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
/
The fabled fourth estate: media freedom, democracy and human rights
(USC Thesis Other)
The fabled fourth estate: media freedom, democracy and human rights
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
THE FABLED FOURTH ESTATE:
MEDIA FREEDOM, DEMOCRACY AND HUMAN RIGHTS
by
Jenifer Whitten-Woodring
_____________________________________________________________________
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
(POLITICS AND INTERNATIONAL RELATIONS)
August 2010
Copyright 2010 Jenifer Whitten-Woodring
ii
Dedication
For my family:
To my brother, Guy, who convinced me to go back to school
To my parents, David and Jo, who have always supported and encouraged me
To my children, Patrick and Alex, who understood and did not complain when I
had to do my homework
And to my husband, Terry, who made it all possible
iii
Acknowledgements
Although writing a doctoral dissertation is a lonely process, it would never have
been possible for me to complete this work without the terrific support, assistance and
advice that I received along the way. Thus, while I am grateful to many for helping me
through graduate school and the dissertation process, I will do my best here to single out
those who are guilty of helping me the most.
The Faculty of University of Southern California’s Politics and International
Relations Program have endowed me with an excellent education. They have expanded
my understanding of the political world and provided me with the scientific knowledge
necessary to pursue some of the puzzles in both domestic and international politics. In
addition, they, along with the USC College, have provided generous financial support
that has made it possible for me to obtain specialized training and participate in
professional conferences.
In particular, my adviser, Patrick James, has provided unfailing support. Although
he is a dedicated workaholic with an ambitious research agenda all his own, he has
always found time to discuss my work and my challenges. I will look back fondly on our
lunches, which were light on carbohydrates and rich in social scientific theorizing and
counseling in academic professionalization. Pat is generous with his time and advice.
From the start, he encouraged me to develop and pursue my own research agenda.
Working with Pat, as a teaching assistant, research assistant and co-author has been an
enriching experience. From the time when I first ran into him (at a salad bar), he showed
a genuine interest in my ideas about media and politics. He has guided me throughout my
iv
graduate education and facilitated my professionalization by encouraging me to present at
conferences and introducing me to others in the discipline. Throughout this process he
has been a strong advocate. As he reviewed this dissertation, painstakingly, chapter by
chapter, he always knew when to demand more, when to compliment and when to throw
me a life-preserving suggestion.
Since I first arrived on campus, I have benefitted from Ann Crigler’s exacting
standards. She challenged and pushed me from the start when I struggled with my first
paper in her American Politics class. After meeting with her to discuss a research project,
I typically experienced an odd combination of devastation and exhilaration. Devastation
because I realized I had much more work to do; exhilaration because I realized that once
this work was completed, I would have accomplished something worthwhile. Her
perceptive critiques have improved this dissertation immeasurably. Additionally, she has
welcomed me into her network of political communication scholars.
Whenever I hit a crisis with my dissertation or graduate career, I could count on
Carol Wise to put everything in perspective and, when necessary, to advocate on my
behalf. Whether we met in person for lattes or teleconferenced via her Blackberry, Carol
made these problems seem much more manageable. Yet, Carol never held back when it
came to assessing my work. I remain grateful for this lack of restraint, and if there is any
lingering “busyness” in this dissertation, it is not her fault. I hope in the future to take
advantage of her expertise in Latin American politics.
Though he is officially listed as the “outside” member of my committee because
he does not teach in the Politics and International Relations program, Philip Seib was
v
very much on the inside when it came to influencing this dissertation and my career path.
Phil’s scholarship in international communication remains a source of inspiration and his
introduction to the community of international communication scholars has been
invaluable. He has lived up to his promise to serve as a “grand inquisitor” and I will live
up to mine to provide a final chapter to this project on the influence of new media as soon
as the data is available.
Speaking of data, I remain indebted to Doug Van Belle for sharing his Global
Press Freedom Dataset. Since I first met him over tea at the International Studies
Association meeting in San Diego, he has generously shared not only his data, but his
ideas about the conceptualization of media freedom. I look forward to continuing our on-
going debate about media freedom as we work on updating the dataset and our book on
media freedom around the world.
In addition to my training at the University of Southern California, I had the
privilege of attending the University of Essex Summer School in Social Science Data
Analysis and Collection, a true haven for nerds, for three summers running. There I
received intensive methods training from Harold Clarke, Jim Granato and B. Dan Wood.
I remain grateful to them for this training and have tried to put it to good use; however,
any mistakes herein are entirely my own.
In addition to the faculty at the University of Southern California, as a graduate
student I have enjoyed studying with a particularly supportive and engaging group of
graduate students at USC. In particular Amy Below, who adopted me as her mentee in
my first year at USC, has become my conference compatriot as well as my coauthor.
vi
Though we are no longer able to meet at the beach for serious discussions about content
analysis, I look forward to continuing our on-going collaboration. Since we sparred in
our religion and conflict seminar, Nukhet Sandal has also become a coauthor and good
friend. As a commuter from San Diego, I appreciated the hospitality from Nukhet, Kate
Svyatets, Denise Gonzalez-Banuelos, and Zeynep Sahin. Likewise, I will always value
the companionship of Lisa Ybarra and Xiangfeng Yang, which made the sociology
methods sequence almost fun.
It is not possible to pursue a graduate degree without the support and assistance of
many friends and family. Because I started a family before I considered pursuing a Ph.D.,
I relied on friends and family to take over many of my responsibilities so that I could
study, finish papers and crunch numbers. To that end, I am very thankful that we moved
into the house across the street from Christen Artino. She is an extraordinary babysitter
and a dear friend. I will always appreciate the love and care she bestowed on my children,
Patrick and Alex, and that she stayed on the job even after the birth of her son, Jack.
Likewise, although she is no longer with us, I will never forget Grace Maloney. Not only
did she cheerfully transport Patrick to baseball and Alex to cheerleading in her Mini
Cooper convertible, but she also provided a safe place to run to in an emergency.
Moreover, she was also always available when I needed to take a break from research and
writing. I miss our treks to Starbucks, Trader Joes and the fruit stand in Valley Center.
My family has always been a wonderful source of support. I never expected my
brother to become my mentor, but Guy Whitten sets a great example as a social scientist.
It was he who convinced me to pursue a Ph.D. and then counseled me as I negotiated my
vii
way through graduate school. I continue to appreciate Guy’s advice as well as his
patience, and that of my sister-in-law Christine Lipsmeyer and my niece Anna, during
our many phone conversations.
I finished this dissertation because my mother, Jo Wright Whitten, moved in
during the week before my deadline. She took care of my family so that I could move to a
hotel to complete the final chapters. This was nothing unusual. She and my father, David
Whitten, earned and used many frequent flyer miles, traveling to take care of my children
so I could attend conferences, complete papers and take my comprehensive exams. They
read drafts of many papers, listened to many stories and were a constant source of
encouragement. It is my greatest hope (and probably a vain one) that I will one day find
an employer who believes in me the way David Whitten does.
Patrick and Alex were justifiably nervous when their mother went back to school.
Yet, they adapted quickly and seemed to understand that sometimes I just had to do my
homework. Kids, I am finally done. I knew that Carleton “Terry” Woodring was an
excellent catch when he bragged about my accomplishments at my high school reunion
so that I did not have to do so. Yet, even that did not prepare me for the vast amount of
patience, love, and support that he has provided. He held our family together while I
pursued my degree. Terry, it is your turn now!
viii
Table of Contents
Dedication ii
Acknowledgements iii
List of Tables ix
List of Figures xiii
Abstract xvi
Chapter 1 Introduction: The Call for Media Freedom around the World 1
Chapter 2 Conceptualizing Media Freedom and Democracy 11
Chapter 3 A Conditional Theory of the Effect of Media Freedom on
Protest and Human Rights 35
Chapter 4: The Role of Media in the Repression-Protest Nexus 68
Chapter 5: Media Freedom, Democracy and Physical Integrity Rights 106
Chapter 6: Media Freedom, Democracy and Women’s Rights 160
Chapter 7: Journalists on the Frontline without Weapons 214
References 230
Appendix: Tables and Figures from Preliminary Analyses 240
ix
List of Tables
Table 1: Criteria for Media Freedom 15
Table 2: The Four Dimensions of Media Freedom 17
Table 3: Media and Regime Types: 1948-1995 32
Table 4: Correlation Analysis of CIRI Women’s Rights Measurements 58
Table 5: Summary of the Effect of Free Media, Autocracy/Democracy
and Respect for Human Rights on Domestic Conflict 80
Table 6: The Effect of Media Freedom and Autocracy/Democracy on the
Probability of a General Strike 83
Table 7: It’s all About Respect—the Effect of Media Freedom and
Autocracy/Democracy on Demonstrations 87
Table 8: Respect Matters with Riots Too—the Influences on the Predicted
Probability of Riots 88
Table 9: Collinearity Diagnostics 103
Table 10: Collinearity Diagnostics Among Components of Democracy 103
Table 11: Models with and without Influential Cases 104
Table 12: Logistic Regression vs. Rare Events Logistic Regression 105
Table 13: Watchdog or Lapdog? The Effect of Media Freedom and
Democracy on Physical Integrity Rights 116
Table 14: It Depends on Democracy: The Effect of the Interaction of Media
Freedom with the Level of Democracy on Physical
Integrity Rights 117
Table 15: It Still Depends on Democracy--The Effect of Media Freedom on
Physical Integrity Rights 5 Years Later 124
Table 16: What is it About Democracy? The Effect of Media Freedom and
Components of Democracy on Human Rights 126
x
Table 17: What is it About Democracy over Time? The Effect of Media
Freedom and Components of Democracy on Human Rights Five
Years Later 127
Table 18: The Effect of Media Freedom and Components of Democracy
on Disappearance 133
Table 19: The Effect of Media Freedom and Components of Democracy
on Disappearance Rights Five Years Later 134
Table 20: The Effect of Media Freedom and Components of Democracy on
the Right to be Free from Political Killing 138
Table 21: The Effect of Media Freedom and Components of Democracy on
the Right to be Free from Political Killing Five Years Later 139
Table 22: Summary of Findings of the Effect of Media Freedom and
Autocracy/Democracy on Physical Integrity Rights 151
Table 23: Correlation Between Independent Variables 154
Table 24: Variance Inflation Factors 155
Table 25: The Effect of Media Freedom and Autocracy/Democracy with
and without Controlling for Development 155
Table 26: The Four Largest (absolute value) DFBETA Scores for
Media Freedom*Autocracy/Democracy 157
Table 27: The Four Largest (absolute value) DFBETA Scores for
Media Freedom*Autocracy/Democracy 157
Table 28: Collinearity Diagnostics 159
Table 29: More Collinearity Diagnostics 159
Table 30: Watchdog or Lapdog? The Effect of Media Freedom and
Autocracy/Democracy (five-part measure) on Women’s Rights 171
Table 31: The Effect of Media Freedom and Democracy on Women’s
Rights Five Years Later 172
xi
Table 32: The Effect of Media Freedom and Components of Democracy on
Women’s Rights 177
Table 33: The Effect of Media Freedom and Components of Democracy on
Women’s Rights 5 Years Later 178
Table 34: The Effect of Media Freedom and Autocracy/Democracy on
Women’s Economic Rights 182
Table 35: The Effect of Media Freedom and Autocracy/Democracy on
Women’s Economic Rights Five Years Later 183
Table 36: The Effect of Media Freedom and Autocracy/Democracy on
Women’s Social Rights 186
Table 37: The Effect of Media Freedom and Autocracy/Democracy on
Women’s Social Rights Five Years Later 187
Table 38: The Effect of Media Freedom and Democracy on Women’s
Political Rights Five Years Later 192
Table 39: The Effect of Media Freedom and Democratic Components on
Women’s Political Rights 5 Years Later 193
Table 40: Summary of Findings of the Effect of Media Freedom and
Autocracy/Democracy on Women’s Rights 206
Table 41: Collinearity Diagnostics 211
Table 42: Collinearity Diagnostics with Democracy Components 212
Table 43: Watchdog or Lapdog? The Effect of Media Freedom and
Autocracy/Democracy (Polity2) on Women’s Rights 213
Table 44: Summary of Findings of the Effect of Media Freedom and
Autocracy/Democracy on Protest and Human Rights 216
Table 45: Summary of Characteristics of Four Non-Democracies with
Media Freedom 223
Table 46: Media Freedom and Levels of Democracy/Autocracy 1948-1995 240
Table 47: Media Freedom and Levels of Democracy/Autocracy 1981-1995 241
xii
Table 48: Binary Decomposition Results 242
Table 49: The Effect of Media Freedom and Components of Democracy
on the Right to be Free from Torture 243
Table 50: The Effect of Media Freedom and Democracy Components on the
Right to be Free from Torture Five Years Later 244
Table 51: The Effect of Media Freedom and Democracy Components on the
Right to be Free from Political Imprisonment 245
Table 52: The Effect of Media Freedom and Democracy Components on the
Right to be Free from Political Imprisonment Five Years Later 246
Table 53: The Effect of Media Freedom and Components of Democracy on
Women’s Economic Rights 247
Table 54: The Effect of Media Freedom and Democracy Components on
Women’s Economic Rights 5 Years Later 248
Table 55: The Effect of Media Freedom and Democracy Components on
Women’s Social Rights 249
Table 56: The Effect of Media Freedom and Democracy Components on
Women’s Social Rights Five Years Later 250
Table 57: The Effect of Media Freedom and Autocracy/Democracy on
Women’s Political Rights 251
Table 58: The Effect of Media Freedom Democratic Components on
Women’s Political Rights 252
xiii
List of Figures
Figure 1: Government Respect for Physical Integrity Rights in Mexico 37
Figure 2: Government Respect for Physical Integrity Rights in Uganda 38
Figure 3: The Role of the Media in the Repression-Protest Nexus 40
Figure 4: International Media in the Repression-Protest Nexus 42
Figure 5: Watchdog 49
Figure 6: Watchdog Barking Up the Wrong Tree 50
Figure 7: Lapdog: Government Controlled Media in a Democracy 51
Figure 8: Lapdog: Government Controlled Media in an Autocracy 52
Figure 9: The Effect of Media Freedom on the Predicted Probability
of a General Strike as Autocracy/Democracy Changes 82
Figure 10: The Effect of Government Respect for Physical Integrity Rights
on the Predicted Probability of a Demonstration 85
Figure 11: The Effect of Government Respect for Physical Integrity Rights
on the Predicted Probability of a Riot 86
Figure 12: Media Freedom, Autocracy/Democracy and Government Respect
for Human Rights in Uruguay 95
Figure 13: Domestic Conflict in Uruguay 99
Figure 14: Authority Patterns in Mexico 1980-2007 112
Figure 15: Authority Patterns in Uganda 1980-2007 113
Figure 16: Authority Patterns in the United States 1980-2007 113
Figure 17: The Marginal Effect of Media Freedom across Autocracy/Democracy 119
Figure 18: The Predicted Effect of the Interaction between Regime
Characteristics and Media Freedom on Government Respect
for Physical Integrity Rights 120
xiv
Figure 19: The Marginal Effect of Media Freedom on Human Rights
Five Years Later 123
Figure 20: The Marginal Effect of Media Freedom on Human Rights
(Depending on Regulation of Participation) 129
Figure 21: The Marginal Effect of Media Freedom on Human Rights Five
Years Later (Depending on Regulation of Participation) 130
Figure 22: The Effect of Media Freedom on Disappearance Rights (Low) 135
Figure 23: The Effect of Media Freedom on Disappearance Rights (High) 135
Figure 24: The Effect of Media Freedom on Probability of Frequent
Political Killing 140
Figure 25: The Effect of Media Freedom on the Predicted Probability
of Frequent Political Killing Five Years Later 141
Figure 26: Physical Integrity Rights, Media Freedom, Autocracy/Democracy
and Regulation of Political Participation in Uganda, 1981-1995 146
Figure 27: Physical Integrity Rights, Media Freedom, Autocracy/Democracy
and Regulation of Political Participation in Mexico, 1981-1995 147
Figure 28: The Marginal Effects of Media Freedom as the Level of
Autocracy/Democracy (without the lagged dependent variable) 158
Figure 29: The Marginal Effect of Media Freedom on Women’s Rights across
Autocracy/Democracy 173
Figure 30: The Marginal Effect of Media Freedom on Women’s Rights
5 Years Later as Regulation of Participation Changes 179
Figure 31: The Marginal Effect of Media Freedom on Women’s Rights
Five Years Later as Executive Constraints Change 180
Figure 32: The Marginal Effect of Media Freedom on Women’s Rights Five
Years Later as Executive Constraints Change 184
Figure 33: The Predicted Probability of Full Respect for Women’s Economic
Rights 184
xv
Figure 34: The Predicted Probability of No Respect for Women’s Social Rights 188
Figure 35: The Predicted Probability of Full Respect for Women’s Social Rights 188
Figure 36: The Predicted Probability of Full Respect for Women’s Political
Rights 190
Figure 37: Media Freedom, Autocracy/Democracy and Women’s Rights
in Burkina Faso 196
Figure 38: Women’s Economic, Social and Political Rights in Burkina Faso 205
Figure 39: Media Freedom and Levels of Democracy/Autocracy 1948-1995 253
Figure 40: The Effect of Media Freedom on the Predicted Probability of
Frequent Political Torture 254
Figure 41: The Effect of Media Freedom on the Predicted Probability of
No Political Torture Five Years Later 255
xvi
Abstract
For more than a century philosophers, politicians, human rights advocates and
ordinary people have called for media freedom. Indeed, spreading media freedom is
widely seen as a key to promoting development, democracy and human rights. The
idea is that independent media will keep government in line and make life better for
citizens. These are grand expectations. Many journalists have risked their lives and
some have died striving to fulfill them. Yet, really, little is known as to whether free
media are up to the task. That then is the focus of this study: under what conditions
are the media able to improve human rights? I present a conditional theory of the
effect of media freedom on protest and government respect for human rights.
Specifically, I hypothesize that the effect of media freedom varies depending on the
presence or absence of democratic characteristics because these determine the
vulnerability of government to public opinion. Much of what we know about the
effect of free media comes from studies of how they function in democracies.
Certainly, leaders who are constrained by political competition, political
participation and whose decision making power is limited are more likely to react
benevolently to media criticism. I argue that leaders who are not restrained by these
democratic characteristics are likely to respond quite differently. Similarly, citizens
who live in democracies have many institutional outlets with which to push for
government reform; those who live in autocracies do not. Therefore, I propose that
critical news coverage is likely to promote protest in non-democracies and that
leaders in these settings are likely to use repression to stifle dissent. I use statistical
xvii
analyses and case studies to test my hypotheses that the effect of media freedom on
protest and human rights depends on democracy. The results suggest that in absence
of democratic institutions, media freedom is associated with increased general strikes
and increased repression of physical integrity rights and women’s social, political
and economic rights.
1
Chapter 1
Introduction: The Call for Media Freedom around the World
"In order to enjoy the inestimable benefits that the liberty of the press ensures, it is
necessary to submit to the inevitable evils it creates."
Alexis de Tocqueville, 1835
For more than a century philosophers, politicians, human rights advocates and
ordinary people have called for media freedom. The idea is that independent media will
keep government in line and make life better for citizens. These are grand expectations.
Many journalists have risked their lives and some have died striving to fulfill them. Yet,
really, little is known as to whether free media are up to the task. That then is the focus of
this study: under what conditions are the media able to improve human rights?
In the first section of this chapter, I discuss how media freedom is an increasingly
popular cause, and why those working to promote democracy and development believe
there is much to gain in spreading media freedom. The second section provides a review
of the status of media freedom around the world. The human costs of media freedom are
the focus of the third section. The fourth section provides an overview of the dissertation.
Spreading Media Freedom around the World
Spreading media freedom is widely seen as a key to promoting development,
democracy and human rights. In 2006 alone, an estimated $242 million was spent on
media assistance programs (CIMA 2008). As part of its efforts to help developing
2
countries, the World Bank advocates for freedom of both print and broadcast media. The
assumption is that independent media can help to hold government accountable:
From Ghana to Indonesia, groups have gone “on the air” and named officials who
have stolen public funds or not delivered a long promised road—and they have
seen an active response. Such public condemnation makes officials change their
calculus of their prerogatives and their responsibilities. (Buckley, Duer, Mendel,
O'Siochru, Price and Raboy 2008: iv)
Clearly, there are advantages to having media perform this watchdog role; however, I
posit that some government leaders will not take kindly to any sort of public admonition,
especially those that force them to rethink how they go about their business.
In addition to minimizing corruption, human rights advocates have argued that
media freedom will lead to improved human rights, “By exposing human rights abuses
and giving voice to marginalised parts of the community, the media can at its best
encourage the proper application of justice and stimulate debates,” (Amnesty
International 2006). Indeed, we know from previous research that democracy is
associated with improved human rights (Poe and Tate 1994; Poe, Tate and Keith 1999;
Davenport and Armstrong 2004; Bueno de Mesquita, Downs, Smith and Cherif 2005).
Yet, it is quite challenging to democratize a country; it is by far easier to advocate for
media independence. In short, media freedom is seen as a quick fix.
Because media freedom is seen as a component of democracy, those who want to
spread democracy view the introduction of media freedom as paving the way for
democracy. It is far easier to urge autocratic leaders to relax their control over the media
than it is to ask them to “be more democratic,” by instituting limitations on their own
power and encouraging political competition. Likewise, in the face of international and/or
3
domestic pressure to reform, it is probably far more palatable to a dictator to instead
allow media freedom, at least on the books, just as he or she might agree to hold
elections, assuming all along that both the media and the elections process can be
covertly controlled. As we shall see in the coming chapters, this has happened in
Uruguay, Mexico, Uganda and Burkina Faso.
