980*
(0.509)
2.011*
(0.512)
2.001*
(0.511)
1.950*
(0.507)
1.655*
(0.697)
H
9a
CovEgo.dlink
0.758
(0.609)
0.655
(0.602)
0.608
(0.610)
0.698
(0.603)
0.710
(0.699)
0.707
(0.640)
0.709
(0.632)
0.712
(0.621)
0.766
(0.632)
CovEgo.elink
0.103
(0.059)
0.098
(0.050)
0.099*
(0.049)
0.102*
(0.044)
0.110*
(0.048)
0.113*
(0.052)
0.108*
(0.049)
0.106*
(0.041)
0.107*
(0.040)
H
9b
CovEgo.ilink
-0.002
(0.020)
0.010
(0.024)
-0.102*
(0.019)
-0.130*
(0.016)
-0.502*
(0.063)
-0.306*
(0.059)
-0.810*
(0.054)
-0.313*
(0.041)
-0.130*
(0.049)
CovAlter.blink
0.650
(0.359)
0.880*
(0.361)
0.751*
(0.329)
0.712*
(0.362)
0.810*
(0.389)
0.823*
(0.409)
0.708
(0.420)
0.785
(0.419)
0.860*
(0.411)
H
10a
CovAlter.dlink
0.542
(0.361)
0.562
(0.349)
0.583
(0.319)
0.559
(0.310)
0.561
(0.314)
0.563
(0.327)
0.552
(0.320)
0.505
(0.319)
0.490
(0.311)
CovAlter.elink
0.489*
(0.119)
0.302*
(0.150)
0.309*
(0.142)
0.298*
(0.149)
0.288
(0.146)
0.289*
(0.122)
0.280
(0.159)
0.298*
(0.124)
0.300
(0.152)
H
10b
CovAlter.ilink
0.130*
(0.086)
0.023*
(0.010)
-0.021
(0.092)
-0.033
(0.091)
-0.102
(0.111)
-0.120
(0.100)
-0.103
(0.098)
-0.030
(0.049)
0.001
(0.082)
* indicates significance at p < 0.05
237
238
Table 27: Summary of Results
Hypothesis Results
H1: During the emergence phase, new organizational
forms are more likely to link to other organizations of
the same type than to different types.
Supported
H2: During the emergence phase, there will be a
proportionally small, but significant, number of links
between a new organizational form and existing
specialist organizations.
Partially
Supported
H3: Links between a new organizational form and the
existing community will increase after the density of the
speciated form begins to increase.
Supported
H4: Media coverage of blogs and social networking sites
(A), the formation of professional associations (B) and
legal recognition (C) will be positively related to the
birth rate of online-based news forms.
4a: Partially
Supported
4b: Supported
4c: Partially
Supported
H5: An increase in the number of links between existing
organizations and new organizational forms will be
positively related to the birth rate of online-based news
forms.
Supported
H6: During periods of transformation, organizations that
actively undertake strategic change initiatives will have
a decreased likelihood of failure
Not
Supported
H7: During periods of transformation, the adoption of
aggressive strategic initiatives will decrease mortality
rates for inertial organizations
Supported
H8: Interorganizational linkages will significantly
reduce the effect of competition on organizational
mortality rates.
Partially
Supported
239
Table 27: Continued
H9A: Established organizations that have a symbiotic
strategy with a high ratio of linkages to new populations
during periods of emergence will have a high proportion
of outlinks to new populations over time.
Partially
Supported
H9B: Established organizations with a low ratio of
linkages to new populations during periods of
emergence will have a low proportion of outlinks to new
populations over time.
Not
Supported
H10A: Established organizations with a high ratio of
linkages to new populations during periods of
emergence will have a high proportion of inlinks to new
populations over time.
Partially
Supported
H10B: Established organizations with a low ratio of
linkages to new populations during periods of
emergence will have a low proportion of inlinks to new
populations over time.
Not
Supported
240
CHAPTER 7: CONCLUSION
Discussion
This study examined a series of research questions related to three critical
aspects of the ongoing transformation of the news media community. First, this
research sought to develop the concept of speciation as a mechanism for
understanding how new organizational forms emerge and compete for resources in
longstanding industries. The examination of form emergence focused specifically on
the rapid and successful emergence of blogs and social networking sites as
alternative sources of daily news. Second, this study examined the process of
legitimation in order to provide a better understanding of the mechanisms that lead to
new organizations becoming established as legitimate populations within an
established community. Lastly, this study examined how existing populations, such
as traditional newspapers, undertake transformational processes in response to the
emergence of new forms of competition.
In addition to the above, this research introduced a new tool for historical
Internet-based research, and applied cutting-edge analytic techniques to the above
hypotheses. The results are discussed in the following sections, covering the
emergence, legitimation and transformation of organizations.
Emergence
The first research question sought to extend the stage model of community
evolution to the evolution of the online news media community. The applicability of
241
the stage model of community evolution is extended here, although it is unclear
whether or not all three stages actually occur within the time period examined; based
on the results of the density analysis, the community as a whole is still in a
maintenance stage and has not reached a threshold of self-sustainability. The
maintenance period is marked by the flow of substantial resources between
populations. The density analysis showed that the online news media community
was in a period of emergence from 1998 through 2004, at which point the
community entered a maintenance stage. Notably, the density analysis relied on an
examination of the density of the main component rather than the overall network
density. This was due, in large part, to the scale free nature of online networks that
resulted in low-density values.
New forms continue to develop and existing populations continue to change,
but the online news community was well established by 2008 as a sizable population
of organizations. Simultaneously, the traditional print newspaper population
continues to bleed resources, losing roughly 25 percent of the industry’s workforce
in a nine year period. This bleeding of resources, however, has benefited emergent
organizational forms as new populations absorb slack resources. A 2010 report
noted, for instance, that many reports of employment loss are overstated in the news
industry because an increasing number of jobs are being recaptured by online news
outlets and alternative forms of media (Edmonds et al., 2010). Similarly, research
has shown that over the past decade both newspaper and television readership has
242
continued to decline, while online sources continue to gain in popularity (Kohut,
2008). Maintenance does not preclude ongoing transformation; rather, it implies that
the community is developed although there may be the potential for continued
change. Self-sufficiency, on the other hand, occurs when the number of competitive
and mutual relationships in a community is in balance, such that the community is
sufficiently robust against competitive pressures (Bryant & Monge, 2008).
Having established the stage model of community development as a high-
level outline for analysis, Hypothesis one, two and three developed the theoretical
mechanism of speciation as a framework for understanding the rapid emergence of
new organizational forms. Organizational evolution is fundamentally a process of
variation, selection and retention, with the development of new organizational forms
being driven by successful variations of existing organizational forms (Campbell,
1965). Existing theory, however, fails to fully explain why existing populations may
experience periods of significant upheaval as the result of an influx of competing
organizational forms. The V-S-R process is a gradual ecological mechanism. Within
this framework, however, speciation provides a mechanism for explaining how new
organizational forms can develop in an adjacent resource space and ultimately reach
a critical mass, creating a rapid disruption in existing, established communities. In
order to examine these hypotheses, two types of network analysis were conducted: a
baseline model was established using statnet and a longitudinal analysis was
conducted using RSiena.
243
In accordance with the process of speciation, hypothesis one proposed that
new organizational forms are likely to develop in clusters of similar organization
types. The study found evidence that blogs were likely to link to other blogs during
the emergence period, whereas online communities were more varied in their linking
patterns. During the emergence phase, blogs accounted for a predominant amount of
new organizational births (29 percent in 2000, and 59 percent in 2001). Online
communities have grown as an organizational form since the early 1990s, however
they did not grow in mass density until after 2004 (boyd & Ellison, 2008). Thus, it is
not altogether surprising that the results were significant for blogging given that the
growth of blogs was significant during this emergence period.
The process of speciation does not, however, occur in isolation. Rather,
speciated forms develop alongside the existing community, but in separate resource
niches. From a biological perspective, this process postulates that speciated
organizational forms interact with established populations, but because they draw on
separate resource pools they are not viewed as competitors during emergence
(Gavrilets et al., 1998). This type of competition was predicted to occur at the fringes
of existing populations, with specialist organizations interacting with speciated
organizational forms. The results of hypothesis two showed that during much of the
emergence period, there was a strong probability of link creation between emergent
organizations and established specialist newspapers. The findings are particularly
relevant from 2001 through 2003, when blogs grew rapidly in density. Although the
244
findings were not consistent across all time periods, they align with Barrett’s (1989)
notion of organizational forms developing at arms length. Ties to existing
organizational forms are a critical tool for new organizations to gain access to
resources, and enable emergent forms to grow towards legitimacy. In this way, early
ties between niche newspapers and blogs fostered the growth of blogs by conveying
an early sense of recognition. It is clear, however, that neither blogs nor online
communities were viewed as a competitive threat during this early period as the
community at large did not react or actively engage with this population.
Early but sparse links are formed during the emergence phase of
organizational growth, but once organizations enter into direct competition the
number of links between forms will increase significantly as full competition
ensures. Hypothesis three showed that there is a positive rate of link formation
between blogs and established organizational forms, and social networking sites and
established organizational forms, in the maintenance stage of community
development – from 2005 onward. For blogs there is a generally constant rate of tie
establishment during the maintenance phase. On the other hand, the rate of the
covariate alter parameter increased over time for online communities, indicating that
over time these organizational forms, particularly social networking sites, became
more connected to established populations. Both results indicate increased
competition between emergent organizational forms and established organizational
forms as the new forms became more established.
245
These findings thus establish the process of speciation as a potential
mechanism for explaining the emergence of new populations of organizations. With
regards to technological innovation, existing research has primarily focused on
technological innovation as a population-level disruption that sparks the creation of
new organization types (Anderson & Tushman, 1990; Tushman & Anderson, 1986).
Online communities and blogs are organizational forms that were sparked by
technological development in the early 1990s. Yet blogs and social networking sites
did not enter into competition with the news media community until the late 1990s.
The punctuated equilibrium model thus does not fully explain how an organizational
form can gradually develop, and then rapidly enter into competition. Speciation, on
the other hand, explains exactly this process. From the perspective of organizational
ecology, this research thus presents evidence of this alternative process of population
development and establishes speciation as a viable framework for examining the
emergence of other organizational populations.
Legitimation
Legitimation is the second critical process examined in this research. This
work was specifically interested in mechanisms for detecting the rapid emergence of
new organizational forms. The majority of existing ecological research has focused
on processes of density dependence, positing that density is a harbinger of
legitimation (Carroll & Hannan, 2000). The process of speciation, however, predicts
that forms can establish in a critical density before entering into competition with an
246
existing population. Thus, this research focused on other mechanisms for
establishing legitimacy of organizational forms beyond density dependence.
Hypotheses four and five, as well as Research question two, thus looked at the
process of legitimation by utilizing a Poisson regression to analyze the various
factors that impacted organizational birth rate in the online news media community.
The central premise of this analysis was that an increase in legitimacy of a new
organizational form would in turn been seen through an increase in the
organizational birth rate for the community.
Hypothesis four looked at media coverage, the formation of professional
associations and positive legal recognition as central processes for legitimacy. First,
Hypothesis four (a) showed partial support for media coverage as a critical
legitimating process. Media coverage of blogs had a strong positive effect on the
organizational birth rate, whereas media coverage of social networking sites did not
have a significant relationship. Overall, the relationship was not terribly strong once
overdispersion was accounted for, but the findings suggest that media coverage
confers a degree of legitimacy onto new organizations (as seen in the birth rate). This
echoes previous research that has found similar effects (Pollock & Rindova, 2003).
The lack of significance in the results for social networking sites is partly due to the
nature of the data: the Web archive data lags actually online events, and although
social networking sites have emerged as notable competitors in the online news
247
arena, their impact was not strongly felt until roughly 2008. Thus, this impact was
not captured in the data presented here.
Hypothesis four (b) found strong support for the relationship between the
formation of professional associations and organizational birth rate. Prior studies
have shown that professional associations are an important indicator of legitimacy
(McKendrick & Carroll, 2001; McKendrick et al., 2003; Rao, 1998); this study
highlights the critical nature of professional associations as an indicator of legitimacy
in the case of rapidly emerging competitive organizational forms. Lastly, hypothesis
four (c) found while there is a positive birth rate between positive legal rulings and
the organizational birth rate, the relationship was moderately significant and
relatively small in magnitude. Thus, although this hypothesis had partial support,
more research is needed to fully examine this effect. Specifically, interviews with
reporters in both online and traditional organizations should explore the effect of
both professional associations and legal rulings on the perceived legitimacy of
organization. In addition to the above, hypothesis five found some support for the
relationship between link formation and organizational birth rate. This relationship
indicated some support for the effect of links between existing and new
organizational forms on the birth rate, although the magnitude of the effect was
relatively weak.
In aggregate, the above findings shed some insight onto the process of
legitimation. In response to the second research question, media converge and the
248
formation of professional associations had the strongest effects, suggested that in the
case of rapidly emerging competition these are the optimal early indicators of form
emergence. From a broader perspective, the results of the legitimation analysis
reveals additional insights into the emergence of new organizational forms.
Interestingly, there was not a strong effect for density dependence: while both the
density of established and emergent organizations had moderately significant
impacts on organizational birth rates, neither effect was particularly large in
magnitude. This further emphasizes the importance of alternative indicators of form
emergence. For instance, the Interactive Advertising Bureau was formed in 1996,
and the Online News Association was formed in 1999. Today both have emerged as
leading authorities on online media, and in the late 1990s there were clear indicators
that a new population of organizational forms being established as a legitimate
competitive sector.
The lack of support for legal rulings as a indicator of legitimacy was
surprising given its prominence in previous research. Ruef (2000), for instance,
noted that legal recognition is one of the most significant indicators that a new form
is established as an institution. A number of legal rulings have established bloggers
as journalists, and in the case of California, guaranteed blogger-journalists that same
legal protection, yet this did not translate through to an increase in birth rates. In part,
this is likely due to a long-standing operational distinction in the news media
industry: the newsroom operations are distinct and separate from the business
249
operations of the online news business. This church-state separation has existed since
the early 1900s, and as a result, legal rulings affecting the legitimacy of the blogger
as journalist are less likely to have the same impact as the formation of professional
associations. One mechanism affects the production process, while the other affects
the business operations. This may explain, in part, why legal rulings did not have the
same impact as in prior studies.
Transformation
When a new organizational form emerges as a legitimate and successful
population, existing organizations must transform or face the risk of failure. Even in
transforming, organizations accept an inherently elevated possibility of failure. In the
case of the newspaper community, it is clear that the print newspaper population had
anticipated the growth of online newspapers for more than a decade. Furthermore,
numerous organizations with the print newspaper population had expected online
technology to be a prominent factor in organizational development since the late
1980s (Boczkowski, 2004a). Despite anticipating the development of digital
technology, the rapid emergence of socially-based forms of news in the 1990s and
2000s forced newspapers to react through a process of transformation and
adaptation. First and foremost, the introduction of online news changed the rules of
the news production business; the use of hyperlinks enabled a fluidity that did not
exist in the printed environment. This is evidenced in the growth of a link economy,
250
through which hyperlinks have developed as a tool for directing information flow
and guiding consumers of news.
In the 1990s, newspaper organizations thus faced a critical decision regarding
strategy selection. Essentially, newspapers either chose to embrace online strategies
or maintained status quo. From an ecological perspective, the effect of strategy was
largely disregarded; inertial organizational forces are generally seen as dominant,
and individual organizational action is not perceived as having a strong effect on
survival. With that perspective, this research following in line with Levinthal (1998)
suggestion that more research is needed to understand the long-term effect of
strategy. Hypotheses six through ten looked at the impact that organizational strategy
had on the transformation process from an ecological perspective.
In terms of this analysis, failure was established as a clear organizational
decision to abandon the existing operational plan: this failure was thus measured as a
change in owner, publisher or editor-in-chief. Hypothesis six, however, showed that
during periods of transformation changes in strategy actually had an increased effect
on organizational failure, although this effect was not overly large in magnitude.
This finding echoes previous ecological research, which established that the
undertaking of strategic change initiatives during transformational periods will
actually increase the rate of failure in existing organizations (Amburgey et al., 1993;
Hannan & Freeman, 1984). Although changes in strategy were found to increase
failure rates, Hypotheses seven and eight support strategic hyperlinking as a tool for
251
reducing organizational failure rates. Hypothesis seven was supported: organizations
that adopted the most aggressive linking strategy significantly reduced the likelihood
of failure. Results for less aggressive strategies were not nearly as strong, further
emphasizing the results of this finding. In addition, hypothesis eight established that
on the whole the formation of interorganizational linkages had a moderate impact on
reducing organizational mortality. In combination, these two findings show that the
concept of a link economy moves beyond simple hyperlinking; long-term
hyperlinking strategies had a direct impact on the survival of online news operations.
Beyond looking at survival rates, the exploration of organizational strategy
also considered the impact of strategy on organizational position within the larger
community network. This study posited that hyperlinks in online environments are
akin to the ecological concept of commensalist and symbiotic relationships.
Organizations that build commensalistic relationships use hyperlinking as a tool to
insulate against outside organizations, and ultimately compete for a restricted pool of
available resources. Symbiotic relationships, on the other hand, occur when
relationships are beneficial to both parties. Hyperlinks provide mutual benefits; they
direct traffic between organizations and by serving as a directory of information
sources the linking organization increases the likelihood that users will return. In
turn, the establishment of these links helps to shape population development over
time. Applying longitudinal network analysis, Hypotheses nine and ten show some
support for the impact of organizational strategy on network position over time.
252
Hypothesis nine was partially supported: organizations with aggressive linking
strategies were likely to continue to out-link aggressively overtime; on the other
hand, there was not enough evidence to draw conclusions about organizations that
chose conservative linking strategies.
Likewise, hypothesis ten was partially supported, showing that organizations
that adopt aggressive linking strategies are likely to receive an increasing proportion
of links over time, whereas organizations that adopt a conservative linking strategy
are likely to receive fewer links over time. Surprisingly, there was no support with
regards to the most conservative strategy, as indicated by the lack of significance of
the CovAlter.ilink parameter, and the positive direction for both CovAlter.ilink and
CovAlter.elink. That said, comparatively the magnitude of the parameter was
significantly smaller for conservative organizations that it was for aggressive
organizations. For instance, in the 2001 – 2002 time period, CovAlter.blink was
strongly significant and positive (CovAlter.blink
2001-2002
= 0.712, SE = 0.362),
whereas the value for experimentation was significantly smaller in magnitude
(CovAlter.elink
2000-2001
= 0.309, SE = 0.142). The same is found when comparing
against isolation strategies. For example, in the 1999 – 2000 time period,
CovAlter.blink was significant and positive (CovAlter.blink
1999-2000
= 0.880, SE =
0.361), and much larger in magnitude than the parameter value for isolation
(CovAlter.ilink
1999-2000
= 0.023, SE = 0.010). Thus, although hypotheses nine (b) and
ten (b) were not supported, directionally these findings highlight the importance of
253
hyerplinking strategy in determining organizational position over time. Additionally,
there is considerable support for the positive and significant impact of aggressive
hyperlinking strategies over time.
Implications for Evolutionary Theory and Social Network Analysis
This study explored the mechanisms of organizational transformation from
the community perspective. The results show the utility of community ecology as a
framework for studying the emergence of new organizations and for examining the
resulting process of organizational transformation. In addition, from a network
perspective this study builds on previous work suggesting the validity of networks
methods as a tool for ecological studies (Contractor, Wasseman, & Faust, 2006; S.
Lee, 2008; Monge & Contractor, 2003; Monge et al., 2009 in press), and shows that
longitudinal network analysis is a critical tool for examining evolutionary
mechanisms in organizational communities. Thus, this dissertation makes a number
of key contributions to the study of networks and organizations.
The development of speciation as a mechanism for form emergence is a
critical contribution to the literature on organizational form emergence. The process
of form emergence is not well understood; rather, research has focused on the nature
of organizational forms and their defining characteristics (Carroll, 1984; Carroll &
Hannan, 2000; Polos et al., 2002). Form emergence is generally attributed to the
process of punctuated equilibrium (Astley, 1985), yet this process assumes the
establishment of the form itself is taken for granted. Speciation, however, shows that
254
forms emerge in open resources niches and are able to develop over time before
entering into competition at a given inflection point. This research has established
speciation as a critical process for understanding form emergence, although more
research is needed to examine the nature of the form itself.
The present study contributed to the literature on legitimation by examining
the process of legitimation during periods of rapid change. As posited by the
speciation mechanism, there are instances where new organizational populations can
develop in parallel with an existing community, only to enter into competition after a
critical density has been reached. Thus, although density dependence may account
for the legitimacy of an individual population, this study shows that other
mechanisms establish the legitimacy of a population within the encompassing
community. Mechanisms such as media coverage and the formation of professional
associations are thus crucial tools for legitimacy with regards to the interaction with
other populations, competition for resources, and ultimately a decrease in
organizational failure rates.
In addition, this study suggests that organizational strategy will increasingly
be an important area for study in the ecological context. Levinthal (1998) observed
that the importance of strategic change is often overlooking because most change is
gradual. This study concurs with the interview statements, but the findings presented
here show that aggressive strategies can have a considerable impact on survival over
time. Although the time period examined was brief by evolutionary standards, it was
255
substantial sample in the development of online media. Furthermore, the concept of a
link economy inherently suggested a increased pace of transformation: online
enterprises, and online relationships, can be established on a scale far greater than
traditional industrial development (Halavais, 2008). Previous studies have primarily
disregarded strategy due to an overemphasis on structural inertia (McKelvey &
Baum, 1999; Phan et al., 2010; Suarez & Oliva, 2005), yet the swift pace of online
development suggests that aggressive strategies can have an effective similar to
technological disruptions. Tushman and Anderson (1986) found that technological
disruptions either enhance or destroy existing competencies within an industry.
This study established an opposite, yet complimentary, perspective:
organizations that adopt an aggressive strategy succeed when that strategy aligns
with the nature of the emergent technology, but if the strategy is incongruous the
organization faces an increased rate of failure. The potential risk of failure was
evidenced in the findings of hypothesis six, which showed an increase in failure rate
for organizations undertaking strategic change initiatives. Thus, strategy is shown to
be an important component of organizational development, yet this perspective is not
incompatible with existing ecological research.
Furthermore, this study emphasizes the important connection between social
network analysis and organizational studies. Social network analysis provides a tool
for examining organizational development in the context of interaction with other
organizations. Monge et. al. (2009 in press) suggest that this perspective is
256
advantageous because it allows for the study of organizational behavior, and the
resulting interorganizational interaction. Longitudinal network analysis extends this
perspective, allowing for the study of organizational networks over time. New
computational tools paired with increased computing power allows for the study of
larger data sets, and larger organizational networks. In addition, as illustrated in this
research, network measures can be used to created new attributes of organizational
development. Network analysis can then be used in concert with existing research
methods to study ecological processes.
Finally, from a methodological perspective, this study also advances a multi-
method approach to organizational ecology. This dissertation utilized network
analysis, generalized linear models and qualitative interview analysis in order to
examine the development of online news media. Qualitative research has been
emphasized by some as a key component in understanding organizational evolution
(DiMaggio, 1994; Fombrun, 1988). As it becomes possible to analyze larger data
networks, it is increasingly important to incorporate qualitative research into these
types of analyses. This research advances this perspective, utilizing qualitative
research to better understand strategic action, and embedding those strategies within
community level network data.
Implications for the News Media Community
In the opening chapter of this dissertation, it was noted that there is a
continued need for new methods and theories to understand the changing nature of
257
journalism (Zelizer, 2007). This study responded to that call, applying ecological
theories and network methods to better understand the transforming nature of news
media. In addition, the findings of this research hold a number of implications for the
news media community. First, this research establishes the long-term importance of
hyperlinking as a strategic tool, and highlights the critical nature of the link
economy. Secondly, this study sheds insight on the development of new
organizational forms; better understanding the development of competitive
organizations will help existing organizations to compete more effectively in the
future.
The nature of hyperlinking was a central issue in this research; hyperlinks
are a key tool for online organizations, and yet many news organizations have
resisted adopting open hyperlinking practices. At its core, this issue focuses on
newspaper organizations’ desire to control information, a battle that was
foreshadowed during the very infancy of the online industry:
On the one hand information wants to be expensive, because it’s so valuable.
On the other hand, information wants to be free, because the cost of getting it
out is getting lower and lower all the time. So you have these two fighting
against each other. (Brand, 1985, p. 49)
In 1985, Stewart Brand made the above comment at the First Hacker’s Conference,
held in November 1984 in Marin County, California. At the time, Brand was talking
about the ongoing debate over whether or not programmers should have to pay for
the software they use. Today, Brand’s commentary aptly describes the issue that lies
at the heart of the ongoing upheaval in the news media community. Hyperlinks help
258
to govern the flow of information online, and as this study demonstrated, a
successful hyperlinking strategy is critical to organizational survival over time Thus,
this work shows that there is merit to Jarvis’ proclamation that the time has come for
newspapers to “do what they do best and link to the rest” (Jarvis, 2008).