The problem for the authoritarian leader is that allowing a little media
independence is like opening Pandora’s Box. Once journalists get a taste of freedom, it
can be difficult to quiet them. Covert threats, withholding of government advertising and
libel suits may help keep journalists from reporting critically about government for the
most part, but as recent history shows, when the political winds shift, journalists will take
off their gloves. This happened in Mexico in the 1990s when alternative media began to
report on government scandals and then the mainstream media followed suit (Lawson
2002). In Uganda when Yoweri Moseveni seized control of the country, he benefited
from positive media coverage, but over time he also felt the pain of media criticism
(Ocitti 2005).
The Decline in Media Freedom
In spite of the efforts to spread media freedom, there are indications that media
independence is actually declining globally. According to Freedom House’s annual
survey, in 2008 only 70 countries had media that were considered free (Karlekar 2009).
“In terms of population, the survey found that only 17 percent of the world’s inhabitants
live in countries that enjoy a Free press,” (Karlekar 2009). In fact, the Freedom House
survey shows media freedom has declined every year since 2001, when 22 percent of the
4
world’s population lived in countries with free media (Freedom House 2002; Karlekar
2009). Although this negative trend spans all regions of the planet, between 2007 and
2008, there were substantial declines in North Africa, the Middle East and the former
Soviet Union (Karlekar 2009). Perhaps it is no accident that media freedom has declined
since 2001. According to Reporters Without Borders (2008), since the 9/11 terrorist
attacks, “It is not economic prosperity but peace that guarantees press freedom.” In the
wake of those attacks, even consolidated democracies like the United States have
imposed limits on media freedom, though the United States is still categorized as free by
Freedom House (2009) and “satisfactory” by Reporters Without Borders (2008).
It is curious that these apparent declines in media freedom have occurred at the
same time that the Internet has become increasingly available. Although the Internet was
expected by many to defy government control, this does not appear to be the case. Instead
“governments are now using traditional means of repression—including lawsuits,
imprisonment, and other forms of harassment—to clamp down on this novel means of
disseminating information,” (Karlekar 2009). Moreover, the emergence of new media has
contributed to a decline in the demand for traditional media, particularly in the United
States where the deteriorating economic situation has “threatened outlets’ financial
sustainability and the diversity of news sources,” (Karlekar 2009).
The landscape of the media has in fact changed dramatically in the last ten years,
largely because of the introduction of new media and the technological advances that
have facilitated global media. If ever there was a need to understand the impact of media
freedom it would be now. Yet, before we can understand the effects of these recent
5
changes, we need to understand the effects of media freedom prior to these changes. For
this reason (and because of data availability), this study focuses on the years before new
media became widely available, 1981-1995.
Paying the Price of Media Freedom
The conception of free media as being so powerful that they can force leaders to
shed long-held habits of corruption and repression places a heavy burden on journalists
who are tasked with fulfilling this watchdog role. Indeed, journalism is a perilous
profession. In 2007, 87 journalists were killed, 1,511 were “physically attacked or
threatened,” 887 were arrested and 67 were kidnapped (Reporters Without Borders
2008).
1
1
These were only the cases that were reported and confirmed. In all likelihood there were far more attacks
on journalists. Under-reporting is especially likely in more repressive regimes.
Even in countries that are highly democratic, journalists are killed for doing their
jobs. Take, for example, 1992. That year in India, reporter Ram Singh Biling was last
seen on January 3, in the custody of police. Though officials have denied it, he is
believed to have died in their custody (Committee to Protect Journalists 2009). Also in
1992, in the United States, newspaper editor Manuel de Dios Unanue was gunned down
in a restaurant in New York City. In this case the perpetrators were not public officials.
According to the Committee to Protect Journalists (2009) the de Dios killing was carried
out by “more than a dozen drug traffickers and businessmen” in retribution for articles he
penned about their illegal activities. In countries that are not consolidated democracies,
the risks are far greater. 1992 was a deadly year for journalists in Peru: Lima television
6
producer Alejandro Perez died when a bomb destroyed his station’s building; Huacho
radio news director Pedro Yauri was never seen again after he was taken by five men,
four of whom wore police uniforms; and Juanjui radio reporter Adolfo Isuiza Urquia was
tortured and killed after receiving threats from suspected drug traffickers (Committee to
Protect Journalists 2009). That same year, 8 journalists were killed in Tajikistan, 9 were
killed in Turkey, and 8 were killed in war-torn Bosnia-Herzegovina. Overall 42
journalists were killed “in direct reprisal” for their work in 1992, making it an average
year, as far as attacks on journalists go (Committee to Protect Journalists 2009). The
point is, reporting critically, whether about government, organized crime, or war, is
dangerous.
Introducing free media in non-democratic states puts journalists at risk. If doing
so truly makes life better, then perhaps it is worthwhile. I am not arguing against media
freedom. If independent media are able to hold government accountable and force
government to improve human rights, then spreading media freedom is indeed excellent
foreign policy, but we should find out first. The goal of this study is to test this
proposition. I hypothesize that rather than improving human rights in autocratic states,
the introduction of media freedom actually leads to more repression because it promotes
resistance. I test my hypotheses with a series of statistical analyses. I explore the
implications of these results in the context of the relationship between media freedom and
government in four non-democracies: Uruguay from 1980 to 1985, and Mexico, Uganda
and Burkina Faso from 1981 to 1995.
7
Overview
It is impossible to analyze the effect of media freedom without at first defining it.
In Chapter 2, I consider the legal, economic, and political impediments to media
independence and then propose a minimum definition of media freedom, similar to that
developed by Van Belle (1997), in which free media are viewed as media that exist
within a structural environment that allows journalists to criticize the government about
issues that could turn the people against the government. I argue that in addition to the
legal, economic and political environments, it is also important to consider the
professional environment, because it is the professional environment that shapes
journalistic agency and this is what largely determines whether journalists perform a
watchdog role. There are countries like the United States, which are comparatively free,
where the journalists often fail to act independently, and then there are countries like Sri
Lanka, which in recent years has certainly not been free, where journalists repeatedly risk
their lives to bring their audience critical information about the government. Journalistic
agency matters.
2
Regime type matters too. In Chapter 3, I present a conditional theory of the effect
of media freedom on protest and government respect for human rights. Specifically, I
hypothesize that the effect of media freedom varies depending on the presence or absence
of democratic characteristics because these determine to a large part the vulnerability of
government to public opinion. Much of what we know about the effect of free media
2
By journalistic agency I mean the decisions and actions that journalists make in contrast and sometimes in
conflict with the structural environment in which they work.
8
comes from studies of how they function in democracies. Certainly, leaders who are
constrained by political competition, political participation and whose decision making
power is limited are more likely to react benevolently to media criticism. I argue that
leaders who are not restrained by these democratic characteristics are likely to respond
quite differently. Similarly, citizens who live in democracies have many institutional
outlets with which to push for government reform; those who live in autocracies do not.
Therefore, I propose that critical news coverage is likely to promote protest in non-
democracies and that leaders in these settings are likely to use repression to stifle dissent.
In this chapter I also outline the research design and data used for this study.
Chapter 4 focuses on the role of the media in the repression-protest nexus. Here I
test the hypothesis that the effect of media freedom on protest varies, depending on
autocracy/democracy. Both the statistical analyses (which include 121 countries from
1981 to 1995) and the brief case study of Uruguay from 1980-1985, provide some
support for this hypothesis. In particular, the results of the statistical analyses indicate
that media freedom was associated with an increased predicted probability of a general
strike in autocracies, but not in democracies. The exploration of the events in Uruguay
during the first half of the 1980s puts these findings in context. During this time the
military junta first relaxed its control of the media, then as the opposition gained
momentum, the generals sought to reign in the media as a cycle of protest and repression
ensued. However, in the long term, Uruguay was able to break free from this cycle as it
democratized.
9
In Chapter 5, I present a study of the conditional effect of media freedom on
physical integrity rights, the rights to be free from political murder, torture, disappearance
and imprisonment. This study tests the hypothesis that the effect of media freedom on
physical integrity rights depends on autocracy/democracy using statistical analyses of
data from 93 countries from 1981 to 1995, and review events in Mexico and Uganda
during the same time period. Overall the findings provide support for the hypothesis of
the conditional effect of media freedom. In particular, in the statistical analyses media
freedom has a positive effect only in the most consolidated democracies, it has no
significant effect on countries in the middle, and it has a negative effect in the most
autocratic countries. Both the cases of Mexico and Uruguay suggest that at least in the
short term, media watchdogging can promote dissent which in turn often leads to
repression. The long term effects remain unclear. While Mexico became more
democratic, Uganda did not.
In Chapter 6, I consider whether the effect of media freedom on physical integrity
rights extends to other human rights, specifically women’s political, economic and social
rights. To test the hypothesis that the effect of media freedom on women’s rights depends
on autocracy/democracy, I conducted statistical analyses using data from 67 countries for
1981-1995 and a case study of Burkina Faso for the same time period. Here again, the
findings from the statistical analysis indicate that, for the most part, media freedom had a
negative effect on women’s rights in autocracies and sometimes had a positive effect, but
only in the most consolidated democracies. The case of Burkina Faso suggests that
improvements in women’s economic and social rights tend to lag behind improvements
10
in women’s political rights. Interestingly, increased media freedom and political
liberalization were associated with increases in women’s political rights and declines in
women’s social and economic rights. In particular, the initial failure and recent progress
in the campaign to eradicate female genital mutilation in Burkina Faso indicates that
persistent media exposure can help to change public opinion, but only if there is some
element in society that is amenable to that change.
In Chapter 7, I review the findings of this study and identify opportunities for
future research. Altogether, there is support here for the hypotheses that the effect of
media freedom on protest and government respect for human rights depends on
autocracy/democracy. In particular these results indicate that while media freedom is
associated with improved physical integrity rights in consolidated democracies, it has
quite the opposite effect on both physical integrity rights and women’s rights in
autocracies, and does not have an effect on countries in the middle. Thus, these findings
show that media freedom does not have a desirable effect on human rights in exactly the
kind of countries that are targeted by many media assistance programs. However, the
cases of Uruguay and Mexico suggest that media freedom may have a long term effect on
democratization in some countries. Therefore, it would be beneficial to study the long
term effect of media freedom on both democracy and corruption.
11
Chapter 2
Conceptualizing Media Freedom and Democracy
“When finally I am killed, it will be the government that kills me”
Lasantha Wickrematunge, 2009
On January 8, 2009, Sri Lankan News Editor Lasantha Wickrematunge was
gunned down as he drove to work. He was the fourteenth journalist to be killed in Sri
Lanka since 2006 (IRIN 2009). Wickremantunge’s killing prompted the World Bank to
issue a statement that these attacks on journalists were making it difficult for the news
media in Sri Lanka “to fulfill its core watchdog function and to carry out its role as a
medium of accountability,” (World Bank 2009). Although many have touted the value of
independent or free media, there is much debate as to what makes media free from
government control.
3
3
I am using the terms “free media” and “independent media” interchangeably here.
Certainly, in Sri Lanka, where journalists who criticize the
government put their lives in peril, the media are not free. Yet, journalists often perform a
watchdog role in countries like Sri Lanka. Wickrematunge persistently criticized the
government, drawing attention to human rights violations committed by both military and
rebel forces (O'Connor 2009), but he was well aware of the risks involved. In his final
12
column, published posthumously, he predicted his own death and blamed the government
(Wickrematunge 2009).
In this Chapter, I develop a conceptualization of media freedom. First I consider
competing definitions of media freedom and the many impediments to media
independence. I then propose a minimal definition of media freedom and reflect on how
this conceptualization of media freedom relates to competing characterizations of
democracy. Finally, I look at the interaction of media freedom with different regime
types.
What Makes Media Free?
Although organizations like the World Bank, the United Nations, Human Rights
Watch and Amnesty International have advocated for media freedom, arguing that it will
serve to make government more responsible and responsive to the needs of citizens, there
is much debate regarding the criteria that enable the news media to act independently.
Some have hypothesized that the relationship between news media and government is
one of “mutual exploitation,” where both seek to exploit the other to maximize their self-
interest and the public gets the byproduct of this process (O'Heffernan 1994: 244). Others
have argued that journalists reliance on sources combined with increasing economic
pressures and the inability of the news media to control the framing of news stories
prevent the news media from serving as a watchdog over government, even in established
democracies (Parenti 1986; Bennett, Lawrence and Livingston 2007). A case in point is
the failure of the U.S. news media to act independently following the 9/11 attacks and the
13
buildup to the Iraq war (Bennett et al. 2007). Yet, the U.S. news media clearly have the
potential to act independently, as evidenced by the reporting of Bush administration’s
dismal response to Hurricane Katrina (Bennett et al. 2007). This is not the case in many
states, such as Myanmar, China and Iran. In looking at the effects of media freedom, it is
important to consider how free the media are one country relative to other countries.
Certainly it is imperative to look beyond the issues of law and ownership and
consider the function and practices of the media. While laws promoting media freedom
are important, to be effective these laws must be enforced. Although government
ownership of media suggests that the government controls the media, this is not always
the case. In England, for example, the British Broadcasting Corporation is government
owned, but the news entities are editorially independent. Similarly, non-government
ownership does not guarantee media freedom. A case in point is Singapore where a
private company, Singapore Press Holdings (SPH), owns most of the country’s
newspapers; however, the government in effect controls SPH because it selects the
managers who in turn hire the journalists (Gomez 2005). Clearly harassment, threats and
attacks on journalists can have a chilling effect that impedes media freedom. Likewise,
economic manipulations, such as media reliance on government advertising or subsidies
for disseminating news, can curtail media independence. Both Reporters Without
Borders and Freedom House have created indexes of press freedom. Table 1 shows the
basic criteria considered in these indices.
While Freedom House focuses on the legal, political, and economic environment, or
the structure, within which the media function, it fails to consider the agency that drives
14
journalistic practices. In his study of Mexican media, Lawson (2002: 189) found that this
agency was crucial as journalists, “driven by a new vision of their place in society”
brought about the emergence of independent media.
4
Yet, in the nineteenth century, most newspapers in the United States were tied to
political parties (Hamilton 2004), and in many parts of the world, independent media
remain closely connected to competing political parties and are transparently partisan
(McQuail, Graber and Norris 2008). Waisbord (2000: 8) identified two models of
journalism: the partisan “journalism of opinion” model and the objective model, which
centers on “the principles of dispassion and political independence.” In both models,
covering politics and serving as a watchdog over government behavior are important
aspects of journalism, at least in theory; however, the degree to which journalists carry
out this watchdog role is largely determined by the intensity of their partisanship or their
professionalism because this shapes the agency with which journalists approach their
The Reporters Without Borders
index does take agency into account because it considers whether the news media engage
in investigative reporting and whether or not reporters criticize government policies, but
neither index considers the professionalism or partisanship of the news media. In
Western news media, professionalism is generally associated with the cultivation of the
professional values of objectivity, fairness, independence, and a sense of responsibility to
serve the news audience (Gurevitch and Blumler 1990). A crucial aspect of this
professionalism is that the media be free from bias.
4
Lawson (2002: 191) is referring to “human agency,” meaning the “decisions and actions of individuals.”
This conceptualization builds on Diamond’s (1992) critique of the tendency of social scientists to focus on
structure at the expense of individual action. This use of agency should not be confused with the principal-
agent which centers on the relationship between political authorities and bureaucrats (Moe 2006).
15
work. A strong sense of public service can drive journalists to overcome challenges from
the political, legal or economic environments. Therefore, I propose that there are four
dimensions of media freedom: the political, legal and economic structures and the
professional agency.
Table 1: Criteria for Media Freedom
FH= Freedom House; RSF=Reporters Without Borders
Media Freedom Controlled Media
Laws protecting media freedom (FH) Laws restricting media freedom—
especially requirement of license
(RSF)
Laws protected freedom of information (FH,
RSF)
Laws restricting freedom of
information
Free access to official and unofficial sources
(FH, RSF)
Lack of access to sources
Lack of laws restricting reporting (FH) Laws restricting reporting
Laws protecting journalists from harassment
and threats (FH, RSF)
Lack of laws protecting journalists
Enforcement of media freedom laws; lack of
enforcement of laws restricting freedom (FH,
RSF)
Lack of enforcement of media
freedom laws; enforcement of laws
restricting freedom
Freedom of media to determine content (FH,
RSF)
Censorship by government
Journalists and media outlets are free from
harassment and violence (FH, RSF)
Journalists and media outlets are
subject to harassment and violence
Freedom from economic manipulations like
taxation, dependency on government
advertising and/or subsidies (FH, RSF)
Subject to taxation, dependent on
government advertising and/or
subsidies
Diversity & transparency of ownership (FH,
RSF)
Concentration of ownership
Journalists resist bribery and gifts (RSF) Journalists accept bribery and gifts
Media engages in investigative reporting (RSF) Media fails to engage in
investigative reporting
Journalists do not self-censor (RSF) Journalists do self-censor
Media criticizes government policies (RSF) Media does not criticize
government policies
Protections extended to foreign journalists
(RSF)
Protections not extended to foreign
journalists
16
These four dimensions are interrelated in that a shift the political dimension could
bring about alterations in the legal or economic dimensions and vice-versa, but the
professional dimension can be juxtaposed to the other three. For example, in the political
dimension, a government that seeks to control the media may resort to legal or economic
means in order to do so. Likewise a change in the legal dimension, for instance new laws
providing media freedom could lead to modifications in the political and economic
dimensions such that the government might employ economic manipulations instead of
engaging in overt censorship. The political, economic and legal structures exist in concert
to establish a level of control over the news media. In contrast, the professional
dimension evolves in response to this level of control. In some cases, the response may
actually be a push back against this control, as demonstrated in the case of
Wickrematunge and his colleagues in Sri Lanka. In other scenarios, the response may be
complacency, where the media fail to perform a watchdog role in spite of legal, political
and economic environments that are remarkably free, as illustrated by the case of the
mainstream media in the United States during the buildup to the Iraq war. During this
time the majority of the media simply parroted government sources and failed to
investigate the validity of this information (Bennett et al. 2007).
5
5
The argument could be made that the complacency of the mainstream media mirrored that of the public in
that there was little opposition to the war; however, standard guidelines in journalistic practice call on
journalists to question all sources, even official sources and that was not done in this case.
Thus, taking into
consideration the role of these four dimensions in allowing news media to control the
agenda and framing of news, I propose that at its deepest level, a free media system
would include all of the criteria outlined in Table 2. Because, as this lengthy list of
17
criteria illustrates, there are so many potential influences on the media, it follows that any
media system will at best be imperfectly free.
Table 2: The Four Dimensions of Media Freedom
Legal Environment
• Provides and enforces
constitutional protection of
media freedom
• Is free from laws that restrict
reporting
Political Environment
• Is free from government censorship
• Allows open access to multiple
competing sources
• Is free from intimidation and
physical violence against
journalists
Economic Environment
• Is free from financial
manipulation by government or
other actors (including
restrictions on production and/or
distribution and reliance on
advertising and/or subsidies)
• Encourages a plurality of
ownership that facilitates
competition among media outlets
• Facilitates the dissemination of
information to citizens
Professional Environment
• Is characterized by autonomy,
distinct professional norms and a
public service orientation (Hallin
and Mancini 2004)
• Encourages journalists to serve as
watchdogs, monitoring and
reporting on government
• Encourages the coverage of
contentious stories
• Encourages news media to serve as
a voice for the marginalized
• Discourages self-censorship
Regarding the legal dimension, constitutional protection of media freedom is
important, but it can also be misleading. For example, Turkey has constitutional
provisions for media freedom, but it also has laws that criminalize reporting on some
politically sensitive issues including depicting as genocide the killing of one and a half
million Armenians in 1915 (VanBelle 2000; Freedom House 2007). Additionally, while
many countries have laws in place that call for freedom of the press, these laws are
meaningless if they are not enforced. Thus, the political environment plays a critical role
in determining media freedom.
18
In the political arena, it is important to consider the flow of information. Van
Belle (2002: 118) argued that free news media “becomes the primary arena for domestic
political competition” when they are available, but this arena is not an even playing field.
According to Entman’s (2004) cascading frames model, information streams down from
elites to the media and then to the mass public, with the possibility of some upstream
flow as public opinion influences media framing which in turn influences elite framing.
Framing is the selection “and highlighting [of] some facets of events or issues, and
making connections among them so as to promote a particular interpretation, evaluation,
and/or solution”(Entman 2004: 5). In addition to the importance of the framing of news,
another aspect of the political environment is the relationship between journalists and
government sources of information.
Even if the news media are free from overt government censorship, some would
argue that the news media are subject to government control because of their reliance on
official sources. This “indexing hypothesis” stipulates that the news media are merely a
megaphone for elites, especially government elites, and that any debate presented in the
news simply reflects the disagreement between these official sources (Hallin 1986;
Bennett 1990). Critics have countered that the hypothesis fails to explain how indexing
actually occurs in journalistic practice and it fails to take into account the “sense of
professional responsibility” that pressures journalists to seek out alternative views
(Althaus 2003: 404). Although professional environments and norms vary across
countries, the ethics codes of many news organizations indicate that journalists pride
themselves on giving a voice to the voiceless (Kovach and Rosenstiel 2001 ). Bennett et
19
al. (2007) argue that it is actually the professional norms governing mainstream
journalism that have often led to the failure of the American press to provide critical
coverage of government behavior. In particular, they posit that in striving to avoid bias
and accusations of agenda setting, journalists give more emphasis to sources they deem
more credible, namely those with greater political power (Bennett et al. 2007).
Certainly, the political environment can have a chilling effect on the media. In
times of war in particular, news media face strong pressure to refrain from news coverage
that is critical of the government. For example, CNN correspondent Christiane Amanpour
said the media was intimidated by the Bush administration and consequently “self-
muzzled” at the beginning of the Iraq war (Johnson 2003).