Indeed, as newspaper organizations have developed over time, editors within
these organizations have begun to realize the importance of linking. As one online
editor at Boston Globe noted:
We’ve always said that it’s important to link., For example, it’s not
enough to just be talking about a UN report and not to actually link
to the UN, or to that report or even to something that mentions the
UN. To try and pretend that you're the only site on the Internet is
very old media and oh so very shortsighted. (Anonymous, 2009d)
As the above editor emphasizes, hyperlinking is a mechanism for providing users
with information that cannot readily be provided by the organization itself. As media
becomes increasingly social, this mechanism will continue to prove more important
to news organizations. The rapid growth of social networking sites such as Flickr,
YouTube, Facebook.com and Orkut continues to propagate the growth of the link
economy and push news organizations to adapt to a linked news environment. For
example, Figure 17 illustrates the integration of social media into The Wall Street
Journal’s online story format:
259
Figure 17: The Wall Street Journal’s Online Format
The first circled portion is an integrated bar that allows users to automatically link
directly to a story from their own sites. The second circled portion provides the
reader with hyperlinks directly to Pfizer and Merck, which are both mentioned in the
story about the Dow Jones Industrial Average’s recent growth.
As this study emphasized, hyperlinking is increasingly critical to the success
of online news organizations. In addition, as traditional news organizations move
into the online space, the choice of hyperlinking strategy is a critical determinant of
success online. For traditional newspapers the hesitancy in linking is derived from a
260
number of different motivations: the creation of hyperlinks to online news outlets
serves as a act of legitimation conferred upon Web sites, and simultaneously the
creation of links essentially creates a free flow of news online. Yet at the same time
creating ties to new organizational forms provides access to information and
knowledge about new business practices (Baum & Oliver, 1991). Practically
speaking, this may overstate the importance of hyperlinks; yet this study focused
only on strong ties – those that had been established three or more times in one year.
Such a tie does not necessarily imply a formal relationship between two
organizations, but it does show a clear awareness of a connection between the two
organizations.
In addition to the above discussion, this research also shows the practical
importance of a number of legitimating processes. For example, the formation of
professional associations was shown to be critical in establishing new populations as
legitimate. Practitioners can thus benefit from actively engaging with existing
professional organizations. In doing so, practitioners can maintain a greater
awareness of community development, and be better prepared when new
organizations emerge as competitors. Professional associations provide a forum for
cultivating existing knowledge about existing communities, but can also indicate the
emergence of new forms. Ironically, media coverage was also found to be a critical
tool for learning about new innovations. News organizations can thus learn by
261
internalizing their own institutions existing coverage of technology and emerging
trends.
Limitations and Future Research
This study has made significant contributions to both existing theories of
organizational development, and existing knowledge about the online news media
community. There are, however, a number of limitations to this study.
Limitations
This study has a number of limitations due to the nature of the data. First, this
study does not capture the full scale of the online news media community. As noted,
the data used in this study represents a sample of the online news media community.
As a result, many niche organizations and isolated clusters or subgroups, were not
included in this study. Furthermore, the data set was restricted to organizations in the
United States, despite the fact that online news media is generally global in terms of
information flow. In addition, although the initial dataset collected information on
roughly 25,000 Web sites, filtering reduced the sample to roughly 2,000
organizations. Thus, although while the number of sites filtered out appears high at
first glance, robust filtering enabled an accurate analysis.
Second, the data were generally restricted to online organizations. The sub-
analysis of online news organizations utilized data about newspaper organizations,
but overall this study did not focus on the characteristics of traditional organizations.
This is particularly restricting with regards to the analysis of organizational strategy.
262
A more comprehensive analysis of the internal decision-making process, and the
internal process of transformation, would yield additional insights, but this is
reserved for future research.
Third, the data were generally limited with regards to the analysis of
organizational performance. Little data were available with regards to online
performance and online success. Thus, beyond measuring network position and
survival, it was not possible to measure the online success on a large scale. As new
datasets become available, this type of analysis should be available in the future, and
will facilitate a more complete analysis of online success. Furthermore,
organizational forms were assumed to be static in this study. There is clear evidence,
however, that as organizations develop in the online news media community many
organizations are adopting hybrid organizational forms. For instance, the integration
of social media into existing organizations is continually redefining the nature of
social networking sites.
In addition to the above, there are also a number of limitations resulting from
the analytical tools chosen for this study. The use of RSiena assumes that
organizations are behavioral actors, functioning as nodes seeking to optimize self-
interest. This may, in certain cases, overemphasize the preference for creating
interorganizational links as organizations seek to optimize network position (Snijders
et al., 2010). Previous studies, however, have found that this assumption does not
generally create significant issues (Van de Bunt & Groenewegen, 2007). More
263
important, given the large scale of the network data, it was not possible to obtain
optimal convergence. In the goodness-of-fit analysis there continues to be evidence
of modulation in the in-degree and out-degree analysis. Parameter values were
generally converged, and non-converged parameters were not used to draw
conclusions, yet a fully converged model would indicate a higher level of reliability
overall, and may increase the significance of certain findings.
Directions for Future Research
This study provides researchers with a number of potential avenues for future
analysis, as discussed below.
Archival Internet Research
This study has shown HistoryCrawl to be a useful tool for archival internet
research. This tool is the first proven Web crawler for extracting archival information
from the Internet Archive, and its use for online research promises to provide a
wealth of information about the development on the Internet and online information.
From the perspective of organizational studies, this tool can be applied to examine a
wide array of evolutionary questions. For instance, future studies could examine the
development of social news media through the lens of collective action. By studying
the historical growth of social media, and the growth of user contribution to social
media Web sites, researchers will be able to better understand how these
organizations have developed as a major online media population.
264
In addition, the use of the HistoryCrawl tool holds particular promise for the
study of global online networks. The present study focused exclusively on the US
news media community, yet the Internet Archive provides a global repository of
organizational data. Future studies would benefit from a focus on the nature of global
information flows, and examining the development of global information networks.
Strategy and Online Development
Furthermore, this study focused extensively on the role of strategy in the
development of online organizations. In turn, this study examined the online aspects
of news organizations, while overlooking the internal structure of newspaper
companies. Organizational strategy in online organizations has been shown, in this
research, to have a clear impact on organizational evolution. A better understanding
of the internal context from which these strategies emerged would reveal additional
insight into the development and subsequent effect of organizational strategy. For
instance, researchers should examine how internal decision makers view online
partnerships and hyperlink relationships in order to assess how these relationships
compare to traditional alliances and partnerships. Furthermore, scholars examining
alliances and partnerships should study the sources of online partnerships to
understand whether they are formed with the same consideration and weight as other
types of alliances.
265
The Nature and Death of Organizational Forms
The research presented in this study examined the development of
organizational forms, but did not extensively consider the nature of individual
organizational forms. A number of previous studies have sought to define the core
characteristics of organizational forms, yet no cohesive theory of forms yet exists
(Polos et al., 2002). As the online news community continues to develop, the forms
within this community will continue to evolve. It is unclear, however, how forms
transition from one state to another, or at what point an existing organizational form
is said to have died. The rapid pace of change in online environments provides a
fertile ground for studying both the boundaries of organizational forms, and the death
of organizations.
Conclusion
This dissertation provides a new perspective for examining the
transformation of organizations. The findings on organizational emergence,
legitimation and transformation suggest that the process of speciation is a powerful
explanatory mechanism for understanding how new organizations emerge. In
addition, this study has demonstrated the importance of considering strategy as a
driver of organizational transformation. In addition, this study introduces a new tool
for archival Internet research, and demonstrates the applicability of longitudinal
network analysis as a tool for examining community development over time.
266
Furthermore, this study developed a better understand of the link economy,
and provides a detailed examination of the development of online news media.
Building on previous literatures on organizational ecology and journalism studies,
this research examines the increasing pace of organizational innovation as a result of
new information communication technology. This dissertation provides substantive
implications for understanding how new technology drives the creation of new
populations of organizations, and how existing organizations struggle to reinvent
themselves.
267
BIBLIOGRAPHY
Abbate, J. (1999). Inventing the Internet. Cambridge, MA: MIT Press.
Abdulla, R. A., Garrison, B., Salwen, M., Driscoll, P., & Casey, D. (2002). Online
news credibility. In M. Salwen, B. Garrison & P. Driscoll (Eds.), Online
News and the Public. Mahwah, NJ: Lawrence Erlbaum Associates.
Ackland, R., O'Neil, M., Bimber, B., Gibson, D., & Ward, S. (2006). New methods
for studying online environmental-activist networks. Paper presented at the
Sunblet Social Network Conference, Vancouver.
Agger, M. (2008). Blogging for Dollars: How do bloggers make money? Slate.
Aldrich, H. (1979). Organizations and Environments. Englewood Cliffs, NJ:
Prentice-Hall.
Aldrich, H. (1999). Organizations Evolving. Thousand Oaks, CA: Sage.
Aldrich, H., & Fiol, C. M. (2007). Fools Rush In? The Institutional Context of
Industry Creation. In A. Cuervo, D. Ribeiro & S. Roig (Eds.),
Entrepreneurship. New York: Springer Berlin Heidelberg.
Aldrich, H., & Ruef, M. (2006). Organizations Evolving. London: SAGE
Publications.
Alonzo, J. S. (2005). Restoring the Ideal Marketplace: How recognizing bloggers as
journalists can save the press. New York University Journal of Legislative
and Public Policy, 9, 751-780.
Althaus, S. L., & Tewksbury, D. (2000). Patterns of Internet and Traditional News
Media Use in a Networked Community. Political Communication, 17, 24-45.
Amburgey, T. L., Kelly, D., & Barnett, W. (1993). Resetting the clock: The
dynamics of organizational change and failure. Administrative Science
Quarterly, 38(1), 51-73.
Amurgey, T. L., Kelly, D., & Barnett, W. (1993). Resetting the clock: The dynamics
of organizational change and failure. Administrative Science Quarterly,
38(1), 51-73.
Amurgey, T. L., & Rao, H. (1996). Organizational Ecology: Past, Present, and
Future Directions. The Academy of Management Journal, 39(5), 1265-1286.
268
Anderson, P. C., & Tushman, M. L. (1990). Technological Discontinuities and
Dominant Designs: A Cyclical Model of Technological Change.
Administrative Science Quarterly, 35.
Andrew, S. A. (2009). Regional integration through contracting networks: An
empirical analysis of institutional collection action framework. Urban Affairs
Review, 44(3), 378-402.
Anonymous. (2009a). Interview with author, Chicago Tribune. Chicago, IL.
November, 2009.
Anonymous. (2009b). Interview with author, The Washington Post. Washington,
DC. October 2009.
Anonymous. (2009c). Interview with author, Las Vegas Sun. Las Vegas, NV.
November, 2009.
Anonymous. (2009d). Interview with author, The Boston Globe. Boston, MA.
August, 2009.
Arends, B. (2009). Will the news survive? MarketWatch.
Arms, W., Aya, S., Dmitriev, P., Kot, B., Mitchell, R., & Walle, L. (2006). A
Research Library Based on the Historical Collections of the Internet Archive.
D-Lib Magazine, 12(2).
Arms, W., Huttenlocher, D., Kleinberg, J., Macy, M., & Strang, D. (2006). From
Wayback Machine to Yesternet: New Opportunities for Social Science.
Astley, W. G. (1985). The Two Ecologies: Populations and community perspectives
on organizational evolution. Administrative Science Quarterly, 30(2), 18.
Bacharach, S. B., Bamberger, P., & Sonnenstuhl, W. J. (1996). The Organizational
Transformation Process: The Micropolitics of Dissonance Reduction and the
Alignment of Logics of Action. Administrative Science Quarterly, 41(3),
477-506.
Bakkalbasi, N., Bauer, K., Glover, J., & Wang, L. (2006). Three Options for Citation
Tracking: Google Scholar, Scopus and Web of Science. Biomedical Digital
Libraries, 3 (7).
Baldasty, G. J. (1992). The Commercialization of News in the Nineteenth Century.
Madison, Wisconsin: The University of Wisconsin Press.
269
Barabasi, A.-L. (2002). Linked: The New Science of Networks. Cambridge, MA:
Perseus Publishing.
Barnett, W., & Carroll, G. R. (1987). Competition and mutualism among early
telephone industries. Administrative Science Quarterly, 32(3), 22.
Barnett, W., & Carroll, G. R. (1995). Modelling Internal Organizational Change.
Annual Review of Sociology, 21(1), 217-236.
Barnhurst, K. G., & Neron, J. (2001). The Form of News: A history. New York City,
NY: The Guilford Press.
Barnhurt, K. G. (2002). New Georgaphy & Monopoly: the form of reports on US
newspaper internet sites. Journalism Studies, 3(4), 477-489.
Barrett, S. C. H. (1989). Mating System Evolution and Speciation in Heterstylous
Plants. In D. Otte & J. A. Endler (Eds.), Speciation and its Consequences.
Sunderland, MA: Sinauer Associates.
Baum, J. A. C., & McKelvey, B. (1999). Variations in organization science: In
honor of David T. Campbell. Thousand Oaks, CA: SAGE Publications.
Baum, J. A. C., & Oliver, C. (1991). Institutional Linkages and Organizational
Mortality. Administrative Science Quarterly, 36(2), 32.
Baum, J. A. C., & Oliver, C. (1992). Institutional embeddedness and the dynamics of
organizational populations. American Sociological REview, 57(4), 20.
Baum, J. A. C., & Oliver, C. (1996). Toward an Insitutional Ecology of
Organizational Founding. The Academy of Management Journal, 39(5),
1378-1427.
Baum, J. A. C., & Singh, J. V. (1994a). Evolutionary Dynamics of Organizations.
New York: Oxford University Press.
Baum, J. A. C., & Singh, J. V. (1994b). Organizational niche overlap and the
dynamics or organizational founding. Organization Science, 5(4), 483-501.
Baum, J. A. C., & Singh, J. V. (1994c). Organizational niche overlap and the
dynamics or organizational mortality. American Journal of Sociology, 100(2),
346-380.
Baym, G. (2010). From Conkrite to Colbert: The Evolution of Broadcast News.
London, UK: Paradigm Publishers.
270
Bergelman, R. A. (1991). Intraorganizational Ecology of Strategy Making and
Organizational Adaptation: Theory and Field Research. Organization
Science, 2(3), 239-262.
Berners-Lee, T., Hall, W., Hendler, J., Shadbolt, N., & Weitzner, D. J. (2006).
Creating a Science of the Web. Science, 313, 769-771.
Betton, J., & Dess, G. G. (1985). The Application of Population Ecology Models to
the Study of Organizations. Academy of Management Review, 10(4), 750-
757.
Boczkowski, P. J. (2004a). Digitizing the News: Innovation in Online Newspapers.
Cambridge, MA: MIT Press.
Boczkowski, P. J. (2004b). The Processes of Adopting Multimedia and Interactivity
in Three Online Newsrooms. Journal of Communication, 54(2), 16.
Boczkowski, P. J. (Forthcoming). Newspaper Culture and Technical Innovation,
1980-2005. In W. R. Neuman (Ed.), Media, Technology, and Society: The
Challenge of Digital Convergence. Ann Arbor, MI: University of Michigan
Press.
Boczkowski, P. J., & Ferris, J. A. (2005). Multiple media, convergenct processes and
divergent products: Organizational innovation in digital media production at
a European firm. The Annals of the American Academy of Political and
Social Science, 597(1), 16.
Boisot, M., & Child, J. (1988). The Iron Law of Fiefs: Bureaucratic Failure and the
Problem of Governance in the Chinese Economic Reforms. Administrative
Science Quarterly, 33(4), 507-527.
Borgatti, S. P. (2002). NetDraw: Graph Visualization Software. Harvard, MA:
Analytic Technologies.
Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002). UCINET 6.0. Natick, MA:
Analytic Technologies.
Bowman, S., & Willis, C. (2003). We Media: How audiences are shaping the future
of news and information. Washington, DC: The Media Center at the
American Press Institute.
boyd, d. m., & Ellison, N. B. (2008). Social Network Sites: Definition, History and
Scholarship. Journal of Computer Mediated Communication, 13(1).
271
Brand, S. (1985). Keep designing: How the information economy is being created
and shaped by the hacker ethic. The Whole Earth Review, 44-49.
Brittain, J. W. (1994). Density-Independent Selection and Community Evolution. In
J. A. C. Baum & J. V. Singh (Eds.), Evolutionary Dynamics of
Organizations. Oxford, UK: Oxford University Press.
Brittain, J. W., & Wholey, D. H. (1988). Competition and coexistence in
organizational communities: Population dynamics in electronics components
manufacturing. In G. R. Carroll (Ed.), Ecological Models of Orgnaizations
(pp. 195-222). Cambridge, MA: Ballinger PRess.
Brown, S. L., & Eisenhardt, K. (1997). The Art of Continuos Change: Linking
Complexity Theory and Time-Paced Evolution in Relentlessly Shifting
Organizations. Administrative Science Quarterly, 42(1), 1-34.
Bruderer, E., & Singh, J. V. (1996). Organizational Evolution, Learning and
Selection: A genetic-algorithm-based model. Academy of Management
Journal, 1996(39), 5.
Bryant, J. A., & Monge, P. R. (2008). The Evolution of the Children's Television
Community: 1953-2003. International Journal of Communication, 2, 160-
192.
Burgelman, R. (1983). A process model of internal corporate venturing in the
diversified major firm. Administrative Science Quarterly, 28(2), 223-244.
Burgelman, R. A. (1983). A process model of internal corporate venturing in the
diversified major firm. Administrative Science Quarterly, 28(2), 223-244.
Bush, G. (1975). Modes of Animal Speciation. Annual Review of Ecology and
Systemics, 6, 339-364.
Butts, C. T. (2008). network: A package for managing relational data in R. Journal
of Statistical Software, 24(2).
Byrant, J. A. (2003). From Networks to Nickelodeon to Noggin: A communication
networks perspective on the evolution of the children's television community.
University of Southern California, Los Angeles, CA.
Cameron, A. C., & Trivedi, P. K. (1986). Econometric models based on count data:
Comparisons and applicatinos of some estimators and tests. Journal of
Applied Econometrics(1), 29-53.
272
Campbell, D. T. (1965). Variation and selective retention in socio-cultural evolution.
In H. R. Barringer, G. I. Blanksten & R. W. Mack (Eds.), Social Change in
Developing Areas: A Reinterpretation of Evolutioinary Theory (pp. 19-48).
Cambridge, MA: Schenkman.
Carroll, G. R. (1984). Organizational Ecology. Annual Review of Sociology, 10, 71-
93.
Carroll, G. R. (1985). Concentration and specialization: Dyanmics of niche width in
populations of organizations. American Journal of Sociology, 90(6), 1262-
1283.
Carroll, G. R., & Delacroix, J. (1982a). Organizational Mortality in the Newsaper
Industries of Argentina and Ireland: An Ecological Approach. Administrative
Science Quarterly, 27(2), 169-198.
Carroll, G. R., & Delacroix, J. (1982b). Organizational Mortality in the Newspaper
Industries of Argentina and Ireland: An Ecological Approach. 1982, 27(2),
169-198.
Carroll, G. R., & Delacroix, J. (1983). Organizational Foundings: An Ecological
Study of the Newspaper Industries of Argentina and Ireland. Administrative
Science Quarterly, 28(2), 274-291.
Carroll, G. R., Dobrev, S., & Swaminathan, A. (2002). Organizational processes of
resource partitioning. Research in Organizational Behavior, 24, 40.
Carroll, G. R., & Hannan, M. (1989). Density Dependence in the Evolution of
Populations of Newspaper Organizations. American Sociological Review,
54(4), 1989.
Carroll, G. R., & Hannan, M. (2000). The Demography of Corporations and
Industries. Princeton, NJ: Princeton University Press.
Cheng, J. L. C., & Kesner, I. F. (1997). Organizational Slack and Response to
Environmental Shifts: The Impact of Resource Allocation Patterns. Journal
of Management, 23(1), 1-18.
Chon, B. S., Choi, J. H., Barnett, G., Danowski, J. A., & Joo, S.-H. (2003). A
Structural Analysis of Media Convergence: Cross-Industry Mergers and
Acquisitions in the Information Industries. Journal of Media Economics,
16(3), 141-157.
273
Cohen, J. (1960). A coefficient of agreement for nominal scales. Eduational and
Psychological Measurement, 20, 37-46.
Cohen, J. (1968). Weighted kappa: Nominal scale agreement with provision for
scaled disagreement or partial credit. Psychological Bulletin, 70, 213-220.
Contractor, N., S., Wasseman, S., & Faust, K. (2006). Testing Multitheoretical,
Multilevel Hypotheses About Organizational Networks: An Analytical
Framework and Empirical Example. Academy of Management Journal,
31(3), 681-703.
Coyne, J. A., & Orr, H. A. (2004). Speciation. Sunderland, MA: Sinauer Associates.
d'Haenens, L., Jankowski, N., & Heuvelman, A. (2004). News in Online and Print
Newspapers: Differences in Reader Consumption and Recall. New Media and
Society, 6(3), 363-382.
Darwin, C. (2003). The Origin of Speices. New York City: Signet Classic.
Deeds, D. L., Mang, P. Y., & Frandsen, M. L. (2004). The Influence of Firms' and
Industries' Legitimacy on the Flow of Capital into High-Technology
Ventures. Strategic Organization, 2(1), 9-34.
Dejean, F., Gond, J.-P., & Leca, B. (2004). Measuring the Unmeasured: An
institutional entrepreneur strategy in an emerging industry. Human Relations,
57(6), 741-764.
Delacroix, J., & Carroll, G. R. (1983). Organizational Foundings: An Ecological
Study of the Newspaper Industries of Argentina and Ireland. Administrative
Science Quarterly, 28(2), 274-291.
Delacroix, J., & Rao, H. (1994). Externalities and Ecological Theory: Unbundling
Density Dependence. In J. A. C. Baum & J. V. Singh (Eds.), Evolutionary
Dynamics of Organizations. Oxford, UK: Oxford University Press.
Delmar, F., & Shane, S. (2004). Legitimating First: Organizing activities and the
survival of new ventures. Journal of Business Venturing, 19, 385-410.
Demers, D. P. (1996). Corporate Newspaper Structure, Profits, and Organizational
Goals. The Journal of Media Economics, 9(2), 24.
Demers, D. P. (1998). Sturcutral pluralism, corporate newspaper structure and news
source peceptions: Another test of teh editorial vigor hypothesis. Journalism
and Mass communication Quarterly, 75(3), 21.
274
Demers, D. P., & Merskin, D. (2000). Corporate News Structure and the Managerial
Revolution. Journal of Media Economics, 13(2), 19.
DiMaggio, P. T. (1994). The challenge of community evolution. In J. A. C. Baum &
J. V. Singh (Eds.), Evolutionary dynamics of organizations (pp. 444-450).
New Yortk: Oxford University Press.
DiMaggio, P. T., & Powell, W. W. (1983). The Iron Cage Revisited: Institutional
isomorphism and collective rationality in organizational fields. American
Sociological Review, 48(2), 14.
Dimmick, J. W. (2003). Media Competitions and Coexistence: The Theory of the
Niche. Mahwah, NJ: Lawrence Equilibrium.
Dimmick, J. W., Chen, Y., & Li, Z. (2004). Competition Between the Internet and
Traditional News Media: The Gratification-Opportunities Niche Dimension.
The Journal of Media Economics, 17(1), 19-33.
Dysart, J. (2002). Making the most of promotional links. Public Relations Tactics,
9(11), 21-22.
Edmonds, R., Appelbaum, J., Morgan, J., Olmstead, K., Guskin, E., Holcomb, J., et
al. (2010). The State of the News Media: An annual report on American
journalism. Washington DC: Pew Project for Excellence in Journalism.
Eisenhardt, K., & Tabrizi, B. N. (1995). Accelerating Adaptive Processes: Product
Innovation in the Global Computer Industry. Administrative Science
Quarterly, 40(1).
Eldredge, N., & Gould, S. J. (1972). Punctuated equilibria: an alternative to phyletic
gradulism in models in paleobiology. Models of Paleobiology.
Epstein, E. (1974). News from Nowhere: Television and the News. New York City:
Vintage Books.
Everitt, B. S., & Hothorn, T. (2006). A handbook of statistical analyses using R.
Boca Raton, FL: Chapman & Hall/CRC.
Fetterly, D., Manasse, M., Najrok, M., & Wiener, J. L. (2004). A Large-Scale Study
of the Evolution of Web Pages. Software - Practice and Experience, 34, 213-
237.
Fishman, M. (1980). Manufacturing the News. Austin, TX: Manufacturing the News.