6
In addition to the political and legal environments, the economic environment can
greatly affect media freedom. First and foremost, media ownership is often perceived as
Similarly Bennett et al (2007)
found that before and during the Iraq war, members of the Bush administration displayed
extraordinary skill in managing the news media, especially Richard Cheney and Donald
Rumsfeld. “The degree to which their original promises and continuing accounts of the
war clashed with available evidence on the ground was matched perhaps only by the
venom they injected into attacks upon those (both newsmakers and journalists) who
would disagree with them in public,” (Bennett et al. 2007: 137). Thus it is difficult for the
news media to challenge the politically powerful, and yet, these are exactly who the news
media must confront if they are to serve as a fourth estate.
6
A 2005 survey found that editors and reporters did censor their reports on the Iraq war in particular to
remove graphic images (Hall and Bear 2005).
20
constraining reporting. Yet, as noted earlier, government ownership or funding of media
does not necessarily equate to government control of media. Instead, a multiplicity of
ownership is the ideal, but as market forces lead to corporate mergers, this diversity is
difficult to achieve and sustain. With ownership becoming increasingly concentrated and
media conglomerates becoming vertically integrated, many news critics have raised
concerns that the push to make news more profitable will result in poor news coverage
that leaves many voices unheard (Gans 1979; Parenti 1986; Bagdikian 1987). Others
have posited that regardless of actual ownership, the wealthy elite will exert control over
both the government and the media (Herman and Chomsky 1988). While few have
argued that media owners would act as censors, there are legitimate concerns that
journalists might engage in self censorship when selecting stories in order to please their
editors and publishers and avoid problems (Gans 1979; Parenti 1986). Indeed, according
to a survey of journalists, self censorship is prevalent in newsrooms (Pew Research
Center for the People and the Press 2000). Another concern regarding ownership is that
as news media come increasingly under the ownership of multinational conglomerates,
the focus will be more on profit-making and less on high quality journalism (Bagdikian
1987). Hamilton (2004) argued that market forces have driven news media coverage in
the United States since the Penny Press Revolution in the 1830s.
7
7
The Penny Press Revolution brought about the transformation of the newspaper industry in the United
States. Before the revolution newspapers were partisan and funded by subscribers. During the 1830s
newspapers began to drop subscription prices and became primarily funded by advertising. In order to
appeal to more advertisers by increasing their readership, newspapers became less partisan and more
objective.
21
While there is compelling evidence that the media are subject to economic
pressures, these same pressures can serve to keep media free from complete elite control,
provided there is sufficient plurality of ownership to allow for competition. There is a
tremendous demand on news organizations to make a profit. The only way they can
accomplish this is to sell their audience to advertisers; therefore mainstream large market
news organizations must cater to the masses and compete against each other to gain a
larger share of the audience (Hamilton 2004). If they fail to do so, they will not survive.
In addition, smaller news organizations that target niche markets will strive to provide
alternative information to audiences that are dissatisfied with what they find in the
mainstream media. Thus, although journalists in the mainstream media may be inclined
initially to self-censor to avoid some controversial or unpopular stories, journalists in
niche media often have incentive to tackle these stories. Ideally, once the niche media
pave the way, the mainstream media will follow, especially in the case of stories about
human rights violations. Lawson (2002) found that in Mexico, media openness led to
increased coverage, first by the niche media and then by more mainstream media, of
government corruption and state-sponsored repression:
The discovery that Mexican audiences devoured scandalous information
only encouraged greater reporting of incendiary and shocking events. For
media owners, scandals sold newspapers and boosted ratings; for
reporters, they helped make careers and satisfied personal desires to
participate in a new kind of journalistic enterprise that would expose the
failings of authoritarian rule. Market forces and journalistic norms thus
encouraged Mexico’s media to expose spectacular instances of
government abuse (Lawson 2002: 139).
Therefore based on economic issues, while elites certainly exert significant pressure on
the news media, they do not enjoy exclusive control over them.
22
One of the most overlooked aspects of media is the role of the professional
environment. Indeed, Lawson (2002: 191) attributed much of the opening of the Mexican
media to a strengthening of professional journalistic norms inspired by “disaffection with
the political system and exposure to foreign models.” Similarly, in looking at the
relationship between the news media and government in Uganda, Ocitti (2005: 101)
found that government efforts to stifle the press with legal restrictions failed because,
many [journalists] had long since decided to risk arrests rather than sacrifice their
journalistic freedom and ethics. It was almost as if the more the government
harassed them the more they attacked the government’s policies and exposed
indiscretions among the governing officials.
These cases illustrate the importance of professional norms in free media, particularly in
transitional governments. In both, the media persisted in critical reporting in spite of
government efforts to muzzle the negative coverage. In particular, professional norms
will influence whether or not journalists engage in self-censorship. This is one of the
most problematic aspects of media freedom. Even media that appear free according to
their legal, political and economic environments, can be quite vulnerable to self-
censorship unless there are strong professional norms which dictate otherwise.
Regardless of the legal, political and economic environment, it is the professional
environment that determines whether media will act freely.
8
8
Of course the presence of strong professional norms is no guarantee that the media will serve as a
watchdog or as a voice for the voiceless, especially in cases involving groups that are cultural outcasts like
the Roma. According to the European Roma Rights Centre (ERRC), not only do the media often fail to
address the plight of the Roma, but in some instances they actually promote anti-Romani sentiment. For
example, according to the ERRC, the Italian newspaper “Corriere della Sera” failed to provide editorial
comment when it printed anti-Romani statements made by Gianfranco Fini, the leader of National Alliance
23
In both Mexico and Uganda, in spite of unfriendly legal, political and economic
environments, the media did in fact act independently, largely because of the professional
environments. (Mexico’s professional environment was heavily influenced by the
Western emphasis on professionalism; whereas Uganda’s professional environment was
shaped by the media’s links to the opposition.) This begs the question as to why an
autocratic government would tolerate free media. I posit that independent media are
sometimes permitted in autocratic settings for the same reason that one-party regimes like
Mexico’s PRI hold elections: because holding elections and allowing some media
freedom lends legitimacy to the government. Moreover, dictators may find that free
media provide useful information. Egorov, Guriev and Sonin (2008) theorized that
autocratic leaders might permit media freedom in order to remain informed about the
performance of lower level bureaucrats in remote regions. Similarly, it follows that free
media would help to facilitate lateral communication across agencies of large
government. Moreover, an independent news media might provide information about
impending rebellions and other threats.
While there are reasons why an autocratic government might allow some press
independence, there are also indications that doing so opens the floodgates because once
somewhat free, the media can become increasingly difficult to contain. For example, in
the case of Mexico, government officials did at times resort to intimidation and
party, and thereby “contributed to an escalation of anti-Romani hostility,” (Italian Politician and Media
Brought Under Scrutiny for Anti-Romani Speech 2007).
24
harassment of critical journalists, but they also sought to maintain a façade of a free
press; thus, this relatively loose form of control facilitated the emergence of independent
media (Lawson 2002). In Uganda, Yoweri Museveni’s National Resistance Movement
relied on the news media to improve and maintain its image when it seized control of the
country in 1986 (Ocitti 2005). In order to use the media in this fashion, the government
“had to demonstrate a noticeable tolerance of the press,” (Ocitti 2005: 93). In each case,
the government sought to gain legitimacy by maintaining the illusion of a free press and
this unlocked the door for independent media. Moreover, as journalistic norms gained
strength, in the form of professional values in Mexico and in the form of intense
partisanship in Uganda, journalists pushed that door wide open.
While it is important to consider the multiple dimensions of media freedom, a
minimal definition of media freedom is more useful analytically. Although the above
listed criteria are necessary for the deepest level of media freedom, if we want media
freedom to mean more than “good media,” we need to identify the minimal requirement
for free media. In studying the relationship between media and government, the most
important role of the media is its ability to perform a watchdog role, monitoring and
criticizing government behavior because when the media performs this function it is able
to act as a forum for political debate regardless of other limitations on its freedom. Of
course even if the media are free to perform a watchdog roll, they may choose not to do
so. In his press freedom index, Van Belle (2000) focuses on the ability or inability of the
news media to criticize the government and thereby serve as an arena for political
competition. In order to facilitate political competition, the media must be capable of
25
criticizing the government about issues that are most important, namely those that could
turn the people against the government. An example of this would be the U.S. news
media’s coverage of the Bush Administration’s failure to respond adequately to the
Hurricane Katrina crisis, which at least temporarily, “produced a crack in the
[Washington] consensus,” (Bennett et al. 2007: 150). This more narrow definition of
media independence is aimed at the function and practices of the media. While the
political, legal, economic and professional environments each play a role in the capability
of the media to serve as a watchdog, I propose that it is the professional environment that
shapes the agency of the media and is therefore most important in determining whether or
not the media actually perform this function.
New Media, Old-Fashioned Control
In looking at the effects of media freedom, it is also important to consider the
impact of new media. In particular, the Internet, satellite television, and cellular phones
have brought about major changes in the media, many of which have the potential to
influence media freedom. Most significantly, it is extremely difficult for governments to
control the flow of information in these new media (Seib 2008). While it is plausible to
track down and imprison a newspaper reporter and stop the printing presses, it is far more
challenging to locate a blogger who is lashing out at the government anonymously from
her laptop computer. While it is possible to silence a subversive radio station, it is
difficult, if not impossible to pinpoint the origin of a chain of text messages mobilizing
26
anti-government protesters. Certainly new media are forcing governments to reevaluate
their policies on information control.
Yet for all of their potential diversity, new media face many of the same
challenges as traditional media when it comes to fulfilling the role of the fourth estate.
Moreover much of the news that is transmitted via new media originates from traditional
news media. While the Internet has created opportunities for citizen journalists to provide
alternative viewpoints, citizen journalists face many of the same challenges as
professional journalists when it comes to news gathering. Both the Internet and satellite
television have contributed to the globalization of the media. While a government may
exert control over the media that operates within its borders, it is far more difficult for
government to control media content that is produced outside of its borders. It is
however possible for government to limit access to the Internet and to some degree
satellite television. While some citizens may be able to circumvent government controls
imposed on the Internet or satellite, the costs of doing so are often prohibitive. It may be
very difficult for people to obtain the technology necessary, and the punishment for those
who are caught could be severe. For example in Myanmar, less than one percent of the
population has Internet access, and those that are able to go online are subjected to
extensive filtering that prevents them from accessing Web-based e-mail and sites with
content concerning “human rights advocacy and democratic reform,” (OpenNet Initiative
2007). Unlike Myanmar, Singapore enjoys one of the world’s highest rates of Internet
penetration (67.2 percent in 2005) (OpenNet Initiative 2007). Yet, like Myanmar, the
government seeks to control the information citizen’s access online. While it employs
27
only minimal Internet filtering, Singapore uses “a combination of licensing controls and
legal pressures to regulate Internet access and to limit the presence of objectionable
content and conduct online,” (OpenNet Initiative 2007). Thus, although the approaches to
Internet regulation are quite different in Singapore and Myanmar, these cases illustrate
that governments can and do engage in limiting freedom of expression in new media.
Myanmar relies economic measures (lack of access) and technology (filtering software),
whereas Singapore primarily employs old fashioned legal and political pressures.
The struggle for control over new media is similar to the offense/defense balance
in security studies. Just as new weapons for attack are balanced by defensive inventions
that naturally follow them (for example artillery and fortifications), so too are
government efforts to control new media followed by new technologies that circumvent
those government efforts. For example, government filters that block sites containing
certain content, may be avoided by using circumventor sites. Still, as new media become
more accessible, even in the most remote parts of the world, the ability of governments to
curtail media freedom will likely be minimized.
Defining Democracy
As with media freedom, democracy can be defined comprehensively or more
narrowly. In his review of different conceptualizations of democracy, Keech
(forthcoming) contrasts the minimal definitions of democracy put forth by Schumpeter
and Riker. Schumpeter characterized a democracy as an “institutional arrangement for
arriving at political decisions in which individuals acquire the power to decide by means
28
of a competitive struggle for the people’s vote,” whereas Riker viewed democracy as
providing the electorate with the opportunity to reject public officials (Qtd. in Keech
forthcoming: 24-25). Elections are central to both of these definitions, but the presence of
elections alone does not ensure democracy; therefore, democracy is generally
conceptualized as multidimensional (Dahl 1998; Marshall, Jaggers and Gurr 2003;
Freedom House 2007).
For example, Dahl (1998: 85-86) identified the following requirements for
democracy: elected officials, elections that are “free, fair, and frequent,” freedom of
expression, accessible alternative sources of information, the right to form and participate
in independent associations, and the inclusion of all adult citizens in the preceding
institutions. While Dahl emphasized the importance of a responsive government, other
conceptualizations have focused on checks and balances on government power (Keech
forthcoming). For instance, the Freedom House index of political rights looks at
anticorruption measures, political pluralism and participation and minority protections in
addition to the electoral process (Freedom House 2007). In contrast to the Freedom
House index and Dahl’s criteria, the Polity index (Marshall, Jaggers, and Gurr 2003)
focuses on authority patterns and considers the “closedness” as well as the “openness” of
political institutions, thereby incorporating measures for autocracy as well as measures
for democracy (Keech forthcoming; Marshall, Jaggers, and Gurr 2003). Additionally, the
Polity series democracy indicator is minimalist in that it focuses primarily on executive
constraints and political competition.
29
The challenge in conceptualizing democracy comes in determining which criteria
are necessary to establish a minimum level of democracy. A consolidated or deep
democracy would meet all of the above standards. Indeed, an argument could be made
that a consolidated democracy would include both free media and complete government
respect for human rights (Linz and Stepan 1996). The problem with this maximalist
definition is that it would be tautological to use it to explain outcomes like human rights
(Munck and Verkuilen 2002). Moreover, “To have ‘democracy’ mean, subjectively, ‘a
good government’ makes it analytically useless,” (Zakaria 2003: 19). Thus, if the goal of
an analysis is to sort out the characteristics that yield greater government respect for
human rights, it is important to use a minimal definition of democracy that focuses on
elections and the characteristics that make elections free and fair such as political
competition, participation and executive constraints.
Zakaria (1997: 43) argues that democracy, narrowly defined as free and fair
elections, without “constitutional liberalism” is “dangerous, bringing with it the erosion
of liberty, the abuse of power, ethnic division, and even war.” He defines constitutional
liberalism as protections of “the individual’s right to life and property, and freedom of
religion and speech” (Zakaria 1997: 26). I argue that freedom of speech, particularly
when manifested as freedom of the press, without democracy is also dangerous because it
provides information and a platform for expressing dissent without any other institutional
outlets for dissent, such as political competition, political participation and executive
30
constraints.
9
It follows that in the absence of these institutional outlets, a cycle of protest
and repression is likely to evolve.
The Interaction of Free Media and Autocracy/Democracy
Some scholars have argued that freedom of speech is more important to a
democracy than the right to vote, that “if citizens have the right to complain, to petition,
to organize, to protest, to demonstrate, to strike, to threaten to emigrate, to shout, to
publish,” government will be more responsible and more responsive (Mueller 1992: 984).
Yet, few have considered the effects of free media in states that lack democratic
characteristics. Because free media itself could be considered an essential component of
democracy, we would expect to find most free media in democratic states and most
controlled media in autocratic states, but this is not always the case.
Table 3 illustrates the distribution of democracies and autocracies with free media and
government-controlled media for all available countries from 1948 to 1995.
10
9
I am not making a normative argument against free speech, rather I am arguing that it is important to
consider how the effects of free speech vary across regime types.
Media
freedom is measured using the Van Belle’s Global Press Freedom Data (1997). This
dataset includes media freedom measurements for states for the years 1948 through 1995.
Media for each state/year is coded as free, partly free, partly controlled, or completely
controlled based on the its ability to serve as an “arena for political competition”
10
Tables A and B in the appendix give the breakdown of states with free media and controlled media by
each level of autocracy/democracy from 1948 to 1995 and from 1981 to 1995 (the years included in the
main analysis), respectively. To see the dispersion of government-controlled media and free media across
the full range of regime types see Figure 39, a kernel density plot, available in the appendix.
31
(VanBelle 2000).
11
Autocracy/democracy is measured using the Polity IV Data
combined Polity score which is coded for each state using a 21-point scale (-10 being the
most autocratic and 10 being the most democratic) (Marshall, Jaggers, and Gurr 2003); I
have used the standard threshold in which states scoring 6 and above are categorized as
democracies and those scoring 5 and below are categorized as autocracies. As expected,
the most common combinations of media and regime type are government-controlled
media in autocratic countries (60%) and free media in democratic countries (26%), but
there are exceptions. For example from 1981 to 1995, both Greece and Portugal were
coded as highly democratic (Marshall, Jaggers, and Gurr 2003), yet their media were
coded as being “restricted” or “somewhat controlled,” meaning that the media in these
countries were unable to function freely because they were not allowed to criticize the
government (Van Belle 2000). During the same time frame, both Uganda and Mexico
were coded as autocracies (Marshall, Jaggers, and Gurr 2003), yet during the same years,
both of these countries had media coded as “partly free,” meaning that the news media
were able to function as free presses because they could criticize the government (Van
Belle 2000).
11
More details about this dataset can be found in the Research Design, Methods and Operationalization
chapter.
32
Table 3: Media and Regime Types: 1948-1995 N=5575
Free Media
Government-Controlled
Media
Democracy
1463
(26%)
313
(6%)
Autocracy
423
(8%)
3376
(60%)
Thus, media freedom exists in a variety of settings. The primary objective of this
study is to ascertain how the effect of free media on government respect for human rights
varies across regime types. The “marketplace of ideas” concept holds promise for
explaining the variation in the effectiveness of free media across regime types. The idea
is that a marketplace of ideas, facilitated by freedom of expression and freedom of the
press, will provide a forum in which elites compete for political support and debate the
merits of different policies if, and only if, the right combination of institutions is in place
(Snyder and Ballentine 1996). Snyder and Ballentine (1996) suggested that just as
economic competition requires well-developed institutions to produce favorable results, a
competitive marketplace of ideas also depends on the presence of certain institutions if it
is to benefit society. Thus, I propose that the impact of free media on government respect
for human rights will depend on the presence of democratic characteristics, in particular
political competition, political participation and executive constraints. While free media
33
are able to report government violations of human rights, in the absence of political
participation and competition and executive constraints, government is less vulnerable to
public opinion and there are no institutional mechanisms with which to hold the
government accountable and push it to be more responsive and respectful of human
rights.
Conclusion
Media freedom can be compromised through legal, economic or political
channels. Because there are so many ways in which media freedom can be violated,
media are never perfectly free, but in order to be considered free at all, the structural
environment of the media must at least allow for journalists to criticize the government
about issues that really matter, meaning those issues that could potentially turn people
against the government. In evaluating media independence, though, it is important to
consider not only the legal, economic and political structure in which the media operates,
but also the role of journalistic agency. Even a system with the most permissive structure
will only have an independently functioning press if journalists are willing to perform a
watch dog role. Likewise, journalists who are motivated to pursue investigative reporting
may overcome many structural obstacles in order to do so. Together these different
aspects shape the news media and determine whether they will serve as a fourth estate.
Because media freedom exists across a range of regime types, it is important also to
consider how the effect of independent news media varies, depending on the presence or
34
absence of democratic characteristics. In the next chapter, I introduce a conditional theory
of the effect of free media on government respect for human rights.
35
Chapter 3
A Conditional Theory of the Effect of Media Freedom on Protest and
Human Rights
“The only security of all is in a free press. The force of public opinion cannot be resisted,
when permitted freely to be expressed. The agitation it produces must be submitted to.”
Thomas Jefferson to the Marquis de Lafayette, 1823 (1900, 2006)
Introduction
For more than 200 years we have assumed that media freedom will benefit
society. Evidence of this assumption can be found in United States foreign policy.
Indeed, the promotion of media freedom in fragile and developing countries remains one
of the primary goals of the United States Agency for International Development, the
justification being that an independent media “can ensure that citizens have access to a
variety of important sources of news and that information is not controlled exclusively by
the state or political-economic interests,” (Vanderwerff 2008: 1). The rationale is that if
people have access to more information, especially information about the government,
they will be able to hold government accountable. Media freedom is associated with
improved government respect for human rights in many countries, including New
Zealand and the Netherlands. Yet, media freedom has not led to improved government
respect for human rights in countries like Mexico and Uganda. Figures 1 and 2 show
depict the levels of government respect for physical integrity rights (the rights to be free
from politically motivated murder, torture, disappearance, and imprisonment) for Mexico
and Uganda respectively for the years 1981 to 1995. Both countries had media that was
36
functionally free during this period, yet both were consistently coded as having very low
government respect for these fundamental rights (Van Belle 2000; Cingranelli and
Richards 2006).
12
Why then are independent news media associated with improved government
respect in some settings, but not in others? In this chapter, I develop a conditional theory
of the effect of media freedom on protest and government respect for human rights. First,
I consider the overlooked role of the media in the repression-protest nexus and argue that
independent news media have the potential to serve as both a source of information and a
mechanism for mobilization. Second, I argue that the effect of media in the repression-
protest nexus varies depending on the presence or absence of democratic characteristics
which make government vulnerable to public opinion. I then introduce a theory of the
conditional effects of media freedom. In the fourth section, I derive a series of hypotheses
from this theory. Finally, I outline the data and methods that I use to test these
hypotheses.
12
The Cingranelli-Richards (CIRI) Human Rights dataset used to generate these figures are described in
greater detail later in this chapter.
37
Figure 1: Government Respect for Physical Integrity Rights in Mexico
0 1 2 3 4 5 6 7 8
CIRI Physical Integrity Rights Index
1980 1985 1990 1995
year
38
Figure 2: Government Respect for Physical Integrity Rights in Uganda
The Role of News Media in the Repression-Protest Nexus
Amnesty International and Human Rights Watch have advocated for media
freedom, arguing that independent news media will serve as a “voice for the voiceless”
by letting citizens know about government repression of human rights (Amnesty
International 2006). Thus, as depicted in Figure 3, the first potential role of the news
media in the repression-protest nexus is to provide information. If the news media follow
through and provide this information
13
13
There are of course many influences that may prevent the news media from providing information that
reflects negatively on the government. These are discussed in detail in Chapter 2.
, then it is up to the people to decide whether or
not to take action against the government. Here the media may play an intervening role
0 1 2 3 4 5 6 7 8
CIRI Physical Integrity Rights Index
1980 1985 1990 1995
year
39
by serving as a mechanism to facilitate and mobilize protest movements. The placement
of the arrows in Figure 3 reflects that in the first case the media acts as a watchdog after
the government has repressed and before the people react, but in the second case, the
media may facilitate mobilization while the people decide how to react because the
availability of the media to rally citizens serves to reduce the costs of organizing protest
movements.
Of course, this is a very simple picture, and it follows that there are many factors
that might come into play regarding the decision of various news organizations to cover
or not cover repression. Moreover, news organizations are diverse and make different
decisions. In fact it is quite likely that fringe media will take the lead in reporting
critically of the government and then the mainstream media may follow suit. Indeed, as
discussed in the previous chapter, this is what happened in Mexico in the early 1990s
(Lawson 2002). Likewise, there will be additional considerations that influence whether
citizens protest, but the focus of this study is on the role of the media in this process.
40
Figure 3: The Role of the Media in the Repression-Protest Nexus
In addition to domestic media, global and/or international media can also play a
role in the repression-protest nexus by bringing government abuses to the attention of the
international community, non-government organizations, international organizations and
governments of other states, and thereby generate international pressure for reform. Keck
and Sikkink (1998) posited that transnational activist networks will form around some
issues and bring about a “boomerang effect” through which domestic actors, who are
powerless to bring about domestic change, are able to connect with international actors
who in turn encourage their governments and international organizations to push the
offending government to reform. I argue that global and international media can play a
key role in this process, though not as direct a role as that hypothesized by the CNN
Government
Represses
People Push for
Reform (or Don't)
Government
Reforms or
Ignores
Media Acts
as Watchdog
Media
Facilitates
Mobilization
41
effect in which media is seen as driving foreign policy (Livingston 1997).
14
Instead, I
argue that global media can potentially play an intervening role by bringing a particular
issue in one country to the attention of people, governments and organizations in other
countries. This possibility, illustrated in Figure 4, begins with the international media
picking up reports of government repression from the domestic media. The international
media then bring the repression to the attention of transnational actors. The transnational
actors may mobilize and form a coalition around the issue which will then lobby
international organizations and/or governments of other states to pressure the repressive
government to reform.
14
As noted by Livingston (1997), there are three basic versions of the CNN effect. In the first, the live
coverage by news media is viewed to force policy makers to make quick decisions. In the second, the news
media are perceived as an impediment because graphic news coverage of military action weaken public
support. The third view posits that the news media actually drive foreign policy by setting the agenda. Here
I am using the CNN effect to describe this agenda-setting role.
42
Figure 4: International Media in the Repression-Protest Nexus
Government
Represses
People Push for
Reform (or Don’t)
Government
Reforms or Ignores
Media Act
as
Watchdog
Media
Facilitates
Mobilization
Formation of
Transnational
Coalitions that Call
for Reform
International Media
act as Watchdog
There is anecdotal evidence that domestic media sometimes cooperate with
international media to bring attention to government repression. Steele (2007) described
how in the early 1990s some journalists in East Timor put human rights activists in
contact with foreign reporters.
According to (editor) Metha Guterres, (the newspaper) Suara Timor Timur also
provided more direct forms of assistance to international media. “If we
interviewed Falintil in the jungle,” he said, “these stories couldn’t be included. So
we would send these stories to Reuters, or AP. In fact I became a regular source
for BBC London. We sent lots of stories – one, sometimes two or three a month.
If they couldn’t be included in STT, we would send them there” (Steele 2007:
267).
The case of East Timor shows that the international media, like the domestic news
media, can play two roles in the repression-protest nexus. First they can provide
information, thereby drawing attention to events in remote corners of the world, and
43
second they can facilitate mobilization of transnational activist networks by publicizing
actions and events staged by these actors. It is important to note here that the influence of
the domestic and international media in the repression-protest is still intervening at best. I
hypothesize that the effect of media coverage is tempered by the degree to which the
repressive regime is vulnerable to public opinion and international pressures. In the next
section I argue that government vulnerability to public opinion depends largely on the
presence (or absence) of democratic characteristics, in particular political competition,
political participation and limitations on executive power.
A Conditional Theory of the Effect of Media Freedom on Protest and Human Rights
The effect of the media, as depicted in Figures 3 and 4, is always intervening
rather than direct. Therefore any theory about the effect of the media, or media freedom,
is conditional, meaning that it depends on other factors. I propose that the effect of media
freedom on human rights depends on the presence or absence of democratic
characteristics. If the news media are independent, it follows that they would be likely to
report on government violations (or toleration of violations) of human rights. How the
media frames the stories about the violations will also influence how the mass public
reacts to the violations. I posit that in any given state, there is a level of cultural tolerance
for human rights violations which is based on public expectations regarding human
rights. Additionally, there is likely to be more cultural tolerance for abuse of some groups
44
than others.
15
My theoretical thinking is illustrated in Figure 5 which depicts the cycle of the
relationship between news media coverage and government respect for human rights in a
democracy. A driving force in this cycle is that there are executive constraints, political
competition and participation and free and fair elections, all of which serve to keep the
government responsive and vulnerable to public opinion. In this scenario, government
violations (or toleration of violations) of physical integrity rights that breech cultural
tolerance would result in significant news media coverage that would likely prompt
public outcry. In addition, the presence of political competition and popular elections
would provide incentives for political elites to co-opt marginalized groups in order to
increase the elites’ base of political support. This in turn would pressure the government
to reform or lead to the election of new leaders with a platform promising more respect
for human rights. It should be noted that the country will not necessarily be completely
free from government violations of human rights; over time the media may become less
vigilant and the government may once again violate physical integrity rights, but in
general, free media would discourage such behavior. Therefore, I propose that in a
democracy where news media are free from government censorship and able to act as a
If this cultural threshold is breeched, it increases the likelihood that the
news media will cover the violation and frame it in a manner that generates public
outrage. It is at this point that there is a difference in the effect of free media depending
on the presence or absence of democratic characteristics.
15
As previously noted, sometimes there is a culture of tolerance for the abuse of some marginalized groups
like the Roma.
45
watchdog over the government and where there are executive constraints, political
competition and participation, there is higher government respect for human rights
because of an interaction between the free media which are likely to cover violations of
human rights and the presence of vibrant institutions through which to hold the
government accountable.
An example of how this cycle works can be found in the Rodney King case. In
1991 in Los Angeles, the videotaped beating of Rodney King garnered much attention
from the media. The coverage of this case drew attention to the problem of police
brutality and racism (Jacobs 1996). In 1992 news of the subsequent acquittal of the police
officers involved in the incident spawned riots in which 54 people were killed, more than
2300 injured and more than 13,200 were arrested (Human Rights Watch 1998). The
media, at both the local and national level continued to cover this story. Although the
officers were acquitted in state court, the case did bring about change. In local
government, Los Angeles Police Chief Daryl Gates stepped down and Mayor Tom
Bradley, who had been in office since 1973, decided not to run again in 1993 (Jacobs
1996). According to Human Rights Watch, the report of the Christopher Commission,
which was formed by Mayor Bradley to investigate the incident, created a “blueprint” for
reforming the Los Angeles police department which Human Rights Watch predicted
would help “to create and maintain a culture of accountability” in departments across the
country (Human Rights Watch 1998). At the federal level, the officers involved in the
King beating were tried on civil rights violations; two were convicted and two were
acquitted. While Human Rights Watch (1998) found that abuse by police “remains one of
46
the most serious and divisive human rights violations in the United States,” it also found
the “pattern or practice” statute of the Violent Crime Control and Law Enforcement Act
of 1994 to be a positive development because it called for the Justice Department to take
action against police departments that fail to address police misconduct. This case shows
how news coverage of government abuse combined with democratic characteristics that
make government vulnerable to public opinion like elections, political competition and
executive constraints serve to make government more responsive and respectful of
human rights.
In contrast, in an autocratic setting, even with free media, the absence of
democratic pressures reverses the cycle. In fact, the very existence of free media indicates
that the autocracy, whether by design or by accident, does not exercise complete control
over its citizens, but the ability of the free media to influence the government is also
limited because of the absence of political competition and political participation make
the government less vulnerable to public opinion. I suggest that in the absence of
democratic characteristics, independent media will serve to lower government respect for
human rights because the interaction of free media, which are likely to cover violations,
and a lack of government vulnerability to public opinion, leave citizens with few means,
other than protest, with which to hold the government accountable and push it to be
responsive. As Figure 6 illustrates, provided they exceed cultural tolerance, violations of
human rights will likely prompt considerable news coverage and place human rights at
the forefront of the public agenda. This would inspire both debate and concern in citizens,
but in an autocracy there are no institutional outlets to push for government reform and
47
little or no incentive for political elites to co-opt the marginalized citizenry. In this case,
there are two possible outcomes. The media framing will influence public perceptions
about the costs of protesting (the expected reaction from the government) and the costs of
not protesting (the expected benefits from protesting and the likelihood that the protest
will succeed). If there are no institutional outlets to push for government reform, and the
repression is so bad that people feel they have nothing to lose, then the costs of not
protesting may well exceed the costs of protesting. This is the protest threshold. If this
protest threshold is not breeched, however, the result could be a culture of fear in which
people are aware of the violations, afraid of their government, and consequently more
submissive. In either case, I predict that the outcome would be further violations of
human rights. A case in point is Iran in the late 1990’s. In 1997, President Khatami
pushed for greater press freedom. Consequently, the press became “the major mobilizing
tool for reformists seeking greater respect for rights,” (Hicks 1999). Although
newspapers reported on the widespread cases of political murder, torture and
imprisonment, protests and calls for reform were met with further repression (Human
Rights Watch 1999). This case illustrates that while independent media in an autocracy
can lead to antigovernment protests, an autocratic government, being less vulnerable to
public opinion, is likely to suppress dissent with repression.
When the media are government-controlled, it follows that they will have little or
no impact on government respect for human rights because censorship constrains the
news media to a lapdog role. In this scenario (illustrated in Figure 7), government
violation of physical integrity rights would either not be covered at all or would be
48
framed as justifiable punishment, perhaps necessary for security purposes. In the
mismatch of a democracy and government-controlled media, democratic pressures would
still constrain government behavior. While the presence of democratic institutions would
provide some incentive for competing political elites to co-opt marginalized groups,
government control of information would likely stifle the flow of communication. Thus,
framing or a lack of coverage might promote distrust of both the media and the
government, but probably not to such a degree as to encourage or discourage further
violation of human rights. Therefore, I propose that in democracies with controlled
media, the interaction between the government and the media will have little or no effect
on government respect for human rights.
Figure 5: Watchdog
Violations exceed
cultural tolerance
Time passes
without
incident, so
media becomes
less vigilant
Government violates
(or tolerates
violations of) human
rights
News coverage prompts
public concern & debate
about human rights
Public and political
elites call for
government reform
and/or change in
leadership
Government
reforms and/or
new leaders are
elected
Media’s watchdog role is
reinforced, so for a time
media is vigilant
Political elites see
opportunity to increase
base of support by co-
opting the cause of the
marginalized
Media report on
violations
Cultural Threshold
The Impact of Free Media &
Democracy on Human Rights
49
Figure 6: Watchdog Barking Up the Wrong Tree
There is no political
incentive for elites to
co-opt the
marginalized; there is
no institutionalized
channel for citizens to
voice outrage
Government violates
(or tolerates
violations of) human
rights
Free media reports
violations
Government seeks to
repress protest and
discourage demands
OR realizes benefit
from repression
because of culture of
fear keeps citizens
submissive
Protest/demands
that government
respect human
rights OR culture
of fear results
News coverage
prompts public
outrage and debate
about human rights
Violations exceed
cultural tolerance
Protest Threshold
The Impact of Free Media &
Autocracy on Human Rights
Cultural Threshold
50
51
Figure 7: Lapdog--Government Controlled Media in a Democracy
Media fails to report
violation or frames it
as justifiable
punishment
Citizens don’t learn of
violation OR do and
don’t trust media or
government
Nothing
happens OR
government
reforms or is
voted out
Government
realizes no benefit
or cost from
violation OR
government sees
cost and reforms
Time passes. Government
behavior continues as before
OR government becomes less
careful
Government
violates human
rights of
marginalized group
52
Figure 8: Lapdog--Government Controlled Media in an Autocracy
Media fails to report
violation or frames it
as justifiable
punishment
Citizens are afraid to
speak out against
government and
there is no incentive
for elites to help
those in the
marginalized group
Fear prompts
citizens to remain or
become more
compliant
Government
recognizes benefit
from violating
human rights
Government violates
human rights of
marginalized group
53
My fourth proposal is that autocracies with government-controlled media will
have less respect for human rights than democracies, but more respect for these rights
than autocracies with free media. As Figure 8 illustrates, with this combination of
government and media, government violations of physical integrity rights will likely
result in no news coverage or news coverage that is framed by the government in a
manner that promotes fear and keeps the public submissive. The government, having
experienced a benefit from controlling the media and from violating human rights, would
likely continue to engage in both behaviors. Taking these four propositions of how free
media and government controlled media function in autocracies and democracies, I
derive a series of testable hypotheses.
Hypotheses
Based on the propositions outlined above, this study will test the following hypotheses:
H
1
Media freedom is positively related to government respect for human rights.
H
2
Media freedom is positively related to protest movements.
H
3
The effect of media freedom on protest movements and government respect for
human rights depends on the presence of democratic characteristics of political
participation, political competition and executive constraints such that:
H
3a
If the regime is democratic and media are free, there will be little protest and
the government will have increased respect for physical integrity rights.
H
3b
If the regime is democratic and the media are government-controlled, there
will be little protest, and the government will have less respect for human rights
54
than democratic regimes with free media, but more respect than autocratic
regimes in general.
H
3c
If the regime is autocratic and media are government-controlled, there will be
some protest and the government will have less respect for human rights than
democratic regimes, but more respect than autocratic regimes with free media.
H
3d
If the regime is autocratic and the media are free, there will be more protest
than in the other scenarios, and the government will have less respect for human
rights than autocratic governments with government-controlled media and
democratic regimes in general.
Research Design, Methods and Operationalization:
In order to test my hypotheses, I need a systematic analysis which can control for
multiple causal factors; therefore, I conduct three series of multiple regression analyses
using the dependent variables of domestic conflict (anti-government demonstrations, riots
and general strikes), government respect for physical integrity rights (politically
motivated murder, torture and imprisonment), and government respect for women’s
social, economic and political rights. In each analysis, I look at the effects of the
independent variables of media freedom (free or government-controlled), the level of
autocracy-democracy, the interaction between media freedom and various control
variables (as outlined below). The multiple regression analyses include data for all
available countries (93 to 122, depending on the study) from 1981 to 1995. This time
span was selected because of data availability, but this time period also provides a wide
55
range of levels of autocracy/democracy with and without free media. Moreover, it covers
the time period immediately prior to the influx of Internet media, thereby making it
possible to get a better understanding of the effect of media freedom separate and apart
from any increased freedoms brought about by the availability of the Internet.
Dependent Variables: Protest Movements, Physical Integrity Rights & Women’s
Rights
To study the effect of media freedom on protest movements (this study is featured in
chapter 4), I use the domestic conflict events data found in the Cross-National Time-
Series Data Archive (Banks 2008). Specifically, I use the following events count
variables:
General Strikes, which includes, “Any strike of 1000 or more industrial or service
workers that involves more than one employer and that is aimed at national
government policies or authority.”
Riots, which includes, “Any violent demonstrations or clash of more than 100
citizens involving the use of physical force.”
Anti-government Demonstrations, which includes “Any peaceful public gathering
of at least 100 people for the primary purpose of displaying or voicing their
opposition to government policies or authority, excluding demonstrations of a
distinctly anti-foreign nature.” (Banks 2008)
This dataset was gathered largely from newspaper reports, mostly from the New York
Times, and is consequently biased geographically because newspapers like the New York
Times allocate much more space to the coverage of domestic events than they do
international events and it is often difficult, if not impossible, for newspapers in one
country to get information about all domestic conflicts in foreign countries, especially
56
those in remote regions of the world (Nam 2006). Consequently, events in countries like
Burma/Myanmar are less likely to be reported than those in the United States, and other
liberal democracies. However this dataset is the only one to provide domestic conflict
data for all available countries for the years covered in this study. Moreover, the tendency
of this dataset to under-report instances of domestic conflict is more likely to minimize
the effects of the independent variables rather than to exaggerate them; therefore, it is the
best available dataset with which to test my hypotheses. To minimize the problems of the
under and over reporting of protests, I simplify these measures and use dichotomous
variables (strike/no strike, demonstration/no demonstration, and riot/no riot) for each
country/year.
To test the hypotheses about the effect of media freedom on a variety of human
rights, I use variables from the Cingranelli-Richards (CIRI) Human Rights Dataset
(Cingranelli and Richards 2006). First, in Chapter 5, I look at government respect for
physical integrity rights, meaning the right to be free from political imprisonment,
murder, disappearance, and torture. Although freedom of expression is considered by
many to be a basic human right, it is important to clarify here that physical integrity
rights are distinct from the right to free speech. This study employs the Physical Integrity
Index, which is constructed from indicators of Torture, Extrajudicial Killing, Political
Imprisonment, and Disappearance (Cingranelli and Richards 1999). It ranges from 0 (no
government respect for these four rights) to 8 (full government respect for these four
rights). This additive index was constructed from indicators on torture, extrajudicial
killing, political imprisonment and disappearance, using Mokken Scaling Analysis
57
(MSA)
16
In Chapter 6, I focus on the effect of media freedom on women’s rights. The CIRI
dataset includes three indicators measuring women’s rights: women’s political rights
(including rights to vote, to run for and hold a political office, and to petition government
officials), women’s economic rights (including rights to equal pay for equal work, free
choice of employment, to work in military and police force, and equality in hiring and
promotion), and women’s social rights (including rights to equal inheritance, to enter
into marriage on basis of equality with men, to initiate a divorce, to travel, to education,
to own property, to be free from genital mutilation and free from forced sterilization).
Each of these variables is measured on a scale of 0 to 3, with 0 being no government
respect for these rights and 3 being complete government respect for these rights. In
reviewing the descriptions of these variables in the CIRI codebook, I suspected that they
would be highly correlated. Running a correlation matrix confirmed this (see Table 4).
The Chronbach’s alpha of .78 is higher than the generally accepted standard of .65, so I
am satisfied that these three variables are in fact measuring the same dimension, women’s
rights. Therefore, I decided to combine them in an additive scale measuring 0 to 9, with 0
meaning no government respect for women’s rights and 9 meaning complete government
, to provide information regarding the “pattern and sequence” of government
respect for specific physical integrity rights as well as an overall level of government
respect for these rights (Cingranelli and Richards 1999). In addition, I also run analyses
using the various indicators to identify any independent effects.
16
Specifically Cingranelli and Richards (Cingranelli and Richards 1999) employed a polychotomous
MSA, which is a cumulative scaling technique to create a unidimensional measurement.
58
respect for women’s rights. In Chapter 6, I include analyses using each indicator as well
as the additive index.
Table 4: Correlation Analysis of CIRI Women’s Rights Measurements
Women’s
Economic Rights
Women’s Political
Rights
Women’s Social
Rights
Women’s
Economic Rights
1
Women’s Political
Rights
.41 1
Women’s Social
Rights
.78 .45 1
Chronbach’s Alpha .78
Independent Variables: Media Freedom & Autocracy-Democracy
As discussed in Chapter 2, when analyzing the effect of media freedom, it is
important to use a minimal definition of media freedom that is consistent over time.
Therefore I have opted to use the Global Press Freedom Dataset, which includes media
freedom measurements for states for the years 1948 through 1995 (Van Belle 2000). Van
Belle used the following five-point coding scheme:
0 – “Press non-existent or too limited to code”
1 – “Press is clearly free and the news media are capable of
functioning as an arena of political competition”
2 – “Press freedom is compromised by corruption or unofficial
influence, but the news media are still capable of functioning as an
arena of political competition”
3 – “Press is not directly controlled by the government, but it is not
capable of functioning as an arena of political competition or
debate”
4 – “Press is directly controlled by the government or strictly
censored” (Van Belle 1997)
59
This is a categorical coding rather than interval scale; the difference between media
coded 2 and media coded 3 is far more substantial than the differences between those
coded 3 and 4 and those coded 1 and 2 (Van Belle 2000). Therefore, I have recoded
categories 1 and 2 as 1 for “free media” and categories 3 and 4 as 0 for “government-
controlled media.” The category 0 is treated as missing data.
In measuring autocracy/democracy, it is critical to restrict the measurement to
institutional democracy because other measurements of democracy that incorporate civil
liberties might also include measurements of media openness and human rights violations
(further discussion of this is available in Chapter 2). In order to avoid a tautology, to test
my hypotheses and account for the variation of media openness across regime types, it is
important to keep these measurements discrete. As it turns out, the Polity IV dataset does
not incorporate media freedom (Choi and James 2006; Marshall and Jaggers 2002). For
the measure of the level of democracy/autocracy, I use the Polity IV dataset’s combined
Polity score (which ranges from -10 to 10, with 10 being the most democratic and -10
being the most autocratic), which is obtained by subtracting the country’s autocracy score
(which ranges from 0=low to 10=high) from the country’s democracy score (which
ranges from 0=low to 10=high) (Marshall and Jaggers 2002). Specifically, I use Polity2,
a version of the Polity variable that has been adapted for time series analysis (Marshall
and Jaggers 2002). Following Davenport and Armstrong (2004), I started with this scale
and then conducted an exploratory analysis to identify threshold effects. Davenport and
Armstrong (2004) found that the relationship between democracy and government
respect for human rights is not linear and conducted a binary decomposition of the Polity
60
democracy measure (in which a dichotomous variable is created for each value of the
scale) to identify a threshold of domestic democratic peace above which democracy does
have a positive and significant influence, but below which it does not. Because I am
looking at the interaction of free media with autocracies as well as democracies, I also
used binary decomposition to come up with an improved measure of the effects of the
level of autocracy/democracy and its interaction with media freedom.