275
Flayhan, D. P. (2001). Cultural studies and media ecology: Meyrowitz's medium
theory and Carey's cultural studies. Atlantic Journal of Communication, 9(1),
21-44.
Fombrun, C. J. (1988). Crafting an institutionally informed ecology of organizations.
In G. R. Carroll (Ed.), Ecological Models of Organizations (pp. 223-239).
Cambridge, MA: Ballinger.
Ford, E. H. (1954). Colonial Pamphleteers. In E. H. Ford & E. Emery (Eds.),
Highlights in the History of the American Press. Minneapolis, MN:
University of Minnesota Press.
Freeman, J. H., & Lomi, A. (1994). Resource partitioning and the foundings of
banking cooperatives in Italy. In J. A. C. Baum & J. V. Singh (Eds.),
Evolutionary Dynamics of Organizations (pp. 269-293). New York City, NY:
Oxford University Press.
Gans, H. (1972). The Famine in American Mass-Communication Research:
Comments on Hirsch, Tuchman and Gecas. The American Journal of
Sociology, 77(4), 9.
Gans, H. (1979). Deciding What's News: A Study of CBS Evening News, NBS Nightly
News and Time. New York City: Pantehon Books.
Gavrilets, S., Li, H., & Vose, M. D. (1998). Rapid parapatric speciation on holey
adaptive landscapes. Proceedings of the Royal Society of London, 263, 1483-
1489.
Gioia, D. A. (1998). From individual to organizational identity. In D. A. Whetten &
P. C. Godfrey (Eds.), Identity in Orgnaizations: Building Theory through
Conversations. Thousand Oaks, CA: Sage Publications.
Glaser, B. J., & Strauss, A. L. (1967). The Discoery of Grounded Theory. Chicago:
Aldine.
Goldenberg, E. (1975). Making the Papers: The Access of Resource-poor Groups to
the Metropolitan Press. New York City: Lexington Press.
Gonzalez-Bailon, S. (2009). Opening the balck box of link formation: Social factors
underlying the structure of the Web. Social Networks, 31(4), 271-280.
Goodreau, S. M. (2007). Advances in exponential random graph (p*) models applied
to a large social network. Socai Networks, 29, 231-248.
276
Gould, S. J. (2007). Punctuated Equilibrium. Cambridge, MA: The Belknap Press.
Gould, S. J., & Eldredge, N. (1977). Punctuated equilibria: The tempo and mode of
evolution reconsidered. Paleobiology, 3, 115-151.
Grant, V. (1963). The Origin of Adaptations. New York City, NY: Columbia
University Press.
Green, H. (2009). With 15.5 Million Active Blogs, New Technorati Data Shows that
Blogging Growth Seems to be Peaking. Business Week.
Greer, J. D., & Mensing, D. (2006). The Evolution of Online Newspapers: A
Longitudinal Content Analysis, 1997-2003. In X. Li (Ed.), Internet
Newspapers: The Making of a Mainstream Medium (pp. 13-32). Mahwah,
NJ: Lawrence Erlbaum Associates.
Halavais, A. (2008). The Hyperlink as an Organizing Principle. In J. Turow & L.
Tsui (Eds.), The Hyperlinked Society: Questioning Connections in the Digital
Age. Ann Arbor, MI: The University of Michigan Press.
Hall, C. M. (2006). The Impact of Tourism Knowledge: Google Scholar, citations
and the opening up of academic space. e-Review of Tourism Reserch, 4(5).
Handcock, M., Hunter, D. R., Butts, C. T., Goodreau, S. M., & Morris, M. (2003).
statnet: Software tools for the Statistical Modeling of Network Data.
Handcock, M., Hunter, D. R., Butts, C. T., Goodreau, S. M., & Morris, M. (2008).
statnet: Software tools for the representation, visualization, analysis and
simulation of network data. Journal of Statistical Software, 24(1).
Hannan, M., Carroll, G. R., & Polos, L. (2003). The organizational niche.
Sociological Theory, 21(4), 32.
Hannan, M., & Freeman, J. H. (1977). The population ecology of organizations.
American Journal of Sociology, 85, 26.
Hannan, M., & Freeman, J. H. (1984). Structural intertia and organizational change.
American Sociological Review, 49, 16.
Hannan, M., & Freeman, J. H. (1986). Where do organizational forms come from?
Sociological Forum, 1(1), 50-72.
277
Hannan, M., & Freeman, J. H. (1988). The Ecology of Organizational Mortality:
American Labor Unions, 1836-1985. The American Journal of Sociology,
94(1), 28.
Hannan, M., & Freeman, J. H. (1989). Organizational ecology. Cambridge, MA:
Harvard University Press.
Haveman, H. A. (1992). Between a Rock and a Hard Place: Organizational change
and Performance under Conditions of Fundamental Environmental
Transformation. Administrative Science Quarterly, 37, 48 - 75.
Haveman, H. A. (1995). The Demographic Metabolism of Organizations: Industry
Dyanmics, Turnover and Tenure Distributions. Administrative Science
Quarterly, 40, 586 - 618.
Hawley, A. H. (1986). Human Ecology: A Theoretical Essay. Chicago: University of
Chicago Press.
Hodgson, G. M. (2010). Darwinian coevolution of organizations and the
environment. Ecological Economics, 69, 700-706.
Hunter, D. R., Handcock, M., Butts, C. T., Goodreau, S. M., & Morris, M. (2008).
ergm: A package to fit, simulate and diagnose exponential-family models for
networks. Journal of Statistical Software, 24(3).
Hutchins, R. M., Chafee, Z., Clark, J. M., Dickinson, J., Hocking, W. E., Lasswell,
H. D., et al. (1947). A Free and Responsible Press: A General Report on
Mass Communication - Newspapers, Radio, Motion Pictures, Magazines, and
Books. Chicago, IL: The University of Chicago.
IBIS. (2008). Newspaper Industry in the United States. New York City: IBISWorld.
Internet Overtakes Newspapers as News Outlet. (2008).). Washington DC: Pew
Center for the People & the Press.
Jarvis, J. (2008, June 30). AP took it to the wire but needs to rethink its role. The
Guardian, p. 6.
Javalgi, R. G., Todd, P. R., & Scherer, R. F. (2005). The Dynamics of Global E-
Commerce: An organizational ecology perspective. International Marketing
Review, 22(4), 420-435.
278
Johnson, V. (2007). What is organizational imprinting? Cultural entrepreneurship in
the founding of the Paris Opera. American Journal of Sociology, 113(1), 97-
127.
Katz, J. (1997). Virtuous Reality. New York, NY: Random House.
Kauffman, S. A. (1993). Origins of Order: Self-organization and Selection in
Evolution. Oxford: Oxford University Press.
Kelly, D., & Amburgey, T. L. (1991). Organizational Intertia and Momentum: A
dynamic model of strategic change. Academy of Management Journal, 34(3),
591-612.
King, G. (1988). Statistical models for political science event counts: Bias in
conventional procedures and evidence for the exponential Poission regression
model. American Journal of Political Science, 32(838-863).
Klinenberg, E. (2005). Convergence: News Proudction in a Digital Age. The Annals
of the American Academy of Political and Social Science, 597(1), 19.
Kobre, S. (1944). The Development of the Colonial Newspaper. Pittsburgh, PA:
Columbia University Press.
Kohut, A. (2008). Audience Segments in a Changing News Environment.
Washington, D.C.: The Pew Research Center for the People and the Press.
Koka, B. R., Madhavan, R., & Prescott, J. E. (2006). The evolution of interfirm
networks: Environmental effects on patterns of network change. Academy of
Management Review, 31(3), 17.
Kollock, P., & Smith, M. (1999). Communities in Cyberspace. London: Routledge.
Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for
categorical data. Biometrics, 33, 159-174.
Lasica, J. D. (2002). Blogging as a Form of Journalism. Online Journalism Review
Retrieved April 14, 2010, from http://www.ojr.org/ojr/lasica/1019166956.php
Lasica, J. D. (2003). Benefits blogging brings to news outlets. Nieman Reports, 57,
72.
Lawless, M. W., & Anderson, P. C. (1996). Generational technology change: Effects
of innovation and local rivalrly on performance. Academy of Management
Journal, 39(5), 33.
279
Lazer, D., Pentland, A., Adamic, L. A., Aral, S., Barabasi, A.-L., Brewer, D., et al.
(2009). Computational Social Science. Science, 323, 721-723.
Leblebici, H., Salancik, G. R., Copay, A., & King, T. (1991). Institutional Change
and the Transformation of Interorganizational Fields: An Organizational
History of the U.S. Radio Broadcasting Industry. Administrative Science
Quarterly, 36(3), 333-363.
Lee, A. M. (1937). The Daily Newspaper in America: The Evolution of a Social
Instrument. New York City, NY: The Macmillan Company.
Lee, S. (2008). The Coevolution of Multmodal, Multiplex, and Multilevel
Organizational Networks in Development Communities. Unpublished
Dissertation, University of Southern California, Los Angeles.
Levinthal, D. A. (1998). The Slow Pace of Rapid Tehcnological Change: Gradualism
and Punctuation in Technological Change. Industrial and Corporate Change,
7(2), 1998.
Lewis, T. (1992). "A Godlike Presence": The Impact of Radio on the 1920s and
1930s. Magazine of History, 6, 26-33.
Li, C. (2007). How Consumers Use Social Networks. Forrester Research.
Li, X. (2006). Introduction. In X. Li (Ed.), Internet Newspapers: The Making of a
Mainstream Medium (pp. 13-32). Mahwah, NJ: Lawrence Erlbaum
Associates.
Lum, C. M. K. (2006). Notes Toward an Intellectual History of Media Ecology. In C.
M. K. Lum (Ed.), Perspectives on Culture, Technology and Communication:
The Media Ecology Tradition. Cresskill, NJ: Hampton Press, Inc.
Lusher, D., & Ackland, R. (2008). A relational hyperlink analysis of an online social
movement: Asylum seeker advoacy groups in Australia. Australian National
University.
March, J. G. (1994). The Evolution of Evolution. In J. A. C. Baum & J. V. Singh
(Eds.), Evolutionary Dynamics of Organizations (pp. 39-49). New York City,
NY: Oxford University Press.
Marquis, A. G. (1984). Written on the Wind: The Impact of Radio during the 1930s.
Journal of Contemporary History, 19(3), 385-416.
280
Matheson, D. (2004). Weblogs and the epistemology of the news: some trends in
online journalism. New Media and Society, 6(4), 443-468.
Mayr, E. (1982). Speciation and macroevolution. Evolution, 36(6), 1119-1132.
McCluhan, H. M. (1962). The Gutenberg galaxy. Toronto, Canada: University of
Toronto Press.
McCullagh, P., & Nelder, J. A. (2003). Generalized Linear Models. London, UK:
Chapman & Hall / CRC.
McKelvey, B. (1982). Organizational Systematics. Berkley, CA: University of
California Press.
McKelvey, B., & Baum, J. A. C. (1999). Donald T. Campbell's Evolving Influence
on Organization Science. In J. A. C. Baum & B. McKelvey (Eds.), Variations
in Organization Science: In Honor of Donald T. Campbell. Thousand Oaks,
CA: SAGE Publications.
McKendrick, D. G., & Carroll, G. R. (2001). On the Genesis of Organizational
Forms: Evidence from the Market for Disk Arrays. Organization Science,
12(6), 661-682.
McKendrick, D. G., Jaffe, J., Carroll, G. R., & Khessina, O. M. (2003). In the bud?
Array producers as a (possibly) emergent organizational form. Administrative
Science Quarterly, 48(1), 60-93.
Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: Formal structures
as myth and ceremony. American Journal of Sociology, 83, 340-363.
Miner, A. S. (1991). Organizational evolution and the social ecology of jobs.
American Sociological Review, 56, 772-785.
Miner, A. S. (1994). Seeking Adaptive Advantage: Evolutionary Thoery and
Managerial Action. In J. A. C. Baum & J. V. Singh (Eds.), Evolutionary
Dynamics of Organizations. New York: Oxford Univeristy Press.
Miner, A. S., Amurgey, T. L., & Stearns, T. M. (1990). Interorganizational linkages
and poplation dynamics: Buffering and tranformational shields.
Administrative Science Quarterly, 35(4), 28.
Minkoff, D. C. (1993). From Service Provision to Institutional Advocacy: The
Shifting Legitimacy of Organizational Forms. Social Forces, 72.
281
Mitchell, A., & Rosenstiel, T. (2009). The State of News Media. Washington, DC:
Project for Excellence in Journalism.