17
Regulation of Chief Executive Recruitment (XXREG) which is a three-
category measure ranging from Unregulated (1), to Designated (2) to
Regulated (3).
Using this
information and the threshold effects identified by Davenport and Armstrong (2004), I
have recoded the Polity2 variable to create a five-part measure for autocracy/democracy
(-10 to -4=0; -3 to 0=1; 1-7=2; 8-9=4; 10=5). To thoroughly test my hypotheses I will
include models with both the Polity2 variable and the five-part variable. Additionally, I
include analyses that look at the independent effects of the Polity Concept and
Component Variables which offer a different breakdown of authority characteristics of
states. These are:
Competitiveness of Executive Recruitment (XRCOMP), which is a three-
category measure ranging from Selection (1), to Dual/Transitional (2)
where “one is chosen by hereditary, succession and the other by
competitive election” and Election (3)
Openness of Executive Recruitment (XROPEN), which is a four category
measure ranging from Closed (1) to Open (4)
The Concept Variable of Executive Recruitment (EXREG) which is a
eight-category scale ranging from Ascription (1), meaning “succession by
birthright” to Competitive Election (8)
17
The results of the binary decomposition can be seen in Table C of the Appendix.
61
Executive Constraints (XCONST or EXCONST) , which is a seven-
category scale ranging from Unlimited Authority (1) where there are no
limits to the executive’s power, to Executive Parity or Subordination (7),
where “accountability groups have effective authority equal to or greater
than the executive in most areas of activity.”
18
Regulation of Participation (PARREG), which is a five category scale
ranging from Unregulated (1) to Regulated (5).
The Competitiveness of Participation (PARCOMP), which is a five
category scale PARREG, to Repressed (1) to Competitive (5)
The Concept Variable of Political Competition (POLCOMP), which is a
ten-category scale ranging from Suppressed (1), where “while no
significant political activity is permitted outside the ranks of the
hegemonic regime, nevertheless, some organized political participation
occurs within the regime through highly circumscribed institutional
channels,” to Institutionalized Electoral (10), where “relatively stable and
enduring political groups compete for political influence with little use of
coercion.” (Marshall and Jaggers 2009: 23-24).
Because there is some discrepancy in the Polity IV codebook as to whether or not
censorship of opposition media is considered in the coding of the Competitiveness of
Competition, these variables will only be used in preliminary analyses. Of the component
variables, executive constraints is the one that largely determines the Polity Score for
each country/year (Gleditsch and Ward 1997).
My third hypothesis, that the effect of media freedom depends on the presence of
democratic characteristics is interactive and leads directly to a multiplicative interactive
specification. Therefore, the interaction between media freedom and the democratic is
measured by multiplying the media freedom score for each country/year by the composite
Polity2 variable and each of the Polity Component and Concept variables.
18
The executive constraints concept and component variables are the same.
62
Control Variables
Of course there are variables other than media freedom and the
democratic/autocratic characteristics that may increase or decrease protest movements
and government respect for human rights. Researchers looking into causes of government
violation of human rights have identified several independent variables including the
country’s population, its level of economic development, and its involvement in
international or civil war (Poe and Tate 1994; McCormick and Mitchell 1997; Poe et al.
1999).
Poe and Tate (1994) proposed that such violations would be more likely when a
government is facing a threat and has a need for greater control over its citizens. It
follows then that involvement in war, both international and civil, has been linked with
increased government violations of the human right to physical integrity (Poe, Tate, and
Keith 1999; Poe and Tate 1994). The threshold for war is generally accepted to be 1000
battle-related deaths (Small and Singer 1982), but I suspected that the threats that might
lead to government violation of human rights probably exist at a much lower threshold.
Preliminary analyses supported this hypothesis, so I look at the impact of armed conflict,
which has a threshold of 25 battle-related deaths per year. A country’s involvement in
international conflict and/or internal conflict is measured using the UCDP/PRIO Armed
Conflict dataset (Gleditsch, Eriksson, Sollenberg and Strand 2002; Strand, Carlsen,
Gleditsch, Hegre, Ormhaug, Wilhelmsen, Wallensteen, Sollenberg, Eriksson, Harbom,
Buhaug and Rod 2005). This dataset includes variables measuring the lower threshold of
armed conflict, 25 battle-related deaths, in addition to variables measuring the standard
63
criteria of 1000 or more battle deaths used by Small and Singer (1982) to define war.
These include interstate armed conflict (type two conflict) between two or more states
resulting in at least 25 battle-related deaths, internal armed conflict (type three conflict)
between the government of a state and internal opposition groups without intervention
from other states, and internationalized internal armed conflict (type four conflict)
between the government of a state and internal opposition groups with intervention from
other states.
19
Regarding the influence of the level of economic development, researchers have
found that countries that are more economically developed will have increased
government respect for physical integrity rights (Poe, Tate, and Keith 1999; Poe and Tate
1994). The rationale here is that governments with strong economies enjoy more security
from the threat of domestic rebellion; therefore, the healthier the economy, the less likely
a government will feel the need to engage in repression (Poe, Tate, and Keith 1999).
Given that gender disparities can have a detrimental impact on women’s socio-economic
rights, I would expect that the level of development is even more of a factor in promoting
respect for women’s rights than other human rights. Specifically, in less developed
Because preliminary results indicated that internal armed conflict and
internationalized internal armed conflict have similar effects on government respect for
physical integrity rights, these variables were collapsed into one measure for internal
armed conflict. To avoid a tautology, I do not use the internal conflict variables in the
series of analyses focusing on domestic conflict.
19
UCDP/PRIO also includes a variable for extrasystemic armed conflict (type 1 conflict) and war which
takes place between a state and a non-state group outside its own territory. This variable was not used
because there was a lack of this type of conflict in the observed states during the observed time period.
64
countries where there are fewer jobs to be had, men may feel more threatened by the
prospect of women entering the workforce. It could also be, as Coleman (2004) argued,
that improving women’s economic rights and welcoming women into the workforce
could boost development. In either case, it is important to control for economic
development when studying women’s rights as well as when studying physical integrity
rights. Likewise development is expected to be inversely related to protest. The
economic development of each country is measured using the country’s real Gross
Domestic Product per capita chain index (RGDPCH) from Penn World Tables (Heston,
Summers and Aten 2002). I have selected the GDP rather than the Gross National
Product (GNP) because the GDP includes all economic activity that occurs within the
borders of the given country. Previous studies have found that population is negatively
related to government respect for human rights because the larger the population the
greater the opportunity for rebellion and repression (Poe and Tate 1994; Poe, Tate, and
Keith 1999). Therefore, I include a log of population for each country/year from Penn
World Tables (Heston, Summers, and Aten 2002).
In looking at women’s rights, it is particularly important to consider cultural
attitudes toward gender equality. To measure these, I would have preferred to use the
Sexual Equality Scale developed by Inglehart and Norris (2003) from the World Values
Survey, but was unable to do so because of missing data issues. However, Inglehart and
Norris (2003) found a correlation between religion and attitudes toward gender equality.
In particular, they found that “the type of religion matters for beliefs about gender
equality far more than the strength of religiousity,” (Inglehart and Norris 2003: 67-68).
65
Specifically, they found that Catholic and Protestant faiths were positively related to
support for gender equality and Orthodox and Muslim faiths were negatively related with
support for gender equality. Therefore, I have opted to control for predominant religious
culture based on data from the CIA’s World Factbook, found in Norris’ Global Indicators
dataset (Norris 2005; Central Intelligence Agency 2009). These are dichotomous
variables for whether or not the state is predominately Catholic, Protestant, Orthodox or
Muslim (because there were too few states that were predominately Orthodox, this
variable was not included in the final analyses). It should be noted that using these
measures does not account for states that do not have a dominant religion.
Finally, it is standard practice to include a lag of government respect for human
rights (Davenport and Armstrong 2004; Poe and Tate 1994; Poe, Tate, and Keith 1999).
The theoretical reason for doing so is that previous respect for human rights may reflect
the cultural tolerance for such violations and also account for persistence in government
behavior regarding human rights. Methodologically this models any autoregression in the
series. However, because the time span covered is brief (15 years) compared to the
number of countries (93), including a lag of the dependent variable in the right side will
likely mask the effects of the independent variables (Achen 2000); therefore, I also
include a model without the lag. For the same reasons, I also use lags of the dependent
variables in the series of analyses looking at domestic conflict.
Using these measurements for the dependent and independent variables, I
designed a models for a multiple regression analyses using a pooled cross-sectional time-
series to test my hypotheses both across countries and across time. As mentioned above,
66
the dependent variable for physical integrity rights analysis and the women’s rights
analysis are the nine-point CIRI Physical Integrity Index and the 10 point women’s rights
index. With dependent variables of this type, I faced a choice between ordinary least
square (OLS) or either ordered logit or ordered probit. I chose to estimate an OLS model
first because of the ease with which it can be interpreted. I use the panel corrected
standard errors procedure to adjust for heteroscedasticity as suggested in Beck and Katz
(1995). Because tests indicated autocorrelation even with the inclusion of the lagged
dependent variable, I allow for a first-order autocorrelation (AR1). For the analyses using
the domestic conflict variables and the various indicators of physical integrity and
women’s rights I use logistic and ordered logistic regression and control for the
unspecified effects of each country by clustering on country code.
The basic model for the multiple regression is:
Government Respect for Human Rights
it
(or Protest
it
)=
α + β
1
Media Freedom
it
+β
2
Autocracy/Democracy
it
+ β
3
(Media Freedom
it
*Autocracy/Democracy
it
)
+ β
4
Government Respect for Human Rights
it-1
+ β
5
ln(GDP/Capita)
it
+ β
6
ln(Population)
it
+ β
7
International Conflict
it
+ β
8
Interstate Conflict
it
+ ε
it
67
Conclusion
In summary, I theorize that the effect of independent media on protest and
government respect for various human rights will vary depending on the vulnerability of
the government to public opinion. Governments with the democratic characteristics of
political competition, political participation and executive constraints offer citizens
institutional means through which to bring about reform, including but not limited to,
elections. In the absence of these institutions, protest may be the most viable means that
citizens have to bring about reform. However, a repressive regime that is threatened by
protest will likely resort to further repression to quell the dissenters. The role of
independent news media in the repression-protest nexus is two-fold. First the media can
provide information about government repression. Second the media can serve as a
facilitator because protest movement leaders can use it to mobilize the masses. I propose
that the effect of the news media is to promote dissent in repressive regimes that lack
democratic characteristics and that this in turn leads to further repression as the
governments of these regimes seek to suppress protest movements. In contrast, I posit
that in democratic regimes, the independent news media have a negative effect on
repression because the news media provides information about repressive behavior, but in
this case, the citizens are more likely to use institutional means, like elections or recalls,
than protest to bring about improved government behavior. In the next three chapters, I
test these hypotheses with a series of analyses and case studies. Chapter 4 focuses on the
effect of independent media on domestic conflict. In Chapters 5 and 6, I consider the
effect of media freedom on physical integrity rights and women’s rights respectively.
68
Chapter 4
The Role of Media in the Repression-Protest Nexus
There comes a time when the operation of the machine becomes so odious, makes you so
sick at heart, that you can't take part, you can't even passively take part, and you've got
to put your bodies upon the gears and upon the wheels, upon all the apparatus, and
you've got to make it stop. And you've got to indicate to the people who run it, the people
who own it, that unless you're free the machine will be prevented from working at all.
Mario Savio, American political activist 1964
Introduction
In December of 1983, German Araujo, director of Radio CX-30 in Montevideo,
Uruguay, went on a hunger strike to protest the government’s closure of his station
(Araujo 1984). The military leaders who controlled the country were angered by the
station’s live coverage of the national convention that nominated an opposition candidate
(Araujo 1984). Araujo recalled the experience in an interview:
The response from listeners and others was tremendous. Every day hundreds of
people congregated outside the building. It was another opportunity for them to
show that they were no longer afraid of any repression the authorities could mete
out (Araujo 1984: 15).
Araujo ended his hunger strike after 12 days when the government determined the station
would be able to reopen (Araujo 1984). Radio CX-30 was back on air on January 17,
1984, just in time to cover the general strikes that paralyzed the Uruguayan capital.
Araujo’s experience provides an example of how media can play both an
informational and an incendiary role in protest movements. This chapter focuses on the
relationship between media and contentious politics, specifically on the links between
69
media freedom and different types of protest. Because there is some dissent around the
definition of contentious politics, I clarify the definition used here in the first section. The
second section elaborates on the potential role of the media in protest movements that
was introduced in the previous chapter. In the third section, I present the findings of a
series of analyses of the relationship between media freedom and demonstrations, riots
and general strikes. I discuss these findings in the context of the case of Uruguay 1980 to
1985 when waves of protest paved the way for that country’s transition to democracy in
the fourth section. In the conclusion, I consider the implications of the findings from the
statistical analyses and the case of Uruguay.
The Controversial Definition of Contentious Politics
As Tilly (2004) so aptly put it, those who study dissent have “contentious
choices” to make. The first of which is the term and definition for the subject. The
choices include “collective action,” “social movements” and the all encompassing
“contentious politics” (Tilly 2004). Collective action studies tend to focus on the
challenges of coordinating dissent, largely in response to Olson’s (1965) work which
showed that it would be far more rational an individual to be a free rider and let others
take the risks and pay the costs of engaging in collective action. Yet while Olson was
writing, many were engaging in collective action in the form of anti-war protests. So,
researchers knew empirically that people would sometimes engage in collective action,
but they could not explain why. Hence, those who focus on collective action tend to look
at how resources and opportunities help to overcome these challenges. Resource
70
mobilization theory offered an answer to this puzzle (McCarthy and Zald 1977). In this
view, social movements are rational reactions to inequities in institutional power
relationships that came about when a shift in resources lowered the costs of mobilization
and improved the chances of success (McCarthy and Zald 1977). The problem with this
approach is first to define these resources and secondly to identify the shifts in resources
(Jenkins 1983).
Social movement research focuses more on processes that bring about political
change, and, as such, include a variety of actions from protests to revolutions. Yet, if
defined too broadly, this term is of little analytical use. For example, Earl (2003) studied
the Branch Davidian cult as a social movement. I posit that this cult, though anti-
government, was more of a separatist organization than a social movement. At the very
least a social movement should be understood as an entity that is actively engaged in
bringing about a change or some sort of response from government. Tilly (2004: 3-4)
defined social movements as sustained, organized collective action against government
employing the “social movement repertoire” (which includes protests, vigils, statements
to the media, and rallies) and displaying messages reinforcing what he terms as the
movement’s “WUNC” (worthiness, unity, numbers of participants and commitment).
This definition is also of limited use because on the one hand, it is a tautology (social
movements use social movement repertoire) and on the other hand, it incorporates an
element of success, at least in organization, into the definition of a social movement,
thereby rendering it useless in the study of popular uprisings that have failed. Also, it
would be difficult to form a social movement matching Tilly’s definition in an autocratic
71
setting, and this is precisely where we need to learn more about the potential for dissent
to bring about political change. In contrast to Tilly’s conception of social movements
there is the all-encompassing conception of contentious politics which are generally
understood to include social movements, revolutions, protests, labor movements, and
shaming rituals (McAdam, Tarrow and Tilly 2001).
It is possible to study the role of the media in facilitating collective action, as part
of the social movement repertoire and as a potential actor in contentious politics.
However, in order to analyze the effect of media freedom on dissent, I need a more
narrow and measurable definition. Therefore, I focus on protest movements which I
define as organized collective action aimed at engaging and bringing about a change in
government (usually, though not always, in respect to government’s treatment of citizens)
using non-institutionalized means of dissent.
20
The Role of News Media in Protest Movements
Protesting against government, especially an autocratic government, is risky
business. In order to analyze to potential involvement of media in protest movements, it
is important to first understand the factors that shape protest. As mentioned above,
changes in resources are one possibility. A similar approach focuses on political process
and structure, looking specifically at the role of political opportunities which can either
20
To measure protest movements I use the domestic conflict indicators for anti-government
demonstrations, riots and general strikes provided by the Cross-National Time-Series Dataset (Banks
2008). As outlined in Chapter 3, because of concerns about bias and because protest is rare, I have
transformed these count variables into dichotomous variables (strike/no strike, demonstration/no
demonstration, riot/no riot). For more information see the data section in the Chapter 3.
72
expand or contract and subsequently influence protest movements (McAdam 1982). With
both of these approaches, political liberalization, including the introduction of media
freedom would be seen as opening the door for the formation and mobilization of protest
movements. Likewise, a reduction in repression would be expected to increase the
likelihood of dissent. However both resource mobilization theory and the political
process model fail to explain why increased repression is sometimes met with increased
dissent. Goldstone and Tilly (2001) posited that rather than conceptualizing threat
(especially in the form of repression) as the opposite of opportunity, the two actually
work in conjunction with each other to drive and form dissent. They proposed to separate
threat from opportunity by redefining opportunity as anything that increases the chances
of the protest’s success and threat as “the costs that a social group will incur from
protest, or that it expects to suffer if it does not take action” (Goldstone and Tilly 2001:
182). In this view then, extensive repression could serve to increase the costs of inaction,
especially when repression is so severe that people perceive they have nothing more to
lose and everything to gain. It is important too that the repression-protest nexus is always
interactive:
An increase in repression or concessions is often followed by more protest, rather
than less. This can only be explained by realizing that opportunity is always in
interaction with current and repressive threats, and that this interaction gives rise
to varied dynamic patterns of protest and contention (Goldstone and Tilly 2001:
193).
Another way of conceptualizing this interaction is to focus on the intersection of
opportunity and willingness (Most and Starr 1993). In this view opportunity represents
the possibilities and willingness the choice. Here media intervene to make certain choices
73
more feasible. It is not that media launch protest movements; rather once the movements
are underway, the media can play a substantial intervening role. I hypothesize that media
can sometimes gain independence as part of this interaction between government and
opposition. For example, in the face of domestic conflict spurred by economic problems
combined with international pressures to reform, the leader of a repressive regime may
opt to allow some political liberalization, perhaps in the form of elections and some
media freedom, both of which the leader believes he can control covertly. This is
precisely what happened in Burkina Faso in the 1990s when Captain Blaise Compaore
responded to domestic and international demands for reform by holding elections and
allowing more media freedom (see more on this in Chapter 6). It is also similar to the
developments in Uruguay in the early 1980s when the military junta agreed to hold a
general election. While the goal of these concessions was to quell dissent, in the case of
Uruguay, the political opening fueled the opposition (Gillespie 1985). It follows too that
in the face of increased opposition, an authoritarian regime will resort to repression.
Thus, there is a pattern of concession and repression. Indeed, Rasler (1996: 149) argued
that “the pattern of tactical moves and countermoves among contending parties” is
important to understanding the development of protest movements.
There are several ways in which media enters into this pattern. First the media can
provide critical information about threats and opportunities. Secondly, free media are in
effect an opportunity because leaders of protest movements can use media to mobilize
support. Thirdly, the way in which the media frames the protest movement and
government response can shape public opinion. In all three cases, the influence of the
74
media is tempered by the extent to which the government controls media, the extent to
which leaders of protest movements have access to media, and the extent to which people
believe what they read and hear.
If the media are free to do so, they can provide information to the people about
government repression, opportunities for political participation, protest movements and
other opposition activity. This in turn can greatly facilitate mobilization of protest
movements. McCarthy and Zald (1977) argued that free media are particularly important
to leaders of protest movements who seek to mobilize people in remote regions. In
researching the role of media in democratization in African and Latin America, Randall
(1993: 636) found that once a protest movement was germinating, the media nurtured it
because, “they could widen awareness of issues and help put some kind of frame on
events. They could mobilize and orchestrate popular protest.”
The struggle over the framing of protest movements is indeed important. As
discussed in Chapter 2, there is much debate as to who controls the framing of news
stories, but certainly the media do not have exclusive control here, and frames are often
handed down to the media by political elites, both those in power and those in opposition
(Entman 2004). Regardless of who controls the frame, media framing of social movement
organizations helps to define the context in which these movement activities are viewed.
“Hypothetically, the more central or salient the espoused beliefs, ideas, and values of a
movement to the targets of mobilization, the greater the probability of their
mobilization,” (Benford and Snow 2000 pg. 621).
75
The relationship between the media and social movements is fueled by mutual
need in that the news media thrive on exciting stories and protest movements depend on
coverage to enhance both mobilization and the impact of any protest activity on the
government. Randall (1993) identified a pattern in which coverage of opposition activity
began in alternative media which was often controlled by the resistance. Over time
though, coverage of the protest movement seeped into the mainstream media for a variety
of reasons, “journalists eager to exercise the full range of their critical and investigative
professional skills were at last given their head by newspaper editors, and beyond them
owners, recognizing that what their readership wanted had changed and that if they didn’t
supply new needs another paper would,” (Randall 1993: 641).
Yet, the relationship between the media and protest movements is not a
relationship between equals. “Movements are generally much more dependent on media
than the reverse, and this fundamental asymmetry implies the greater power of the media
system in the transaction,” (Gamson and Wolfsfeld 1993: 116). Because of this
dependency, leaders of protest movements may strive to make their events more
appealing to the media (McCarthy and Zald 1977). However, the tendency of the media
to focus on the more extreme protest actions can have a backlash on protest movements
(O'Neill 2004). News coverage of violent protests draws attention away from the issues
and this in turn minimizes the legitimacy of the social movement. “The mainstream of the
movement has had a difficult battle itself to get media attention as a result and have, for
their own protection, had to move rapidly to marginalize radical groups and to adopt
explicitly nonviolent action manifestos” (O'Neill 2004: 243).
76
Media also can play a role in the longevity of a protest movement. Schock (1999:
368) argued that “the capacity to sustain challenging movements in the face of repression
depends significantly on the presence of influential allies and the ability to produce and
receive accurate information.” If media are sufficiently free they can provide this
information. Moreover through their connections with international media, independent
domestic media may also help draw the attention of “influential allies.”
Thus, if sufficiently free, media can play a substantial intervening role in the
repression-protest nexus. Specifically, I propose that media freedom in an autocratic
setting will serve to fuel protest movements. Likewise, I hypothesize that, contrary to
resource mobilization theory and the political process model, increased repression will
fan the flames of dissent. In an autocratic state where there are no institutionalized
mechanisms for people to express their discontent, the threat of repression combined with
the opportunity offered by free media create an environment that is primed for protest. Of
course, repression could also be seen as raising the costs of protesting because a
repressive regime is likely to retaliate with more repression; however, if the repression is
severe, people may feel that they have little left to lose. The potential effect of democracy
is similarly two-sided. One the one hand, democratic institutions offer people less costly
means of expressing discontent than protest, but on the other hand these same democratic
institutions could reduce the costs of protesting.
In the previous chapter, I outlined a theory of conditional media effects on human
rights. This theory is depicted in Figure 6 in Chapter 3, which shows a cycle in which
government repression in an autocratic state is covered by independent media and this
77
coverage subsequently leads to protest movements, which in turns leads to further
repression. In this chapter I focus on the middle step of this cycle, the effect of media
freedom on protest. Specifically, I look at the effects of media freedom,
autocracy/democracy, the interaction of media freedom and/autocracy democracy and
repression on the probability of a general strike, a demonstration, or a riot. It is important
to note that, according the criteria used in the Cross-National Time-Series Dataset, a
general strike is defined as a strike involving a thousand or more workers and “more than
one employer that is aimed at national government policies or authority,” (Banks 2008).
In contrast, an anti-government demonstrations and riots require a minimum of only 100
participants (Banks 2008). Therefore, by these definitions alone, general strikes require
far more participation and organization. Because general strikes are, according to this
definition, larger scale protests, I anticipate that the media will have a greater effect on
general strikes than it does with anti-government demonstrations and riots. As Van Belle
(2002: 120) argued, the difficulty with contentious politics is in mobilizing people to act
collectively:
If the violence and criminal acts were spread out over several days instead of all
occurring simultaneously, the police force could probably handle it. It is the fact
that it is all happening all at once that overwhelms the police, prevents them from
arresting the perpetrators and allows the mob to run free.
If the government controls the media, it may be able to convince people to suppress their
dissatisfaction about the government. In contrast, if the media is free, it is able to
facilitate collective action as opposition groups make use of the media to mobilize
citizens to protest. Expanding on the hypotheses outlined in the previous chapter, those
tested here are:
78
H
2
If there is Media freedom, there is increased likelihood of a general strike, riot
or anti-government demonstration.
H
3
The effect of media freedom on protest depends on the presence of democratic
characteristics of political participation, political competition and executive
constraints such that:
H
3a
With the interaction of media freedom and democracy, there is
decreased likelihood of a general strike, riot and/or anti-government
demonstration.
H
3b
With the interaction of government-controlled media and democracy,
there is no effect on the likelihood of a general strike, riot and/or
antigovernment demonstration.
H
3c
With the interaction of government-controlled media and autocracy,
there is no effect on the likelihood of a general strike, riot or anti-
government demonstration.
H
3d
With the interaction of free media and democracy, there is increased
likelihood of a general strike, riot or anti-government demonstration.
H
4
If there is repression, there is a greater likelihood of a general strike, riot or
antigovernment demonstration.
H
5
If there is democracy, there is less likelihood of a general strike, riot or
antigovernment demonstration.
In the next section, I present a study that tests these hypotheses about the relationship
between media freedom, autocracy/democracy, repression and protest.
Analyzing the Effect of Media Freedom on Protest
To test the above hypotheses, I conducted a series of logistic regression analyses
of the predicted probability of a country/year having a general strike, anti-government
demonstration, or a riot. Overall, the findings (summarized in Table 5) provide some
support for the hypothesis H
3
, that free media does have an effect on domestic conflict,
79
but that the effect depends on autocracy/democracy. Specifically, media freedom was
associated with an increased predicted probability of a general strike, but only in
autocratic states. However, neither media freedom nor autocracy/democracy had a
significant effect on the probability of a demonstration or a riot. So, H
2
was not supported
because media freedom had no independent effect on any form of protest. Moreover, H
5
was not supported because autocracy/democracy was actually associated with an
increased predicted probability of a general strike. (This finding is not completely
surprising because democracy does decrease the potential costs of protest.) Across all
analyses, government respect for physical integrity rights reduced the probability of
domestic conflict; thus H
4
is supported, decreased government respect for human rights
(meaning an increase in repression) was associated with an increased probability of all
three types of protest. Previous domestic conflict significantly and substantially increased
the probability of a general strike, demonstration or riot. Population did not have a
significant effect on general strikes, but did have a significant and positive effect on
demonstrations and riots. Thus, the more populous a state, the greater the probability that
it will experience a demonstration or riot. Interestingly, development did not make a
difference with domestic conflict. Likewise, interstate war did not have a significant
effect.
80
Table 5: Summary of the Effect of Free Media, Autocracy/Democracy and
Respect for Human Rights on Domestic Conflict
General
Strike
Demonstration Riot
Media Freedom +* NS NS
Autocracy/Democracy (five part) +* NS NS
Media
Freedom*Autocracy/Democracy
-* NS NS
General Strike (-1 year) + + +
Demonstration (-1 year) + + +
Riot (-1 year) + + +
Respect for Physical Integrity
Rights
- - -
Population (logged) NS + +
GDP/capita (logged) NS NS NS
Interstate War NS NS NS
NS=not significant; “+”=significant and positive; “-“=significant and negative
*interpretation of interactive term needs to be considered in conjunction with
components of the term (see figures)
General strikes are rare events. Of the 1494 cases included in this study, only 230
of them had one or more general strike (these analyses covered 122 countries from 1981
to 1995) (Banks 2008). As the results from a series of logistic regressions (shown in
Table 6) indicate, a general strike in the past year was a significant predictor of a general
strike. Specifically, a country with a previous general strike had 4.5 greater odds of
having a general strike, holding other predictors constant at their mean. For example,
from 1981 to 1995, Uruguay experienced 16 general strikes (6 of which occurred in 1990
in response to efforts to privatize the economy), France had 9, and Bangladesh had 22.
Likewise previous demonstrations and previous riots also increased the odds of a general
strike by factors of 1.9 and 1.7 respectively. In contrast, every increase in government
respect for physical integrity rights decreased the chances of a general strike. For
81
example, a state with no government respect for these rights had a predicted probability
of .17 of having a strike, whereas a state with complete government respect for these
rights had a predicted probability of only .08 of having a strike, holding all other
predictors constant.
As shown in Model 1 of Table 6, media freedom did not have an independent
effect on the predicted probability of a general strike and autocracy/democracy had only a
slight significant and positive independent effect. However, as shown in model 2, the
interaction of media freedom and autocracy/democracy did have a statistically significant
effect. This effect is best interpreted graphically. As depicted in Figure 9, in the most
consolidated democracies the predicted probability of a general strike is about the same
for states with free media as it is for states with controlled media. As democracy
decreases, media freedom is associated with a higher predicted probability of a general
strike. Overall, the magnitude of this effect is slight, but in autocracies, media freedom is
associated with an increased predicted probability of a general strike.
21
21
Because general strikes are rare, they are more difficult to predict. For this reason, the standard error
increases as the predicted probability increases. Logistic regression tends to underestimate the probability
of rare events, therefore, in reality the probability of a general strike may be higher than what is depicted in
Figure 2. For more on this, see the technical report at the end of this chapter.
82
Figure 9: The Effect of Media Freedom on the Predicted Probability of a General
Strike as Autocracy/Democracy Changes
0 .1 .2 .3
Probability of General Strike
0 1 2 3 4
Autocracy/Democracy
Free Media Controlled Media
Table 6: The Effect of Media Freedom and Autocracy/Democracy on the Probability of a General Strike
Dependent Variable:
General Strike
Model 1 Model 2 Model 3 Model 3
Odds Ratio
General Strike (-1 year) 1.835 (.230)*** 1.792 (.237)*** 1.511 (.239)*** 4.53***
Media Freedom .491 (.339) 1.444 (.518)** 1.539 (.487)** 4.66**
Autocracy/Democracy (five part) .2864 (.126)* .432 (.131)** .428 (.131)** 1.53**
Media Freedom*Autocracy/Democracy --- -.427 (.176)* -.440 (131)* .64*
Demonstration (-1 year) --- --- .642 (.207)** 1.9**
Riot (-1 year) --- --- .553 (.201)** 1.74**
Respect for Physical Integrity Rights -.172 (.047)*** -.149 (.046)** -.115 (.045)* .89*
Population (logged) .145 (.088) .152 (.090) .092 (.086) 1.1
GDP/capita (logged) -.060 (.155) -.001 (.167) -.024 (.160) .98
Interstate War -.175 (.360) -.118 (.359) -.305 (.358) .73
Constant -3.036 (1.270)* -3.822 (1.370) -3.545 (1.348)** ---
N 1494 1494 1494 1494
Wald χ
2
181.63 194.58 227.49 227.49
BIC -9810.892 -9812.334 -9828.678 -9828.678
McKelvey & Zavoina’s R
2
.25 .26 .30 .30
*p<.05 ** p<.01 ***p<.001
83
84
Compared to general strikes, anti-government demonstrations were more
common. In this sample there were 416 country/years with demonstrations. Among those
countries that had demonstrations between 1981 and 1995, the United States had 21, the
Dominican Republic had 19, and South Africa had 76 (26 of which occurred in 1985
during the anti-Apartheid movement) (Banks 2008). Unlike general strikes, media
freedom and autocracy/democracy did not have a significant effect on the predicted
probability of demonstrations. Instead, the results (shown in Table 7) indicate that
government respect for physical integrity rights and population had significant effects on
the predicted probability of a demonstration. In particular, every increase in government
respect for human rights decreased the chances of a demonstration. Figure 10 depicts this
effect and shows that, holding all other predictors constant, a state with no respect for
physical integrity rights had a predicted probability of .40 for a demonstration, whereas a
state with full respect had a predicted probability of .17 for a demonstration. The effect of
population was significant and positive, meaning that the greater the population, the
greater the odds of having a demonstration. As with general strikes, previous domestic
conflict was associated with a higher predicted probability for a demonstration. States
with a previous demonstration had 2.6 greater odds for a demonstration and states with a
previous general strike or a previous riot had 1.8 greater odds for demonstration, holding
all other predictors constant.
85
Figure 10: The Effect of Government Respect for Physical Integrity Rights on the
Predicted Probability of a Demonstration
.1 .2 .3 .4 .5
Probability of Demonstration
0 2 4 6 8
Government Respect for Physical Integrity Rights
Predicted Probability 95% upper limit
95% lower limit
86
Figure 11: The Effect of Government Respect for Physical Integrity Rights
on the Predicted Probability of a Riot
.05 .1 .15 .2 .25 .3
Probability of Riot
0 2 4 6 8
Government Respect for Physical Integrity Rights
Predicted Probability
95% upper limit 95% lower limit
Table 7: It’s all About Respect—the Effect o Media Freedom and Autocracy/Democracy on Demonstrations
121 Countries, 1981-1995
Dependent Variable:
Demonstration
Model 1 Model 2 Model 3 Model 3
Odds Ratio
Demonstration (-1 year) .970 (.158)*** 1.316 (.149)*** .965 (.157)*** 2.62***
Media Freedom .096 (.211) .604 (.398) .524 (.371) 1.69
Autocracy/Democracy (five part) .111 (.083) .205 (.093)* .165 (.091) 1.17
Media Freedom*Autocracy/Democracy --- -.222 (.140) -.188 (.136) .83
General Strike (-1 year) --- --- .587 (.182)** 1.80**
Riot (-1 year) --- --- .592 (.178)** 1.81**
Respect for Physical Integrity Rights -.154 (.035)*** -.157 (.036)*** -.145 (.036)*** .87***
Population (logged) .233 (.054)*** .260 (.061)*** .235 (.055)*** 1.27***
GDP/capita (logged) .198 (.115) .193 (.126) .225 (.119) 1.25
Interstate War -.338 (.314) -.285 (.323) -.311 (.311) .73
Constant -4.858 (.887)*** -5.028 (.968)*** -5.175 (.931)***
N 1492 1492 1492 1492
Wald χ
2
194.71 175.58 195.75 195.75
BIC -9352.355 -9336.039 -9347.381 -9347.381
McKelvey & Zavoina’s R
2
.25 .23 .25 .25
*p<.05; **p<.01;*** p<.001
87
Table 8: Respect Matters with Riots too—the Influences on the Predicted Probability of Riots
121 countries, 1981-1995
Dependent Variable:
Riot
Model 1 Model 2 Model 3 Model 3
Odds Ratio
Riot (-1 year) 1.688(.199)*** 1.681 (.198)*** 1.243 (.215)*** 3.47***
Media Freedom -.159 (.277) .120 (.511) -.011 (.494) .99
Autocracy/Democracy (five part) .034 (.096) .070 (.097) .001 (.097) 1.00
Media Freedom*Autocracy/Democracy --- -.127 (.176) -.091 (.174) .91
General Strike (-1 year) --- --- .483 (.214)* 1.62*
Demonstration (-1 year) --- --- .664 (.204)** 1.94**
Respect for Physical Integrity Rights -.149 (.047)** -.142 (.047)** -.125 (.047)** .88 **
Population (logged) .341 (.078)*** .344 (.078)*** .330 (.075)*** 1.39***
GDP/capita (logged) .048 (.132) .062 (.137) .044 (.132) 1.04
Interstate War -.574 (.359) -.552 (.356) -.588 (353) .56
Constant -4.977 (1.051)*** -5.169 (1.105)*** -5.058 (1.034)***
N 1494 1494 1494 1494
Wald χ
2
149.34 149.59 172.78 172.78
BIC -9722.789 -9716.225 -9723.530 -9723.530
McKelvey & Zavoina’s R
2
.25 .26 .28 .28
*p<.05; **p<.01;*** p<.001
88
89
Riots were less common than anti-government demonstrations, but more common
than general strikes. Out of 1494 country/years included in these analyses, 262
experienced one or more riot. These include the United Kingdom which experienced 34
riots, Nigeria which had 20 and India which had 140 (many of these were ethnic riots—
there were 26 each in 1983 and 1984). Once again, the results of the logistic regressions
(shown in Table 8) indicate that previous domestic conflict was a significant predictor of
the probability of a riot. Specifically, states with a previous riot had 3.5 greater odds of
experiencing a riot, those that had a previous strike had 1.6 greater odds of a riot, and
those that had a previous demonstration had 1.9 greater odds of a riot. Neither media
freedom nor autocracy/democracy had a significant effect. Instead, as with
demonstrations, population had a significant and positive effect on the predicted
probability of a riot. As with both general strikes and demonstrations, government respect
for physical integrity rights decreased the likelihood of a riot. In particular, states with
full respect for these rights had a predicted probability of .09 for a riot, compared to .21
for states with no respect for these rights, holding all other predictors constant at their
means (this effect is depicted in Figure 11).
These results suggest that repression and all types of domestic conflict increase
the likelihood of future strikes, anti-government demonstrations and riots; however,
media freedom and autocracy/democracy only make a difference with general strikes.
Perhaps this is because, according to the definition used here, general strikes require
greater participation (ten times that of anti-government demonstrations and riots) and
therefore entail greater organization and commitment than do anti-government
90
demonstrations and riots. Thus, it is not surprising that media freedom makes a difference
with general strikes rather than the other two types of dissent. If measurements with
larger participation thresholds were available for the other two types of domestic conflict,
the results might be different. In any case, the finding that media freedom increases the
predicted probability of a strike in autocratic settings supports H
3
, the hypothesis that
media freedom will increase dissent in autocracies.
It is somewhat counterintuitive that autocracy/democracy does not have a
significant effect on riots and demonstrations. This could be because on the one hand
democracy lowers the cost of such dissent, but on the other hand, democracy offers
dissenters other institutional means with which to exercise their voice. Across the board,
repression of physical integrity rights is associated with increased probability of domestic
conflict. This finding suggests that regardless of the level of autocracy/democracy,
repression will nurture dissent. In the next section, I explore the implications of these
results in the context of the protest movements that accompanied Uruguay’s transition to
democracy in the 1980s.
Uruguayan Military Regime’s Humiliation Caught on Camera
The case of Uruguay in the first half of the 1980s offers a unique opportunity to
examine the effects of free media in a country that has undergone substantial changes in
both media freedom and regime type in the past century. From the mid 1950s to the early
1970s, Uruguay was a democracy and Uruguayan media enjoyed substantial freedoms.
However, starting in the late 1960s, a rebel group so threatened Uruguay’s national
91
security that by 1973 Uruguay was under martial law with substantial constraints on all
civil liberties, including media freedom. Thus as the Uruguayan independent media
began to emerge in the first half of the 1980s, the military government remained firmly
autocratic. Moreover, having risen to power by virtue of its ability to suppress rebellion,
Uruguay’s military had a brutal reputation for repressing all opposition. This confluence
of characteristics makes Uruguay a particularly interesting case.
In the 1990s, Uruguay’s military dictatorship bowed to international pressures and
allowed Uruguayans to vote on a constitution that would have institutionalized the armed
forces’ control of the country. The generals were so assured of their victory that they
allowed the votes to be counted on television (Steele 1984). In campaign advertisements
broadcast on radio and television and printed in newspapers, the military reminded
citizens of the national security threats posed by leftist guerillas prior to the military’s
intervention and claimed the new constitution would improve national security (Goodsell
1980). Buoyed by the success of a similar measure in Chile, the military expected to win
and knew that if it did not allow Uruguayans to vote the constitution would be
condemned by the United States (Schumacher 1980). The military was so confident the
constitution would prevail and so eager to gain legitimacy that it kept the election free
from fraud and allowed some public debate.
Early independent opinion polls showed the Government leading by more than 2
to 1, though the poll takers said that almost half the people they approached
refused to respond. By last week at least one poll had data showing that the gap
had closed. But it was not published, partly out of fear of the Government's
reaction, a poll taker said. The change appeared to begin on Nov. 14, t he night of
the first televised debate between drafters of the constitution and opponents. For
two and a half hours viewers saw Government officials being criticized to their
face on national television. (Schumacher 1980)
92
In addition to the overt criticism of the constitution, members of Uruguay’s two political
parties (the Blancos and the Colorados) launched a secret campaign through cassette
tapes that were passed from person to person.
22
The defeat of the plebiscite stunned the generals. The military once again
muzzled the media and banned political activity and for months refused to comment on
the failed constitution (Schumacher 1980; Araujo 1984). Yet, not all media were silent.
The weekly Opinar, continued to carry news about the plebiscite (Gorney 1981). In all,
though the repression continued and the opposition remained underground during and
after the vote, this event marked the beginning of the end for the military regime and
breathed life into the protest movement. As one member of the Blanco party stated, “It
was a protest. We didn’t choose between two constitutions. It was just getting out the
hatred—no, no, no,” (Gorney 1981).
These recordings featured speeches by
party leaders which urged people to vote “no” (Gorney 1981).
Uruguay: From Democracy to Dictatorship to Democracy
In the first half of the 20
th
century Uruguay was dubbed the Switzerland of Latin
America in recognition of its economic development, democratic government and
welfare state (Hudson and Meditz 1990).
22
Since Uruguay’s independence in 1830, its political system has been dominated by the Colorado (called
the Colorados) and the National (called the Blancos) parties. These parties evolved around two local
leaders who gained followers because of “their power, bravery or wealth” and were eventually each in turn
elected president (Hudson and Meditz 1990). Loyalty to these parties was more tradition-based than issue-
oriented, but in general, the Colorado Party tended to be the party of choice for labor unions and urban
residents, whereas the National Party was seen as the more conservative of the two and appealed to rural
citizens and farming groups. These two parties dominated until the late 1980’s when a third party, the
Broad Front gained popularity (Hudson and Meditz 1990).
93
During the sixty-year period from 1870 to 1930, foreign immigrants flooded into
Uruguay, mainly from Spain and Italy, to improve their standard of living. A
historical study of social and economic development ranked Uruguay fourth
among all independent nations in the world in the 1880s. (Hudson and Meditz
1990)
Life for Uruguayans changed dramatically in the late 1950s when the economy began a
downward spiral marking the beginning of what some termed the “Latin
Americanization” of Uruguay (Hudson and Meditz 1990). During the 1960s Uruguay’s
democratic government faced increasing threats from an urban guerilla movement known
as the National Liberation Movement-Tupamaros (MLN-T) (Marshall and Jaggers 2007).
These threats paved the way for the military to seize control of the country in 1973. In
spite of the defeat of the plebiscite in 1980, the military remained in power until the
elections of 1984. Gillespie (1985: 102) argued that by then the military had little choice
but to hand over control of the country due to the economic situation which was so
desperate that civil unrest was imminent because, “real earnings fell to less than a half of
their 1968 value by 1984.” The quandary for the military was that they did not want to
relinquish power to their political enemies for fear of reprisals.