Mitchelstein, E., & Boczkowski, P. J. (2009). Between tradition and change: A
review of recent research on online news production. Journalism, 10(5), 562-
586.
Monge, P. R., & Contractor, N., S. (2003). Theory of Communication Networks.
New York City: Oxford University Press.
Monge, P. R., Heiss, B., & Margolin, D. (2008). Network evolution in organizational
communities. Communication Theory.
Monge, P. R., Lee, S., Fulk, J., Weber, M., Shen, C., Schulz, C., et al. (2009 in
press). Research Methods for Studying Evolutionary and Ecological
Processes in Organizational Communities. Management Communication
Quarterly.
Morgan, G. (1986). Images of Organizations. Beverly Hills, CA: Sage Publications.
Mott, F. L. (1950). American Journalism. New York City, NY: The Macmillan
Company.
Murphy, J., Hashim, N. H., & O'Connor, P. (2008). Take Me Back: Validating the
Wayback Machine. Journal of Computer Mediated Communication, 13, 60-
75.
Murray, J. (2007). Circulation Facts, Figures & Logic (Publication. Retrieved
December 1, 2008, from Newspaper Association of America:
Murray, S. (2003). Media convergence's third wave: Content streaming.
Convergence, 9(1), 8-18.
Nelson, R. R., & Winter, S. G. (1982). An Evolutionary Theory of Economic change.
Boston: Belknap Press.
O'Reilly, T. (2005). What is Web 2.0. Design Patterns and Business Models for the
Next Generation of Software, 30.
O’Murchu, I., Breslin, J. G., & Decker, S. (2004). Online social and business
networking communities. Paper presented at the ECAI 2004 Workshop on
Application of Semantic Web Technologies to Web Communities.
282
Padgett, J. F. (2001). Organizational Genesis, Identity, and Control: The
Transformation of Banking in Renaissance Florence. In J. E. Raush & A.
Casella (Eds.), Networks and Markets. New York City: Russell Sage
Foundation.
Park, H. W. (2003). Hyperlink network analysis: A new method for the study of
social strcuture on the web. Connections, 25(1), 49-61.
Park, H. W., Barnett, G., & Nam, I.-Y. (2002). Hyperlink-affiliation network
structure of top Web sites: Examining affiliates with hyperlink in Korea.
Journal of the American Society for Information Science and Technology,
53(7), 592-601.
Park, H. W., & Jankowski, N. W. (2008). A Hyperlink Network Analysis of Citizen
Blogs in South Korea Politics. Javanost: The Public, 15(2), 57-74.
Park, H. W., & Thelwall, M. (2008). Link analysis: Hyperlink patterns and social
structure on politician's Web sites in South Korea. Qual Quant, 43, 687-697.
Parsons, P. R. (2008). Blue Skies: A history of cable television. Philadelphia, PA:
Temple University Press.
Patterson, T. (2007). Creative destruction: An exploratory look at news on the
Internet. Boston: Joan Shorenstein Center on the Press, Politics and Public
Policy.
Pavlik, J. V. (1999). New Media and News: Implications for the Future of
Journalism. New Media Society, 1(1), 54-59.
Pentland, A., Fletcher, R., & Hasson, A. (2004). DakNet: Rethinking connectivity in
developing nations. Computer, 37(1), 78-83.
Perrow, C. (1991). A Society of Orgnaizations. Theory and Society, 20(6), 725-762.
Phan, D. D., Schmidt, M. B., & Chen, J. Q. (2010). Organizational Ecology Success
Factors in the Business: A Case Study of Fingerhut, Inc. Information Systems
Management, 27, 82-91.
Pickett, C. M. (1977). Voices of the Past: Key Documents in the History of American
Journalism. Columbus, Ohio: Grid Publishing.
Pollock, T. G., & Rindova, V. P. (2003). Media Legitimation Effects in the Market
for Initial Public Offerings. The Academy of Management Journal, 46(5),
631-642.
283
Polos, L., Hannan, M., & Carroll, G. R. (2002). Foundations of a theory of social
forms. Industrial and Corporate Change, 11(1).
Postman, N. (1979). Teaching as a conserving activity. New York: Delta.
Postman, N. (1982). The Disappearance of Childhood. New York: Delacorte Press.
Postman, N. (1985). Amusing Ourselves to Death. New York: Viking Press.
R: A language and environment for statistical computing (Publication (2010). from R
Foundation for Statistical Computing: http://www.R-project.org
Rainie, L., & Horrigan, J. (2005). Trends 2005: Internet Evolution. Washington, DC:
Pew Internet & American Life Project.
Rao, H. (1998). Caveat Emptor: The Construction of Nonprofit Consumer Watchdog
Organizations. The American Journal of Sociology, 103(4), 912-961.
Rao, H. (2007). Interorganizational Ecology. In J. A. C. Baum (Ed.), Companion to
Organizations. Malden, MA: Blackwell Publishing.
Rao, H., & Singh, J. V. (1999). Types of Variation in Organizational Populations:
The Speciation of New Organizational Froms. In J. A. C. Baum & B.
McKelvery (Eds.), Variations in Organization Science (pp. 63-77). Thousand
Oaks, CA: Sage Publications.
Reese, S. D., Ruigliano, L., Hyun, K., & Jeong, J. (2007). Mapping the Blogosphere.
Journalism, 8(3), 235-261.
Regan, T. (1993a). Internet as a journalist's tool. Nieman Reports, 47, 26-30.
Regan, T. (1993b). Weblogs threaten and inform traditional journalism. Nieman
Reports, 57, 68.
Regan, T. (2000). Technology is changing journalism. Nieman Reports, 54, 6-10.
Rettberg, J. W. (2008). Blogging. Malden, MA: Polity Press.
Rheingold, H. (2003). Smart Mobs: The Next Social Revolution. Cambridge, Mass.:
Basic Books.
Rieseberg, L. H. (2001). Chromosomal rearrangement and speciation. Trends in
Ecology and Evolution, 16(7), 351-358.
284
Robins, G., Pattison, P., Kalish, Y., & Lusher, D. (2007). An introduction to
exponential rendom graph (p*) models for social networks. Social Networks,
29(2), 19.
Romanelli, E. (1991). The Evolution of New Organizational Forms. Annual Review
of Sociology, 17(1), 79-103.
Romanelli, E., & Tushman, M. L. (1994). Organizational Transformation as
Punctuated Equilibrium: An Empirical Test. Academy of Management
Journal, 37(5), 1141-1166.
Ross, S. S. (1998). Journalists' Use of On-line Technology and Sources. In D. L.
Borden & K. Harvey (Eds.), The Electronic Grapevine: rumor, reputation,
and reporting in the new on-line environment (pp. 143-160). Mahway, NJ:
Erlbaum.
Rossum, G. V. (2010). The Python Language Reference (Vol. 2.6.5). San Francisco,
CA: Python Software Foundation.
Ruef, M. (2000). The emergence of organizational forms: A community ecology
approach. American Journal of Sociology, 106, 57.
Ruef, M., & Scott, W. R. (1998). A Multidimensional Model of Organizational
Legitimacy: Hospital Survival in Changing Institutional Environments.
Administrative Science Quarterly, 43(4), 877-904.
O'Grady v. Superior Court of Santa Clara County, H028579 C.F.R. (2006).
Salancik, G. R. (1977). Commitment and control of organizational behavior and
belief. In B. M. Staw & G. R. Salancik (Eds.), New Directions in
Organizational Behavior (pp. 1-54). New York City, New York: St. Clair
Press.
Schumpeter, E. (1945). The Control of Germany and Japan. The American Economic
Review, 35(3).
Schwarzlose, R. (1987). Newspapers: A reference guide. New York City, NY:
Greenwood Press.
Scott, E. (2004). "Big Media" Meets the "Bloggers": Coverage of Trent Lott's
Remarks at Strom Thurmond's Birthday Party. Cambridge, MA: Harvard
University.
285
Shaaber, M. A. (1954). Forerunners of the Newspaper in America. In E. H. Ford &
E. Emry (Eds.), Highlights in the history of the American Press. Minneapolis,
MN: University of Minnesota Press.
Shirky, C. (2008). Here Comes Everybody. London, England: Penguin Books.
Shudson, M. (1978). Discovering the News: A Social History of American
Newspapers. New York: Basic Books.
Shumate, M., & Dewitt, L. (2008). The north/south divide in NGO hyperlink
networks. Journal of Computer-Mediated Communication, 13, 405-428.
Shumate, M., & Lipp, J. (2008). Connective collective action online: An examination
of the hyperlink network structure of an NGO issue network. Journal of
Computer Mediated Communication, 14, 178-201.
Shumate, M. D. (2003). The coevolution of a population with a community,
organizations and the evironment: The emergence, evolution, and impact of
HIV/AIDS NGOs. University of Southern California, Los Angeles, CA.
Simpson, G. (1944). Tempo and Mode in Evolution. New York City, NY: Columbia
University Press.
Smythe, T. C. (2003). The Gilded Age Press, 1865-1900 (Vol. 4). Westport, CT:
Praeger.
Snijders, T. (2009). Longitudinal Methods of Network Analysis. In B. Meyers (Ed.),
Encyclopedia of Complexity and System Science (pp. 5998-6013). Boston,
MA: Springer Verlag.
Snijders, T., Pattison, P., Robins, G., & Handcock, M. (2006). New specifications for
exponential random graph models. Sociological Methodology, 36(1), 99.
Snijders, T., Steglich, C. E. G., Schweinberger, M., & Huisman, M. (2007a). Manual
for RSiena. Oxford, UK: University of Groningen: ICS / Department of
Sociology; University of Oxford: Department of Statistics.
Snijders, T., Steglich, C. E. G., Schweinberger, M., & Huisman, M. (2007b). Manual
for SIENA version 3.1. Oxford, UK: University of Groningen: ICS /
Department of Sociology; University of Oxford: Department of Statistics.
Snijders, T., Steglich, C. E. G., & Van de Bunt, G. G. (2010). Introduction to actor-
based models for network dynamics. Social Networks(32), 44-60.
286
Stinchcombe, A. (1965). Social Structure and Organizations. In J. G. March (Ed.),
Handbook of Organizations. Chicago: Rand McNally.
Stovall, J. G. (2004). Web Journalism: Practice and Promise of a New Medium.
Boston, MA: Pearson Education.
Suarez, F. F., & Oliva, R. (2005). Environmental change and organizational
transformation. Industrial and Corporate Change, 2005(14), 6.
Suchman, M. (1995). Managing Legitimacy: Strategic and Institutional Approaches.
Academy of Management Review, 20(3), 571-610.
Swaminathan, A. (1995). The proliferation of specialist organizations in the
American wine industry: 1941- 1990. Administrative Science Quarterly,
40(4), 653-680.
Swaminathan, A. (2001). Resource Partitioning and the Evolution of Specialist
Organizations: The Role of Location and Identity in the U.S. Wine Industry.
Academy of Management Journal, 44(6), 1169-1185.
Swidler, A. (1986). Culture in Action: Symbols and Strategies. American
Sociological Review, 51(2), 273-286.
Sylvie, G., & Witherspoon, P. D. (2002). Time, Change and the American
Newspaper. Mahwah, NJ: Lawrence Earlbaum Associates.
Tan, J., & Tan, D. (2005). Environment-Strategy Co-Evoultion and Co-Alignment:
A staged model of Chinese SOEs under transition. Strategic Management
Journal, 26, 141-157.
Tremayne, M. (2004). The Web of Context: Applying Network Theory to the Use of
Hyperlinks in Journalism on the Web. Journalism and Mass Communication
Quarterly, 81(2), 237-253.
Tremayne, M. (2006). Applying Network Theory to the Use of External Links of
News Web Sites. In X. Li (Ed.), Internet Newspapers: the making of a
mainstream medium (pp. 49-64). Malwah, NJ: Lawrence Earlbaum
Associates.
Trends & Numbers: Total Paid Circulation (Publication (2007). Retrieved December
1, 2008, from Newspaper Association of America:
Tuchman, G. (1978). Making News. New York City: Free Press.
287
Turkle, S. (1995). Life on the Screen: Identity in the Age of the Internet. New York:
Simon and Schuster.
Tushman, M. L., & Anderson, P. C. (1986). Technological discontinuities and
organizational environments. Administrative Science Quarterly, 31(3), 27.
Tushman, M. L., & Romanelli, E. (1985a). Organizational evolution: A
metamophosis model of convergence and reorientation. Research in
Organizational Behavior, 7, 52.
Tushman, M. L., & Romanelli, E. (1985b). Organizational Evolution: A
Metamorphosis Model of Convergence and Reorientation. In L. L.