Throughout its tenure, the military dictatorship maintained a strict police state.
Political activity, including participation in political parties, was banned. Moreover, the
government maintained a computerized system ranking citizens based on their political
ideology into A, B and C categories.
Those falling in category ‘C’ could not get posts in public employment and could
not get passports. These were people who had belonged to a banned political or
student group; membership even a long time ago was sufficient to be placed in
94
‘C’. People given a ‘B’ were believed to be of questionable loyalty and would be
watched, whereas ‘A’ meant a ‘good’ citizen (Heinz and Fruhling 1999).
In 1975 with nearly 7000 political prisoners, Uruguay had “probably the highest
proportion of citizens jailed for political reasons in the world,” (Riding 1984). Many of
those imprisoned were tortured and some disappeared (Hudson and Meditz 1990)
The pattern of Uruguay’s eventual democratization is depicted in Figure 12 which
shows that improvements in media freedom preceded Uruguay’s democratization.
Government respect for physical integrity rights varied, but was generally low in the
early 1980’s. Initially, government respect for human rights declined as media freedom
improved, but as Uruguay democratized, government respect improved dramatically.
95
Figure 12: Media Freedom, Autocracy/Democracy and Government Respect for
Human Rights in Uruguay
23
The Uruguayan Media
Prior to the imposition of martial law, Uruguay’s media was free in that they were
able to criticize the government. Hudson and Meditz (1990) posited that the news
media’s affiliations with political parties impeded their ability to serve as a Fourth
23
This figure was generated using STATA 10. The measure for autocracy/democracy is based on the Polity
score, the measure for media freedom is from the Global Press Freedom Index and the measure for
government respect for physical integrity rights is from the CIRI Human Rights dataset. For ease of
interpretation, each of these measures has been normalized so that they all are between 0 and 1.
Additionally media freedom has been transposed, so that 0 is government controlled media and 1 is free
media. For the autocracy/democracy measure 0=autocracy and 1=democracy. Likewise, 0=no government
respect for human rights and 1= full government respect.
.2 .4 .6 .8 1
1980 1985 1990 1995
year
Media Freedom Autocracy/Democracy
Government Respect for Physical Integrity Rights
96
Estate. Similarly Randall (1993: 632) described the Uruguayan media prior to the
military dictatorship as “a lively combative press and private sector radio, both generally
aligned with the main political parties.” The Global Press Freedom Index lists the media
in Uruguay as free from 1948 to 1974 and 1984 to 1995, meaning that, at the very least,
the media was able to criticize the government during these periods.
During the military regime, it was difficult for the media to function
independently. According to the director of Radio CX-30, German Araujo, prior to 1980
it was challenging to provide political information.
I was determined to give as much news as possible, but no commentary was
allowed about Uruguay’s internal affairs. Our silence, rather than the open
support for the military government which people heard on other radio stations,
was the most significant thing we could do then. This made life difficult with the
authorities—they withdrew all official advertising, for example, but it did gain us
lots of listeners, (Araujo 1984: 15).
Rather than report on the situation in Uruguay, Araujo reported on the democratization of
Spain and the military governments in Argentina and Chile and “People learned ‘to listen
between the lines,’” (Araujo 1984: 15). 1980 was a turning point in media freedom
because the station was then able to critique the proposed constitution and even arranged
for listeners to monitor the election against fraud (Araujo 1984). After 1980 the
government began to permit more media freedom, but it was a gradual process (Araujo
1984).
97
1982 to 1984: The Protests
The military government’s announced plan to return control of the country to the
civilians did not appease the people. If anything, the setting of the timetable seemed to
fuel the opposition. The plan was to elect political party leaders in 1982 to prepare for a
presidential election in 1984. From 1982 to 1984, there were a series of protests,
culminating in a massive general strike in January of 1984. The first of these occurred in
October of 1982, when the Colorado Party launched its campaign for the elections slated
for the following month with what some termed Uruguay’s “first public political
demonstration in 10 years” (1982).
The election was held to select about 500 people to attend the political party
conventions where the leaders of each party would be selected in preparation for the
presidential election in 1984 (Burns 1982). On the day of the election, Uruguayans
celebrated:
“Liberty, liberty,” some yelled as they strode through the streets. “It’s going to
end, it’s going to end. The military dictatorship is going to end,” others shouted.
Drivers of the city’s multitude of old Ford cars showed their approval, letting out
a few rusty horn blasts,” (Burns 1982)
With their votes, the Uruguayans made known their disdain for the junta. The majority of
those selected were anti-military candidates (Burns 1982).
In 1983, protests became more common as the opposition began to organize
regular demonstrations. In part, the protests were aimed at pressuring the military to keep
to its timetable for the elections. In September of 1983, one such protest drew more than
25,000 workers and students marched through the streets of Montevideo (Diehl 1983).
The junta responded by arresting a prominent member of the Blanco Party (Goodsell
98
1983). Indeed, in its waning days, the military dictatorship of Uruguay became
increasingly repressive, arresting political dissidents, prohibiting most political activity
and censoring the media (Schumacher 1984).
El Dia, a leading daily, was banned for four days for publishing an interview from
Buenos Aires with Mr. Ferreira Aldunate [banned member of Blancos Party]. An
opposition radio station has been shut down for a month for its live coverage of
the national convention that nominated Mr. Ferreira, and, for the first time ever,
the Government has imposed prior censorship. The country’s 14 political
weeklies must be cleared by the military before they can be sold. A number of
issues have been rejected. (Schumacher 1984)
In addition to this crackdown on political communication, the generals began to hint that
they might postpone the election.
On January 18, Montevideo was paralyzed by a general strike that left the streets
deserted, shops shuttered and factories stalled (1984). The government banned the union
confederation, sent in soldiers to break up a sit-in at a textile factory and confiscated three
news magazines that ignored a ban on news coverage of the strike (1984). Citizens
responded with public pot-banging. Strikes continued throughout 1984, though only the
one in January reached the threshold of a thousand participants (Banks 2008). In addition
to the strikes, there were non-violent anti-government demonstrations and riots. Figure 13
illustrates the pattern of protests.
99
Figure 13: Domestic Conflict in Uruguay
24
In spite of the generals’ threats that the unrest might force them to delay the
election, it went on as scheduled:
A foghorn is sounded in the Montevideo city centre, making the beginning of the
poll. The horn is sounded only for momentous events, such as the declaration of
the Second World War or the assassination of President Kennedy. Bus and car
drivers joined in with their horns, momentarily breaking the calm which had taken
over the capital’s streets after the election fever of the past few weeks (Coad
1984).
Julio Maria Sanguinetti, the Colorado Party candidate, won the election. He restored
media freedom on March 1, 1985, the day he was sworn in. (Hudson and Meditz 1990).
24
This Figure shows the pattern of domestic conflict in Uruguay from 1981 to 1995. It was generated using
STATA 10 and the domestic conflict data from the Cross-National Time-Series Dataset (Banks 2008).
Aside from the protests in the first half of the 1980s, there were also general strikes and some anti-
government demonstrations in the 1990s in response to President Luis Alberto Lacalle’s efforts to privatize
the economy (Christian 1990).
0 2 4 6
1980 1985 1990 1995
year
General Strikes Riots
Anti-government Demonstrations
100
Overall, the contentious relationship between the Uruguayan government and
citizens during the first half of the 1980s is much closer to Rasler’s (1996) concept of a
pattern of moves and countermoves than either resource mobilization theory or the
political process model. For example, in response to international pressures and economic
problems, the generals opened the door to political participation with the 1980 plebiscite
and allowed the media more freedom. The opposition took advantage of this opening and
began its campaign to undermine the military with its underground circulations of the
cassettes and overt criticism in the opposition media. Likewise, Goldstone and Tilly’s
(2001) argument that protest movements are shaped by both threats and opportunities is
supported. The people protested against the military with their vote against the
constitution, and the generals responded by clamping down on political participation and
political communication. However, the people were empowered by the process and saw
the opportunity to push for reform. Support for the military government diminished and
in spite of the threat posed by the junta’s reputation for repression, anti-government
protests ensued. Clearly, the combination of increasing media freedom and on-going
repression helped jumpstart the opposition’s “vote no” campaign. Once the resistance
gained momentum, the military’s repressive response and subsequent threats to postpone
the election only served to fuel the massive protests in 1983 and 1984. Throughout this
process, media like German’s Radio CX and the weekly newsmagazine Opinar provided
information about both the government and the protest movement.
101
Conclusion
The statistical analyses presented here provide support for the hypothesis that
media freedom has an effect on protest that varies depending on the presence or absence
of democratic institutions. Specifically, media freedom was associated with an increased
predicted probability of a general strike in an autocracy. It did not have a significant
effect in consolidated democracies, nor did it have a significant effect on the predicted
probability of an anti-government demonstration or a riot, but this is not surprising given
the much higher threshold of participation for a general strike (1000) than for an anti-
government demonstration or riot (100). Consequently, I posit that in a non-democracy,
media freedom will likely have a significant effect on the probability of a large scale
protest, but not on that of a small scale protest.
As the case of the protest movement in Uruguay illustrates, contentious politics
are interactive and the role of the media is at the center of this interaction. The military
junta and opposition leaders tried different tactics and adapted new strategies in the effort
to gain power and control. Similarly some journalists seized the opportunity to provide
information and report critically when the military relaxed its control and some defied
military orders to keep quiet when the military clamped down. Clearly the media,
especially those affiliated with the opposition, played a role in mobilizing the resistance
in the months leading up to the defeat of the plebiscite and the year of protests and strikes
that ended in the presidential election. Even when the media were greatly restricted, some
persisted in providing information about political possibilities through their coverage of
political activity in other countries.
102
Going back to the cycle depicted in Figure 6 (Chapter 3), both the statistical
findings and the case of Uruguay provide support for idea that media coverage of
government repression will lead to protest. While the protests in Uruguay did initially
prompt a repressive response from the military junta, the repression eventually subsided
and Uruguay’s subsequent democratization allowed it to break out of the cycle of
repression and dissent. As Uruguay returned to democracy, government respect for
human rights improved dramatically. However, this is not always the case. In the next
chapter I move to the next segment of the cycle and consider the relationship between
media freedom, autocracy/democracy and government respect for human rights.
Technical Report
Because media freedom is often considered to be an aspect of democracy, it was
important to consider that there might be an issue with multicollinearity with the media
freedom and autocracy/democracy variables. Table 9 shows the results of collinearity
diagnostics which reveal that there is no problem with the variance inflation factors for
the model without the interactive term (Model 1), but that there are some issues in the
model with the interactive term (Model 2). This was expected because any interactive
term by definition is going to have some collinearity with its constituent terms. However,
the primary problem with collinearity is that it tends to inflate standard error and
therefore make it difficult to obtain significant results (Achen 1982). Therefore, because
the interactive term was significant, the collinearity here is of little concern.
Multicollinearity was a substantial problem with components of democracy (as shown in
103
Table 10). Although I tried various combinations of the components of democracy, the
multicollinearity was so extreme that I was unable to obtain significant results. Therefore,
these models are not included in the findings.
Table 9: Collinearity Diagnostics
VIF VIF
Strike (-1 year) 1.19 1.19
Media Freedom 2.09 6.65
Autocracy/Democracy (5
part)
3.34 4.16
Media Freedom*
Autocracy/Democracy
--- 10.76
Demonstration (-1 year) 1.49 1.50
Riot (-1 year) 1.46 1.46
Respect for Human Rights 1.60 1.65
GDP/capita (logged) 2.14 2.24
Population (logged 1.30 1.30
Interstate Conflict 1.07 1.07
Mean VIF 1.74 3.20
Table 10: Collinearity Diagnostics Among Components of Democracy
VIF VIF VIF
Strike (-1 year) 1.23 1.21 1.21
Media Freedom 1.73 1.73 1.46
Regulation of Participation 80.29 41.62 ---
Competitiveness of
Participation
307.36 --- 104.00
Executive Constraints 191.08 43.23 ---
Executive Recruitment 139.64 --- 104.63
Demonstration (-1 year) 1.51 1.51 1.49
Riot (-1 year) 1.46 1.31 1.46
Respect for Human Rights 1.58 1.57 1.57
GDP/capita (logged) 1.69 1.31 1.57
Population (logged 1.34 1.66 1.32
Interstate Conflict 1.07 1.07 1.07
Mean VIF 60.83 9.64 21.98
104
I conducted a variety of diagnostics of the residuals from the logistic regressions
and identified several country/years as potential influential cases. Table 11 shows the
analysis for General Strikes with the cases included (model one), without the influential
cases of Uruguay 1984, Panama 1989, Greece 1991, and Philippines 1984 & 1985 (model
2) and without the cases identified as having a higher DFBETA score (this measures the
influence of a particular observation on each predictor) Spain 1991, France 1991,
Uruguay 1984, Argentina 1982 and Brazil 1983. Excluding these cases did not
substantively change the results.
Table 11: Models with and without Influential Cases N=661
Dependent Variable:
General Strike
All Cases Without
Influential
Cases
Without
DFBETA
Cases
General Strike (-1 year) 1.511 (.239)*** 1.507 (.246)*** 1.528
(.239)***
Media Freedom 1.539 (.487)** 1.546 (.518)** 1.363 (.506)**
Autocracy/Democracy (5 part) .428 (.131)** .423 (.127)** .434 (.132)***
Media Freedom
*Autocracy/Democracy
-.440 (131)* -.436 (.185)* -.391 (.183)*
Demonstration (-1 year) .642 (.207)** .635 (.208)** .647 (.209)**
Riot (-1 year) .553 (.201)** .542 (.208)** .525 (.209)*
Respect for Physical Integrity
Rights
-.115 (.045)* -.122 (.046)* -.125 (.045)**
Population (logged) .092 (.086) .097 (.086) .098 (.086)
GDP/capita (logged) -.024 (.160) -.027 (.165) -.090 (.160)
Interstate War -.305 (.358) -.331 (.382) -.766 (.401)
Constant -3.545 (1.348)** -3.532 (1.386)* -3.069 (1.329)*
N 1494 1489 1489
Wald χ
2
227.49 232.89 221.91
BIC -9828.678 -9794.029 -9805.583
McKelvey & Zavoina’s R
2
.30 .30 .31
p<.05; **p<.01;*** p<.001
105
The diagnostics did indicate that the models were much better at predicting the
probability of no domestic conflict than they were at predicting the probability of
domestic conflict. I expect that in general these analyses have underestimated the
probability of domestic conflict. Mass protests, especially general strikes, are rare events
and logistic regression tends to underestimate the probability of rare events (King and
Zeng 1999). Therefore I also ran these models using a special procedure aimed at
correcting some of these estimation problems (Tomz, King and Zeng 1999 ). The results
were substantively the same. Table 12 shows the comparison of the results for the general
strikes analyses.
Table 12: Logistic Regression vs. Rare Events Logistic Regression
Dependent Variable:
General Strike
Logistic
Regression
Rare Events
Regression
General Strike (-1 year) 1.511 (.239)*** 1.492 (.237)***
Media Freedom 1.539 (.487)** 1.536 (.484)**
Autocracy/Democracy (five part) .428 (.131)** .421 (.129)**
Media Freedom*Autocracy/Democracy -.440 (131)* -.441 (.174)*
Demonstration (-1 year) .642 (.207)** .638 (.206)**
Riot (-1 year) .553 (.201)** .546 (.199)**
Respect for Physical Integrity Rights -.115 (.045)* -.113 (.045)*
Population (logged) .092 (.086) .092 (.085)
GDP/capita (logged) -.024 (.160) -.022 (.159)
Interstate War -.305 (.358) -.275 (.356)
Constant -3.545
(1.348)**
-3.523 (1.34)**
N 1494 1494
Wald χ
2
227.49
BIC -9828.678
McKelvey & Zavoina’s R
2
.30
*p<.05; **p<.01;*** p<.001
106
Chapter 5
Media Freedom, Democracy and Physical Integrity Rights
“Silence never won rights. They are not handed down from above; they are forced by
pressures from below.”
Roger Nash Baldwin
Introduction
On New Year’s Day 1994, hundreds of armed peasants took over radio stations
and seized control of five towns in the south of Mexico in what became known as the
Zapatista uprising. The rebels called themselves the Ejercito Zapatista de Liberacion
Nacional (EZLN) and called for the resignation of President Salinas claiming that the
government had ignored the plight of the indigenous people.
“We are the product of 500 years of victimization that began with the
Spanish,” said one guerrilla leader dressed in new army fatigues, a red
bandana around his neck, and an automatic rifle slung over his shoulder.
“We are dying of hunger and disease,” he said reading a statement on
television. “We have nothing, absolutely nothing: not a decent roof, nor
land, nor work, nor education…Today, we say: Enough!” (Scott 1994).
The uprising captured the attention of the Mexican and the international media
alike. In addition to documenting the fighting and negotiations between the EZLN and
the Mexican government, the Mexican media, traditionally hesitant to criticize the
government, provided coverage of the Mexican military’s brutal treatment of suspected
rebels. “As they reported on events like the EZLN uprising, many journalists in Mexico
were pleased that their work supported social movements and activities such as human
rights,” (Hughes 2006: 8). Indeed, the idea that independent news media can support the
107
oppressed by drawing attention to their plight and thereby putting pressure on the
government to change is a common theme from human rights organizations (Amnesty
International 2006). Yet, there is no guarantee that drawing attention to human rights
violations will actually reduce repression. This chapter focuses on the relationship
between and among media freedom democratic institutions and government respect for
physical integrity rights.
In the first section, I review the relationship between media freedom and
democracy as it pertains to physical integrity rights. Specifically, I expand on the
potential interactive role of the media freedom and autocracy/democracy that was
introduced in Chapter 3. In particular, the study presented here tests the relationship
depicted in the cycles illustrated in Figures 5-8 in the Chapter 3. The second section
features the results of a series of statistical analyses looking at both the contemporaneous
and future effects of media freedom and democratic institutions on government respect
for physical integrity rights. The third section discusses these findings in the context of
two non-democracies with independent media, Mexico and Uganda from 1981 to 1995.
During this time Mexico was democratizing, whereas Uganda remained authoritarian. In
both states, government respect for physical rights was quite low. Thus the two cases
offer different perspectives of the effect of media freedom on physical integrity rights in
the absence of democracy. In the concluding section, I consider the implications of these
findings.
108
Media Freedom, Democratic Institutions and Physical Integrity Rights
It is well established that countries that are fully democratic are less likely to
repress physical integrity rights (the rights to be free from political imprisonment, torture,
extrajudicial killing and disappearance) (Poe and Tate 1994; Poe et al. 1999; Davenport
and Armstrong 2004). There is far less consensus as to what exactly it is about
democracy that promotes greater respect for human rights. Recent studies indicate that
the relationship between democracy and human rights is not linear, meaning that there is
not an equivalent decrease in repression for every increase in democracy; rather there are
thresholds above which democracy is associated with a decrease in repression and below
which it is not (Davenport and Armstrong 2004). In fact, Fein (1995: 2) argued that
physical integrity rights are more likely to be violated “as democracy is extended before
it is fully institutionalized (More Murder in the Middle).” Bueno de Mesquita, Downs,
Smith and Cherif (2005) looked at the independent effects of components of democracy
and found that the most significant aspect of democracy in reducing repression was
political participation at its highest level, multi-party competition,
25
Building on this work, I add media freedom to the mix of democratic
characteristics and test the interactive effects of these variables on physical integrity
rights combined as well as individual types of physical integrity rights. First it is
important to consider in greater detail each of the six indicators of autocracy/democracy
and also found that
only consolidated democracy was associated with greater respect for human rights.
25
The measure used for political participation in the study by Bueno de Mesquita et al. (2005) comprises
two indicators: Regulation of participation and competitiveness of participation.
109
contained within the Polity IV dataset. Descriptions of each of these component
variables are provided in Chapter 3. The three executive recruitment variables measure
the regulation, competitiveness, and openness of executive recruitment. executive
constraints, provides information about the degree to which there are institutional checks
and balances on executive decisions. Gleditsch and Ward (1997) found that this aspect of
autocracy/democracy largely determines the composite Polity score of any given state.
Competitiveness of participation ranges from “repressed” to “competitive” and intuitively
seems in step with the other components which include categories that range from most
autocratic to most democratic.
26
(1) Unregulated: “Political participation is fluid; there are no enduring
national political organizations and no systematic regime controls on
political activity. Political groupings tend to form around particular
leaders, regional interests, religious or ethnic or clan groups, etc.; but the
number and relative importance of such groups in national political life
varies substantially over time”(Marshall and Jaggers 2009). No cases in
this study fall in this category. An example is Syria from 1942 to 1957.
The coding of regulation of participation is less intuitive
and requires further explanation. This variable measures the extent to which there are
rules that govern political participation (Marshall and Jaggers 2009). On the face of it,
this sounds simple; however, as the authors of the codebook admit, participation is
highly regulated in both one party states and consolidated democracies, “but they do so in
different ways, the former by channeling participation through a single party structure,
with sharp limits on diversity of opinion; the latter by allowing relatively stable and
enduring groups to compete nonviolently for political influence,” (Marshall and Jaggers
2002: 24). The coding for this dimension uses the following five-category scale:
26
As mentioned in Chapter 3, although the codebook specifically states that civil liberties are not included
in the coding, suppression of opposition media is referred to as one possible indicator that competitive of
participation should be coded as “suppressed”; therefore I exclude this variable from the primary analyses.
110
During this time Syria experienced a series of coups and there were no
enduring political coalitions (Collelo 1987)
(2) Multiple Identity: “There are relatively stable and enduring political
groups which compete for political influence at the national level–parties,
regional groups, or ethnic groups, not necessarily elected–but there are
few, recognized overlapping (common) interests” (Marshall and Jaggers
2009). An example is Argentina 1983-2007. Although political
participation in Argentina is somewhat competitive, the political arena is
defined by “Personality-based factions and social movements, rather than
political parties” (Marshall and Jaggers 2007).