Cummings & B. M. Staw (Eds.), Research in Organizational Behavior (Vol.
7, pp. 171-222). New York City, NY: JAI Press Inc.
Usher, J. M., & Evans, M. G. (1996). Life and Death along Gasoline Alley:
Darwinian and Lamarckian Processes in a Differentiating Population. The
Academy of Management Journal, 39(5), 1428-1466.
Van de Bunt, G. G., & Groenewegen, P. (2007). An actor-oriented dynamic network
approach: The case of interorganizational network evolution. Organizational
Research Methods(10), 463-482.
Van de Ven, A. H., & Garud, R. (1994). The Coevolution of Technical and
Inistutional Events in the Development of an Innovation. In J. A. C. Baum &
J. V. Singh (Eds.), Evolutionary Dynamics of Organizations (pp. 425-443).
New York, NY: Oxford University Press.
van Duijn, M. A. J., Zeggelink, E. P. H., Huisman, M., Stokman, F. N., & Wasseur,
F. W. (2003). Evolution of Sociology Freshmen into a Friendship Network.
Journal of Mathematical Sociology(27), 153-191.
Vaughn, L., & Thelwall, M. (2003a). Scholarly Use of the Web: What are the key
inducers of links to journal Web sites? Journal of the American Society for
Information Science and Technology, 54(1), 490-497.
Vaughn, L., & Thelwall, M. (2003b). Scholarly use of the Web: What are the key
inducers of links to journal web sites? Journal of the American Society for
Information Science and Technology, 54(1), 29-38.
Veglis, A. (2007). Cross-media publishing by U.S. newspapers. The Journal of
Electronic Publishing, 10(2).
288
Ventresca, J. J., & Mohr, J. W. (2002). Archival research methods. In J. A. C. Baum
(Ed.), The Blackwell Companion to Organizations (pp. 805-828). Malden,
MA: Blackwell.
ver Hoef, J. M., & Boveng, P. L. (2007). Quasi-Poisson vs. negative binomial
regression: How should we model overdispersed count data? Ecology,
88(11), 2766-2772.
Wasserman, S., & Faust, K. (1994). Social Network Analysis: Methods and
Application. Boston: Cambridge University Press.
Waterman, D. (2005). Hollywood's Road to Riches. Cambridge, MA: Harvard Press.
Weber, M. (1984). The Protestant Ethic and the Spirit of Capitalism. Longon:
Unwin.
Wellman, B. (1997). An electronic group is vritually a social network. In S. Kiesler
(Ed.), Culture of the Internet. Malhwah, NJ: Lawrence Erlbaum Associates.
White, D. M. (1949). "The Gatekeeper": A case study of the selection of news.
Journalism Quarterly, 27(1), 383-390.
Whittaker, R. H. (1974). Communities and Ecosystems. New York City, NY:
MacMillan Publishing Co.
Wise, K., Bolls, P. D., & Schaefer, S. R. (2008). Choosing and Reading Online
News: How Available Choice Affects Cognitive Processing. Journal of
Broadcasting & Electronic Media, 52(1), 69-85.
World Press Trends. (2007).). Paris: World Association of Newspapers.
World Press Trends 2007. (2007).). Paris: World Association of Newspapers.
Zajac, E., & Westphal, J. D. (2007). Iterorganizational Economics. In J. A. C. Baum
(Ed.), Companion to Organizations. Malden, MA: Blackwell Publishing.
Zelizer, B. (2007). Introduction: On Finding New Ways of Thinking About
Journalism. Political Communication, 24, 111-114.
289
APPENDIX A:
SEED LIST OF WEB SITES FOR HISTORYCRAWL
Table A-1: List of initial Web sites used for HistoryCrawl
Organization Location URL
AnnArbor.com Ann Arbor, MI www.annarbor.com
The Atlanta Journal-
Constitution
Atlanta, GA www.ajc.com
The Austin Statesman Austin TX www.statesman.com
The Baltimore Sun Baltimore, MD www.baltimoresun.com
The Christian Science
Monitor
Boston, MA www.csmonitor.com
The Boston Globe Boston, MA www.boston.com
The Boston Herald Boston, MA www.bostonherald.com
The Buffalo News Buffalo, NY www.home.buffalo.com
The News & Observer Chapel Hill, NC www.newsobserver.com
The Charlotte Observer Charlotte, NC www.charlotteobserver.com
News Today Chennai, China www.newstodaynet.com
The Chicago Sun-
Times
Chicago, IL www.suntimes.com
The Cincinnati
Enquirer
Cincinnati, OH www.news.cincinnati.com
The Cleveland Plain
Dealer
Cleveland, OH www.cleveland.com
The Columbus
Dispatch
Columbus, OH www.dispatch.com
The Dallas Morning
News
Dallas, TX www.dallasnews.com
DFW.com Dallas, TX www.dfw.com
The Denver Post Denver, CO www.denverpost.com
The Detroit Free Press Detroit, MI www.freep.com
The Detroit News Detroit, MI www.detnews.com
The Ft. Lauderdale Sun
Sentinel
Ft. Lauderdale, FL www.sun-sentinel.com
The Virginian Pilot Hampton Roads, VA www.hamptonroads.com
The Hartford Courant Hartford, CT www.courant.com
The Houston Chronicle Houston, TX www.chron.com
The Indianapolis Star Indianapolis, IN www.indystar.com
Newsday Islip, NY www.newsday.com
290
Table A-1: Continued
The Florida Times-
Union
Jacksonville, FL www.jacksonville.com
The Kansas City Star Kansas City, KS www.kansascity.com
The Tennessean Knoxville, TN www.tennessean.com
The Las Vegas Sun Las Vegas, NV www.lasvegassun.com
The Arkansas
Democrat Gazette
Little Rock, AK www.arkansasonline.com
The Guardian London, UK www.guardian.co.uk
The British
Broadcasting
Corporation (BBC)
London, UK www.bbc.co.uk
Reuters London, UK www.reuters.com
Los Angeles Daily
News
Los Angeles, CA www.dailynews.com
The Huffington Post Los Angeles, CA www.huffingtonpost.com
The Courier-Journal Louisville, KY www.courier-journal.com
The Commercial
Appeal
Memphis, TN www.commercialappeal.com
The Miami Herald Miami, FL www.miamiherald.com
The Milwaukee Journal
Sentinel
Milwaukee, WI www.jsonline.com
Google News Mountain View, CA news.google.com
The New Orleans
Times-Picayune
New Orleans, LA www.nola.com
The New York Post New York City, NY www.nypost.com
Investor’s Business
Daily
New York City, NY www.investors.com
Sporting News New York City, NY www.sportingnews.com
Variety New York City, NY www.variety.com
The Wall Street Journal New York City, NY www.wsj.com
The Associated Press New York City, NY www.ap.org
The Oklahoman Oklahoma, City OK www.newsok.com
Omaha World-Herald Omaha, NE www.omaha.com
The Orlando Sentinel Orlando, FL www.orlandosentinel.com
The Palm Beach Post Palm Beach, FL www.palmbeachpost.com
The Philadelphia
Inquirer
Philadelphia, PA www.philly.com
The Arizona Republic Phoenix, AZ www.azcentral.com
Pittsburgh Post-Gazette Pittsburgh, PA www.postgazette.com
291
Table A-1: Continued
The Oregonian Portland, OR www.oregonlive.com
The Richmond Times
Dispatch
Richmond, VA www.timesdispatch.com
The Press-Enterprise Riverside, CA www.pe.com
The Democrat and
Chronicle
Rochester, NY www.democratandchronicle.com
The Sacramento Bee Sacramento, CA www.sacbee.com
The San Antonio Free
Press
San Antonio, TX www.mysanantonio.com
The San Diego Union
Tribune
San Diego, CA www.signonsandiego.com
The San Francisco
Chronicle
San Francisco, CA www.sfgate.com
The San Jose Mercury
News
San Jose, CA www.mercurynews.com
The Orange County
Register
Santa Ana, CA www.ocregister.com
The Seattle Times Seattle, WA www.seattletimes.nwsource.com
St. Louis Post-Dispatch St. Louis, MO www.stltoday.com
The Star Tribune St. Paul, MN www.startribune.com
St. Paul Pioneer Press St. Paul, MN www.twincities.com
The St. Petersburg
Times
St. Petersburg, FL www.tampabay.com
Yahoo! News Sunnyvale, CA news.yahoo.com
The Tampa Bay
Tribune
Tampa Bay, FL www.tampatrib.com
The Star-Ledger Trenton, NJ www.nj.com
The Contra Costa
Times
Walnut Creek, CA www.contracostatimes.com
The Washington Times Washington, DC www.washingtontimes.com
The Record and Herald
News
Woodland Park, NJ www.northjersey.com
292
APPENDIX B:
LIST OF INTERVIEW SUBJECTS
Table B-1: List of Interview Subjects
Interview With Organization Location Date Length
Director, Online Strategy The Boston Globe Boston, MA August, 2009 60 min.
Online Editor The Boston Globe Boston, MA August, 2009 33 min.
Web Producer San Diego Union Tribune San Diego, CA September, 2009 35 min.
Online Editor San Diego Union Tribune San Diego, CA September, 2009 82 min.
VP, Interactive Las Vegas Sun Las Vegas, NV September, 2009 140 min.
Sr. Producer Las Vegas Sun Las Vegas, NV September, 2009 45 min.
Sr. Reporter Las Vegas Sun Las Vegas, NV September, 2009 52 min.
Online Editor Las Vegas Sun Las Vegas, NV September, 2009 36 min.
Producer Las Vegas Sun Las Vegas, NV September, 2009 34 min.
Blogger Las Vegas Sun Las Vegas, NV September, 2009 45 min.
Senior Vice President Bakersfield Californian Bakersfield, CA September, 2009 68 min.
Blogger Bakersfield Californian Bakersfield, CA September, 2009 24 min.
VP, Marketing AnnArbor.com Ann Arbor, MI September, 2009 67 min.
Editor AnnArbor.com Ann Arbor, MI September, 2009 54 min.
Director, Online Strategy The Orange County
Register
Santa Ana, CA September, 2009 60 min.
Sr. Online Editor The Orange Country
Register
Santa Ana, CA September, 2009 25 min.
Online Editor The Orange County
Register
Santa Ana, CA September, 2009 31 min.
SVP, Digital Operations Washington Post Washington, DC October, 2009 82 min.
293
Table B-1: Continued
SVP and Publisher,
Express
Washington Post Washington, DC October, 2009 45 min.
Director, Online Washington Post Washington, DC October, 2009 62 min.
Reporter Washington Post Washington, DC October, 2009 45 min.
Reporter Washington Post Washington, DC October, 2009 61 min.
Director, Online
Communities
CNN Digital Washington, DC October, 2009 62 min.
VP, Strategy and
Operations
New York Times New York City, NY October, 2009 68 min.
Director, Strategy New York Times New York City, NY October, 2009 72 min.
Reporter New York Times New York City, NY October, 2009 34 min.
Jason Fry, Web CMS
Evangelist
Eidos Media New York City, NY October, 2009 71 min.
Online Director The Baltimore Sun Baltimore, MD October, 2009 72 min.
Online Producer The Baltimore Sun Baltimore, MD October, 2009 58 min.
Reporter The Chronicle of Higher
Education
Washington, DC October, 2009 47 min.
Rachel Sklar, Blogger Mediaite New York City, NY October, 2009 80 min.
Blogger The Huffington Post New York City, NY October, 2009 35 min.
Producer MSNBC New York City, NY October, 2009 61 min.
Online Editor MSNBC New York City, NY October, 2009 58 min.
Sr. Online Director MSNBC New York City, NY October, 2009 31 min.
Online Director MSNBC New York City, NY October, 2009 34 min.
Producer NBC.com/TodayShow.com New York City, NY October, 2009 30 min.
Executive Editor iVillage Networks New York City, NY October, 2009 50 min.
Communities Editor Reuters London, UK October, 2009 58 min.
293
294
Table B-1: Continued
Sr. Online Manager The Guardian London, UK October, 2009 65 min.
Communities Editor The Telegraph London, UK October, 2009 60 min.
Sr. Online Editor The Telegraph London, UK October, 2009 32 min.
Editor, Online The Wall Street Journal New York City, NY October, 2009 34 min.
Senior Reporter, Online The Wall Street Journal New York City, NY October, 2009 30 min.
Sr. Producer, Social
Media
West Coast News na November, 2009 67 min.
Manager Online
Operations
The Chicago Tribune Chicago, IL November, 2009 65 min.
Sr. Editor, Online The Chicago Tribune Chicago, IL November, 009 35 min.
Director, Online
Operations
The BBC London, UK November, 2009 72 min.
294
295
APPENDIX C: INTERVIEW PROTOCOL
[PRECEEDED BY IRB PROTOCOL ! Discussion of IRB policies with interview
subject, review IRB release and obtain participant’s signature, discuss recording of
interviews and obtain permission to use digital recorder]
OVERVIEW:
The purpose of this study is to explore the strategies that news organizations have
used to disseminate information online, and to understand how organizations have
linked and partnered with one another online organizations. Theoretically, this
research will help practitioners and researchers alike to understand how
organizations respond to rapid and dramatic technological changes.
In order to understand how news organizations have responded, we will spend the
next hour discussing specifically how your company has responded to changes in the
industry over the past five years. I’ll first ask some general questions about your
general approach to online news and your company’s response to recent industry
changes, and then dig deeper by asking specifically about blogs and social
networking.
INTRODUCTION:
To start, I’d like to learn a little bit more about your role within [INSERT
COMPANY NAME HERE]:
1. Please describe your role and responsibilities at your company:
2. How long have you been working at [INSERT COMPANY NAME HERE]:
GENERAL ONLINE NEWS:
Next, I’d like to talk about [INSERT COMPANY NAME HERE]’s general approach
to the online news industry.
3. [For companies that publish offline] When did your newspaper first begin
publishing online?
4. Can you please describe your company’s general strategy with regards to online
news?
296
5. [For companies that publish offline] How much overlap and interaction is there
between your online and offline operations?
CHANGES IN THE INDUSTRY:
Clearly this is a changing industry. Much has been written in the past few years
about the decline of traditional news organizations, the rise of social media and
social news. I’d like to now talk about [INSERT COMPANY NAME HERE]’s
response to recent industry trends and competitive threats:
6. In your opinion, what are the three largest competitive threats your company has
faced in the past five years? past year?
7. The past five years have been marked by significant technological change in the
news industry. Which new technologies have most dramatically changed the nature
of your business, and how?
8. How has your company responded to these changes? What major initiatives have
you undertaken in the past five years in response to new technology?
8b. [FOLLOWUP TO SPECIFIC CHANGE INITIATIVES –
INTERVIEWER TO FOLLOWUP ON SIGNIFICANT CHANGE
INITIATIVES MENTIONED BY THE INTERVIEWEE]
What was the time frame for that change initiative [FOCUS ON
DURATION]?
SOCIAL MEDIA (BLOGS & SOCIAL NETWORKING):
Lastly, I’d like to talk specifically about [INSERT COMPANY NAME HERE]’s
attitudes towards recent trends in social media.
[PROCEED TO QUESTION 10 IF GOOGLE NEWS IS DISCUSSED AS A
COMPETITIVE THREAT IN QUESTION 6]
9. Google News launched in 2002, and by 2003 it was the top source for online
news. How did [INSERT COMPANY NAME HERE] respond to this threat?
297
[PROCEED TO QUESTION 11 IF WIKIPEDIA IS DISCUSSED AS A
COMPETITIVE THREAT IN QUESTION 6]
10. This past year, a number of major audience trackers, including Nielsen, Pew and
ComScore, have reported the Wikipedia is now one of the most visited sources for
online news. Does [INSERT COMPANY NAME HERE] consider this a major
threat?
10b. [IF YES] How has [INSERT COMPANY NAME HERE] responded, or
how to you plan to respond?
[PROCEED TO QUESTION 12 IF FACEBOOK.COM IS DISCUSSED AS A
COMPETITIVE THREAT IN QUESTION 6]
11. Facebook.com launched in 2003 and has been credited with the growing
dominance of social networking as a media platform. How has [INSERT
COMPANY NAME HERE] adapted to social networking? What is the general
strategic approach to social networking, given that many online consumers use this
space to repost news stories or comment on major events?
[PROCEED TO WRAP-UP IF BLOGS ARE DISCUSSED AS A COMPETITIVE
THREAT IN QUESTION 6]
12. Many newspapers are now adopting blogs as a means of providing rapid updates
to online users through news briefs and live blogging. Other companies invite
readers to participate in blogs, even providing frequent posters with their own blogs.
What role do blogs occupy at [INSERT COMPANY NAME HERE]?
WRAP-UP
We’ve covered quite a bit in this discussion; thank you for your time. Before we wrap
up is there anything else that you would like to mention?
[CONCLUDE INTERVIEW]
298
APPENDIX D: STATNET GOODNESS OF FIT DIAGRAMS 1998 – 2007
Figure D-1: 1998 Goodness-of-Fit Out-Degree
299
Figure D-2: 1998 Goodness-of-Fit In-Degree
Figure D-3: 1999 Goodness-of-Fit Diagrams Out-Degree
300
Figure D-4: 1999 Goodness-of-Fit Diagrams In-Degree
Figure D-5: 2000 Goodness-of-Fit Diagrams Out-Degree
301
Figure D-6: 2000 Goodness-of-Fit Diagrams In-Degree
Figure D-7: 2001 Goodness-of-Fit Diagrams Out-Degree
302
Figure D-8: 2001 Goodness-of-Fit Diagrams In-Degree
Figure D-9: 2002 Goodness-of-Fit Diagrams Out-Degree
303
Figure D-10: 2002 Goodness-of-Fit Diagrams In-Degree
Figure D-11: 2003 Goodness-of-Fit Diagrams Out-Degree
304
Figure D-12: 2003 Goodness-of-Fit Diagrams Out-Degree
Figure D-13: 2004 Goodness-of-Fit Diagrams Out-Degree
305
Figure D-14: 2004 Goodness-of-Fit Diagrams In-Degree
Figure D-15: 2005 Goodness-of-Fit Diagrams Out-Degree
306
Figure D-16: 2005 Goodness-of-Fit Diagrams In-Degree
Figure D-17: 2006 Goodness-of-Fit Diagrams Out-Degree
307
Figure D-18: 2006 Goodness-of-Fit Diagrams In-Degree
Figure D-19: 2007 Goodness-of-Fit Diagrams Out-Degree
308
Figure D-20: 2007 Goodness-of-Fit Diagrams In-Degree
309
Table E-1: Correlation Matrix for Birth Rates
1 2 3 4 5 6 7 8 9
1. Failures (Deaths) 1
2. Births 0.46 1
3. Internet Users 0.82 0.08 1
4. Environmental Events 0.32 0.52 0.3 1
5. Org. Links 0.76 0.03 0.66 0.37 1
6. News Org. Forms 0.49 0.85 0.14 0.46 0.16 1
7. Establish. Orgs Forms 0.21 0.87 -0.2 0.12 -0.28 0.74 1
8. Strategy: Blogification 0.65 0.44 0.62 0.54 0.72 0.65 0.14 1
9. Strategy: Densification 0.7 -0.09 0.12 0.24 0.96 -0.02 -0.37 0.54 1
10. Strategy: Experimentation 0.77 0.07 0.20 0.4 0.80 0.15 -0.25 0.71 0.96
11. Strategy: Isolation 0.82 0.77 0.45 0.35 0.39 0.78 0.62 0.59 0.3
12. Change in Strategy 0.63 0.35 0.17 0.51 0.77 0.45 0.03 0.94 0.62
13. Num. of Newspapers 0.22 0.87 -0.21 0.12 -0.29 0.75 0.20 0.14 -0.37
14. News Media Coverage 0.88 0.14 0.47 0.21 0.9 0.21 -0.08 0.6 0.41
15. Media Coverage of Blogs 0.64 -0.19 0.15 0.18 0.93 -0.13 -0.43 0.48 0.47
16. Media Coverage of SNS 0.92 0.23 0.14 0.25 0.88 0.31 0.01 0.62 0.86
17. Positive Legal Rulings 0.32 0.2 0.53 0.52 0.49 0.15 -0.11 0.3 0.57
18. Professional Assoc. 0.42 0.75 0.01 0.46 -0.03 0.57 0.67 0.29 -0.15
APPENDIX E: CORRELATION MATRIX FOR BIRTH
AND FAILURE RATE VARIABLES
309
310
Table E-2: Correlation Matrix for Failure Rates
10 11 12 13 14 15 16 17 18
1. Failures (Deaths)
2. Births
3. Internet Users
4. Environmental Events
5. Org. Links
7. News Org. Forms
8. Establish. Orgs Forms
9. Strategy: Blogification
10. Strategy: Densification
11. Strategy: Experimentation 1
12. Strategy: Isolation 0.42 1
13. Change in Strategy 0.79 0.5 1
14. Num. of Newspapers -0.26 0.63 0.03 1
15. News Media Coverage 0.41 0.57 0.63 -0.09 1
16. Media Coverage of Blogs 0.44 0.2 0.59 -0.44 0.9 1
17. Media Coverage of SNS 0.88 0.67 0.63 0.01 0.60 0.84 1
18. Positive Legal Rulings 0.53 0.22 0.29 -0.12 0.45 0.47 0.4 1
19. Professional Assoc. 0.02 0.65 0.31 0.67 0.09 -0.17 0.17 -0.02 1
310
Abstract (if available)
Abstract
This study presents an examination of the organizational process of transformation, specifically examining how new organizations emerge as the result of new information communication technology, and how existing organizations emerge in response. In aggregate, the process of transformation is examined in three stages. First, this study looks at the nature of organizational forms, and seeks to understand how organizational forms emerge in rapidly changing competitive environments. Second, this study examines the process of legitimation in an attempt to better understand how emerging organizational forms are established as legitimate. Third, this research examines the process of organizational transformation, and seeks to introduce organizational strategy as a critical determinant of transformational success for existing organizations.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Organizational mimicry in American social movement communities: an analysis of form communication effects on the evolution of crisis pregnancy centers, 1989-2009
PDF
Ecology and network evolution in online innovation contest crowdsourcing
PDF
The coevolution of multimodal, multiplex, and multilevel organizational networks in development communities
PDF
Staying ahead of the digital tsunami: strategy, innovation and change in public media organizations
PDF
The informationalization of race: communication technologies and genomics in the information age
PDF
The Silicon Valley startup ecosystem in the 21st century: entrepreneurial resilience and networked innovation
PDF
The evolution of multilevel organizational networks in an online gaming community
PDF
Contagious: social norms about health in work group networks
PDF
Intermedia agenda setting in an era of fragmentation: applications of network science in the study of mass communication
PDF
Using public relations to gain legitimacy in an emerging market: nanotechnology firms and the news media
PDF
The evolution of knowledge creation online: what is driving the dynamics of Wikipedia networks
PDF
Digital creativity and innovation in Chinese social network sites industry
PDF
Communities of innovation: cyborganic and the birth of networked social media
PDF
Transformative learning in the public sphere: the educational dimensions of a broad-based community organizing movement
PDF
A multitheoretical multilevel explication of crowd-enabled organizations: exploration/exploitation, social capital, signaling, and homophily as determinants of associative mechanisms in donation-...
PDF
Conspicuous connections as signals of expertise in networks
PDF
Sustaining employee engagement and resilience through continuous transformation in digital organizations
PDF
The evolution of multidimensional and multilevel networks in online crowdsourcing
PDF
Digital storytelling as participatory media practice for empowerment: the case of the Chinese immigrants in the San Gabriel Valley
PDF
Effects of economic incentives on creative project-based networks: communication, collaboration and change in the American film industry, 1998-2010
Asset Metadata
Creator
Weber, Matthew Scott
(author)
Core Title
Media reinvented: the transformation of news in a networked society
School
Annenberg School for Communication
Degree
Doctor of Philosophy
Degree Program
Communication
Publication Date
08/04/2010
Defense Date
06/07/2010
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
community ecology,longitudinal network analysis,network structure,networks,news media,OAI-PMH Harvest,online news media,organizational ecology,organizational evolution,social network analysis,strategy
Place Name
London
(city or populated place),
United Kingdom
(countries),
USA
(countries)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Monge, Peter R. (
committee chair
), Castells, Manuel (
committee member
), Kennedy, Mark T. (
committee member
)
Creator Email
matthesw@usc.edu,mattsweber@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m3269
Unique identifier
UC1155726
Identifier
etd-Weber-3786 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-373334 (legacy record id),usctheses-m3269 (legacy record id)
Legacy Identifier
etd-Weber-3786.pdf
Dmrecord
373334
Document Type
Dissertation
Rights
Weber, Matthew Scott
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
community ecology
longitudinal network analysis
network structure
networks
news media
online news media
organizational ecology
organizational evolution
social network analysis
strategy