(3) Sectarian: “Political demands are characterized by incompatible interests
an intransigent posturing among multiple identity groups and oscillate
more or less regularly between intense factionalism and government
favoritism, that is, when one identity group secures central power it favors
group members in central allocations and restricts competing groups'
political activities, until it is displaced in turn (i.e., active factionalism).
Also coded here are polities in which political groups are based on
restricted membership and significant portions of the population
historically have been excluded from access to positions of power (latent
factionalism, e.g., indigenous peoples in some South American countries)”
(Marshall and Jaggers 2009). An example is Turkey from 1993 to 2007
which has a political system with a “dual nature” where it is on the one
hand a constitutional democracy and on the other hand a factional system
dominated by the military (Marshall and Jaggers 2007). Although no
political parties are banned, members of parties that are out of favor with
the government face frequent legal penalties for their political activities
(Marshall and Jaggers 2007).
(4) Restricted: “Some organized political participation is permitted without
intense
factionalism but significant groups, issues, and/or types of conventional
participation are regularly excluded from the political process” (Marshall
and Jaggers 2009). An example is Burkina Faso from 1980-1996. After a
series of coups, Burkina Faso President Compaore introduced a multiparty
constitution and held elections in 1991. Opposition parties are permitted,
however, the fairness of these elections has been contested and Compaore
and his Congress for Democracy and Progress Party remain in power as of
this writing (Marshall and Jaggers 2007).
(5) Regulated: “Relatively stable and enduring political groups regularly
compete for political influence and positions with little use of coercion.
No significant groups, issues, or types of conventional political action are
regularly excluded from the political process” (Marshall and Jaggers
111
2009). An example is Australia 1980-2007. Australia is a constitutional
democracy with four durable parties where elections are “competitive and
free from coercion” (Marshall and Jaggers 2007).
Interestingly, regulation of participation does not always change in step with other
dimensions of autocracy/democracy. Consider, for example, how regulation of
participation has evolved in Mexico since the 1980’s. As Figure 14 depicts, in 1988,
regulation of participation changed from “sectarian” to “multiple identity” as the
Institutional Revolutionary Party’s (PRI) legitimacy was challenged by claims of
widespread election fraud. As political participation became less regulated and less
organized, executive power became more constrained (as reflected in executive
constraints in Figure 14). Interestingly, the Mexican media remained consistently in the
“partly free” category and regulation of participation first decreased and then remained
consistently low, as executive constraints increased and Mexico became increasingly
(though never fully) democratic.
112
Figure 14: Authority Patterns in Mexico 1980-2007
27
27
This figure was generated using the Regulation of Participation, Executive Constraints and a recode of
the Polity Score from the Polity IV dataset. These variables have been normalized (transformed to a 0 to 1
scale) for ease of interpretation.
.2 .4 .6 .8 1
1980 1990 2000 2010
year
Regulation of Participation Executive Constraints
Autocracy/Democracy
113
Figure 15: Authority Patterns in Uganda 1980-2007
Figure 16: Authority Patterns in the United States 1980-2007
0 .2 .4 .6 .8 1
1980 1990 2000 2010
year
Regulation of Participation Executive Constraints
Autocracy/Democracy
1 .2 .6 .8 .4 0
1980 1990 2000 2010
year
Regulation of Participation Executive Constraints
Autocracy/Democracy
114
In contrast, as depicted in Figure 15, Uganda experienced a reversal of the shift
that took place in Mexico. Following the military coup of 1985 which brought an end the
regime of President Milton Obote, the failure of the military government and the
subsequent rise to power of Yoweri Museveni, regulation of participation increased from
“sectarian” to “restricted.” This shift came about in part because President Museveni’s
“no-party” political system (in place from 1986 to 2005) restricted political activity to the
so-called “Movement” in an effort to prevent political divisions along ethnic and
religious lines (Marshall and Jaggers 2007). Thus, the rules of the political game were
clear, though decidedly undemocratic. In contrast, this same regime change led to a
dramatic reduction in executive constraints from an intermediate level to the lowest level
of “unlimited authority.” Executive constraints increased somewhat in 1993, but
limitations to the executive authority have remained “slight to moderate” since then.
During the same time period the Ugandan media remained in the “partly free” category.
Although the cases of Mexico and Uganda seem to indicate that regulation of
participation is negatively related to democracy, a third example illustrates that this is not
the case. The flat line across the top of Figure 16 shows the authority trends in the United
States. This figure illustrates that in a consolidated democracy, executive constraints,
regulation of participation, and autocracy/democracy are each at their highest levels.
Likewise, the media in the United States is consistently coded as “free.”
While we know that consolidated democracies are less likely to violate physical
integrity rights (Poe et al. 1999; Davenport and Armstrong 2004; Bueno de Mesquita et
al. 2005), we know very little about the effects of lower levels of these dimensions and
115
how they interact with civil liberties like media freedom to influence government
treatment of citizens. In the following section I present findings from a series of analyses
that collectively suggest that the effect of media freedom varies importantly depending on
the level of autocracy/democracy and also on the extent of regulation of political
participation. Moreover, this pattern is consistent over time and across different types of
physical integrity rights.
Findings Part 1: The Contemporaneous Effect of Media Freedom on Human Rights
The results of the statistical analyses focusing on the contemporaneous effect
media freedom on government respect for physical integrity rights support the hypothesis
that these effects are conditional, that they do in fact depend on the level of
autocracy/democracy. The results shown in Table 13 indicate that media freedom does
not have a significant independent effect on government respect for physical integrity
rights once I control for autocracy/democracy. However the addition of the interaction
between media freedom and the level of democracy/autocracy (shown in Table 14)
reveals that there is an effect of media freedom that varies depending on the level of
democracy/autocracy. These results are best interpreted graphically (Brambor, Clark, and
Golder 2006).
Table 13: Watchdog or Lapdog? The Effect of Media Freedom and Democracy on Physical Integrity Rights
1981-1995 N=1395
Model A Model B Model C Model D Model E Model F Model G
Lag Physical Integrity Rights .848***
(.034)
.832***
(.036)
.826***
(.037)
.787***
(.041)
.671***
(.048)
.680***
(.048)
---
Free Media .249**
(.089)
-.031
(.091)
-.013
(.09)
-.005
(.091)
-.026
(.088)
.041
(.088)
.169
(.172)
Autocracy/
Democracy
(five-part)
--- .126***
(.126)
.08*
(.033)
.085*
(.034)
.129***
(.033)
--- .351***
(.062)
Autocracy/
Democracy (Polity2)
--- --- --- --- --- .018**
(.007)
---
GDP (logged) --- --- .1**
(.038)
.138**
(.042)
.153***
(.04)
.188***
(.042)
.450***
(.089)
Population (logged) --- --- --- -.115***
(.028)
-.103***
.025)
-.102***
(.025)
-.399***
(.041)
Interstate Armed Conflict --- --- --- --- .128
(.164)
.147
(.167)
.047
(.22)
Internal Armed
Conflict
--- --- --- --- -1.086***
(.166)
-1.072***
(.167)
-1.856***
(.202)
Constant .637***
(.165)
.607***
(.156)
-.119
(.264)
.819**
(.345)
1.304***
(.349)
1.127**
(.352)
4.718***
(.735)
R
2
.76 .77 .77 .76 .77 .77 .38
*p<.05 ** p<.01 ***p<.001
116
Table 14: It Depends on Democracy: The Effect of the Interaction of Media Freedom and Democracy on
Physical Integrity Rights 1981-1995, N=1395
Model
1
Model
2
Model
3
Model
4
Model
5
Model
6
Model
7
Model
8
Lag Physical
Integrity Rights
.848***
(.034)
.832***
(.036)
.819***
(.038)
.816***
(.038)
.774***
(.043)
.650***
(.05)
.665***
(.049)
---
Free Media .249**
(.089)
-.031
(.091)
-.357*
(.176)
-.275
(.178)
-.316
(.181)
-.478**
(.178)
-.180
(.129)
-.828**
(.319)
Autocracy/
Democracy
(five-part)
--- .126***
(.033)
.079*
(.035)
.05**
(.356)
.05
(.036)
.08*
(.034)
--- .187**
(.067)
Autocracy/
Democracy
(Polity2)
--- --- --- --- --- --- .012
(.007)
---
Free Media
*Auto/Dem
-- -- .142*
(.064)
.113
(.065)
.135*
(.066)
.195**
(.066)
.041*
(.018)
.481***
(.121)
GDP (logged) --- --- --- .084*
(.038)
.12**
(.041)
.126**
(.039)
.166***
(.041)
.355***
(.092)
Population
(logged)
--- --- --- --- -.119***
(.028)
-.106***
(.026)
-.101***
(.026)
-.389***
(.04)
Interstate Armed
Conflict
--- --- --- --- --- .097
(.167)
.12
(.171)
.022
(.219)
Internal Armed
Conflict
--- --- --- --- --- -1.122***
(.167)
-
1.107***
(.168)
-
1.869***
(.195)
Constant .637***
(.165)
.607***
(.156)
. 705***
(.176)
.077
(.297)
1.084**
(.399)
1.69***
(.402)
1.352***
(.377)
5.471***
(.733)
R
2
.76 .77 .76 .76 .76 .77 .77 .40
*p<.05 ** p<.01 ***p<.001
117
118
Figure 17 shows the marginal effects of media freedom on human rights across a
range of regimes. As regime type changes from most autocratic (0) to most democratic
(4), the influence of media freedom changes from negative and statistically significant for
the most autocratic regimes to insignificant for regimes in the middle to positive and
statistically significant for only the most democratic regimes. This supports the
theoretical arguments presented in Chapter 3 that the effect of media freedom varies
depending on the presence (or absence) of democratic characteristics. Regardless of the
value of media freedom, the impact of a one unit increase in the five-part
autocracy/democracy score on government respect for physical integrity rights is positive
and statistically significant.
In order to see how these two variables work together to influence a country’s score on
the CIRI Physical Integrity Rights Index, I used Clarify Software (Tomz, Wittenberg, and
King 2003; King, Tomz, and Wittenberg 2000). Figure 18 shows that free media is
predicted to have a detrimental effect on human rights in extreme autocracies and a
positive effect on human rights in consolidated democracies. This figure (from Model 6
in Table 14) depicts the predicted values of government respect for physical integrity
rights from a simulation in which media freedom and regime type are varied while all
other variables are held constant at their mean or modal values. Figure 18 illustrates that
in an autocracy, having media that serve as a watchdog over government behavior could
actually result in worse government behavior; whereas media that are government-
controlled and play more of a lapdog role will have little effect on government respect for
human rights. As we shall see later in this chapter the case of Mexico in the 1980s and
119
early 1990s exemplifies this effect. The country was still far from a consolidated
democracy and yet the media was mostly free. As the media began to report more
critically on government activity in the early 1990s, government respect for physical
integrity rights remained quite low and decreased substantial between 1993 and 1994
(Cingranelli and Richards 2006).
Figure 17: The Marginal Effect of Media Freedom across Autocracy/Democracy
-1 -.5 0 .5
Marginal Effect of Media Freedom
0 1 2 3 4
Regime Characteristics
Marginal Effect of Media Freedom
95% Confidence Interval
Dependent Variable: Government Respect for Physical Integrity Rights
120
Figure 18: The Predicted Effect of the Interaction between Regime Characteristics
and Media Freedom on Government Respect for Physical Integrity Rights
28
This finding supports the hypothesis that the potential for free news media to have a
positive effect on government respect for human rights depends on the presence of
democratic institutions (H
3
). Thus, the impact of media freedom on government respect
for physical integrity rights appears more complicated than the simple positive or
negative relationships proposed in H
1
. In fact, the findings here do not support this
hypothesis. Instead, consistent with the cycle depicted in Figure 6 in Chapter 3, media
freedom appears to have a negative impact on government respect for human rights in
autocracies (states with a Polity score between -10 and 0 or categories 0 and 1 of the five-
28
This graph was produced using the Clarify software (King, Tomz and Wittenberg 2000; Tomz,
Wittenberg and King 2003).Thanks to Laron Williams for providing the code for adapting Clarify for the
panel corrected standard errors model.
4 4.5 5 5.5 6
Predicted Level of Physical Integrity Rights
0 1 2 3 4
Autocracy/Democracy
95% confidence interval
free media
95% confidence interval
government-controlled media
121
part autocracy/democracy measure). An example is Croatia from 1993-1995 which was
quite autocratic (with a Polity score of -3 to -5), but did have independent media (Van
Belle 2000; Marshall et al. 2003). During this time government respect for physical
integrity rights in Croatia ranged from 2 to 4 (with 8 being full respect for these rights)
(Cingranelli and Richards 2006). In countries that have some democratic institutions,
those with a combined Polity score of 1 to 7 (category 2 of the five-part
autocracy/democracy measure), the effect of free media does not appear to be different
from that of government-controlled media. Only in states that are fully democratized,
those with Polity scores of 8 and above (categories 3 and 4 of the five-part measure),
does the effect of free media appear to have a statistically significant positive effect on
government respect for human rights and the effect is more pronounced in states like
Costa Rica and the Netherlands which during the period covered in this study had free
media and the highest Polity score of 10 (category 4 of the five part measure) (Van Belle
2000; Marshall et al. 2003). During most of this time Costa Rica had full respect for
physical integrity rights; the Netherlands had full respect during the entire time span
(Cingranelli and Richards 2006).
As expected, the level of development (GDP per capita) has a statistically
significant and positive effect and both population and internal armed conflict have a
statistically significant and negative effect on government respect for physical integrity
rights. Interestingly, international armed conflict did not have a significant effect on
human rights. This finding is consistent across all models, suggesting that international
conflict has little influence on human rights compared to domestic conflict. It does make
122
sense intuitively that conflict within a state would influence how the government treats
citizens, as opposed to conflict with other states.
Findings Part 2: The Effects of Media Freedom on Human Rights Five Years Later
The next step in my analysis was to test whether or not media freedom has a
different effect over time. As mentioned earlier, Bueno de Mesquita et al. (2005)
theorized that democratic institutions might influence subsequent government repression
of or respect for human rights and posited that it would take about five years for
government to respond to any institutional shifts. Other research has found that it
generally takes about five years for a change in institutions to make a difference in the
economy, so it seemed reasonable that this same time might have a corresponding effect
on human rights (Bueno de Mesquita, Smith, Siverson and Morrow 2003). Therefore in
this next series of analyses, I tested the effect of media freedom on government respect
for physical integrity rights five years later. These results (shown Model 1 in Table 15)
indicate that five years later, media freedom has an independent statistically significant
and negative effect on government respect for physical integrity rights. However, as
shown in Model 2 of Table 15, the model with the interactive term is a better fit,
suggesting that the effect of media freedom is dependent on the presence (or absence) of
democratic characteristics. The extent of this variation is depicted in Figure 19 which
illustrates that five years later media freedom only has a statistically significant and
positive effect in the most consolidated democracies and that it has a statistically
significant and negative effect in countries in the middle and lower levels of the adjusted
123
Polity autocracy/democracy score.
29
Figure 19: The Marginal Effect of Media Freedom on Human Rights Five Years
Later
The results for the control variables were consistent
with the findings in the contemporaneous analyses.
29
Unlike the analyses looking at the contemporaneous effects, this analysis was more sensitive to different
specifications. I did not get significant results using an ordered logistical analysis; however panel corrected
standard errors seems to be the more appropriate method given the dependent variable. As more data
become available it will be important to consider these effects over a longer time span.
-2 -1 0 1
Marginal Effect of Media Freedom
0 1 2 3 4
Autocracy/Democracy
Marginal Effect of Media Freedom
95% Confidence Interval
Dependent Variable: Physical Integrity Rights Index (+5 years)
124
Table 15: It Still Depends on Democracy--The Effect of Media Freedom on Physical
Integrity Rights 5 Years Later
Dependent Variable:
Physical Integrity Index (+ 5 years)
Model 1 Model 2
Physical Integrity Rights Index (current) .290 (.064)*** .273 (.062)***
Media Freedom -.408 (.139)** -1.621 (.279)***
Autocracy-Democracy .358 (.053)*** .225 (.057)***
Media Freedom*Autocracy-Democracy --- .511 (.092)***
GDP/capita (logged) .453 (.076)*** .340 (.068)***
Population (logged) -.318 (.046)*** -.335 (.048)***
Interstate Conflict -.102 (.162) -.092 (.162)
Intrastate Conflict -.496 (.155)** -.514 (.151)**
Constant 2.394 (.558)*** 3.621 (.589)***
R
2
.78 .80
*p<.05 ** p<.01 ***p<.001
Findings Part Three: Pulling Apart Democracy
Thus far, the statistical analyses indicate that media freedom does have an effect
on government respect for human rights, but that the effect varies depending on the level
of autocracy/democracy. The next step in this study breaks down democracy into its
components. The Polity IV dataset provides measures for the indicators used to compose
the Polity score.
30
These include executive recruitment, executive constraints, regulation
of participation and competitiveness of participation.
31
30
Detailed definitions of the Polity IV variables can be found in Chapter 3.
The results in Model 1 of Table
16 show that when all the components of democracy are included, only regulation of
participation has a significant effect on government respect for physical integrity rights;
however, diagnostic tests revealed that the democracy variables are highly correlated,
31
Because the regulation of chief executive recruitment, competitiveness of executive recruitment and
openness of executive recruitment variables are highly correlated, I used the executive recruitment concept
variable, which comprises the three recruitment indicators.
125
thereby making it difficult to obtain statistically significant results.
32
32
Details of these and other diagnostics can be found in the technical report at the end of this chapter.
Although these
problems with multicollinearity persisted in the analyses looking at human rights five
years later, Model 1 in Table 17 shows that media freedom had a statistically significant
and negative effect and all of the indicator variables except for competitiveness of
participation had a statistically significant and positive effect on government respect for
physical integrity rights five years later. To address these multicollinearity issues, I
dropped the competitiveness of participation variable, in part because of its lack of
significance and also because of concerns noted in Chapter 3 that there might now be
some consideration of opposition media in the coding of this dimension. In addition, I
dropped the executive recruitment variable because it was highly correlated with
executive constraints and previous research has documented the theoretical importance of
the executive constraints variable in determining the Polity score (Gleditsch and Ward
1997). Once these variables were removed, both executive constraints and regulation of
participation have statistically significant and positive effects on government respect for
physical integrity rights in the contemporaneous as well as the five years later model.
Interestingly, in the five-years-later model (Model 2 in Table 17) the effect of media
freedom becomes insignificant.
Table 16: What is it About Democracy? The Effect of Media Freedom and Components of Democracy on
Human Rights N=1036
Dependent Variable:
Physical Integrity Index (current)
Model 1 Model 2 Model 3 Model 4
Physical Integrity Rights Index (-1 year) .638 (.052)*** .651 (.051)*** .651 (.051)*** .638 (.052)***
Media Freedom -.025 (.081) .009 (.082) -.041 (.229) -.359 (.150)*
Executive Recruitment .253 (.198) --- --- ---
Executive Constraints .267 (.243) .581 (.171)** .566 (.186)** .506 (.169)**
Media Freedom*Executive Constraints --- --- .068 (.279) ---
Regulation of Participation .754 (.159)*** .748 (.152)*** .749 (.158)*** .274 (.199)
Media Freedom*Regulation of
Participation
--- --- --- .673 (.233)**
Competitiveness of Participation .357 (.244) --- --- ---
GDP/capita (logged) .054 (.041) .083 (.040)* .080 (.039)* .077 (.040)
Population (logged) -.115
(.026)***
-.117 (.026)*** -.117 (.027)*** -.123 (.027)***
Interstate Conflict -.034 (.139) -.026 (.140) -.032 (.142) -.021 (.136)
Intrastate Conflict -.863
(.147)***
-.852 (.147)*** -.855 (.149)*** -.851 (.147)***
Constant 1.576
(.356)***
1.493 (.357)*** 1.514 (.359)*** 1.981 (.433)***
R
2
.89 .89 .89 .89
*p<.05 ** p<.01 ***p<.001
126
Table 17: What is it About Democracy over Time? The Effect of Media Freedom and Components of
Democracy on Human Rights Five Years Later N=1110
Ordinary Least Squares Analyses using Panel Corrected Standard Errors *p<.05 ** p<.01 ***p<.001
Dependent Variable:
Physical Integrity Index (+5 years)
Model 1 Model 2 Model 3 Model 4
Physical Integrity Rights Index (current) .222 (.063)*** .231 (.062)*** .231 (.062)*** .218 (.062)***
Media Freedom -.289 (.134)* -.233 (.135) -.217 (.389) -1.072 (.253)***
Executive Recruitment .783 (.320)* --- --- ---
Executive Constraints .795 (.387)* 1.459 (.222)*** 1.469 (.240)*** 1.344 (.215)***
Media Freedom*Executive Constraints --- --- -.019 (.458) ---
Regulation of Participation 2.261
(.227)***
2.222 (.222)*** 2.222 (.228)*** 1.303 (.301)***
Media Freedom*Regulation of
Participation
--- --- --- 1.352 (.340)***
Competitiveness of Participation .344 (.440) --- --- ---
GDP/capita (logged) .198 (.056)*** .240 (.056)*** .238 (.057)*** .219 (.055)***
Population (logged) -.383
(.047)***
-.399 (.047)*** -.399 (.047)*** -.401 (.045)***
Interstate Conflict -.097 (.147) -.092 (.148) -.089 (.149) -.087 (.148)
Intrastate Conflict -.468 (.147)** -.469 (.148)** -.461 (.150)** -.476 (.146)**
Constant 3.299
(.539)***
3.421 (.555)*** 3.432 (.559)*** 4.312 (.648)***
R
2
.80 .80 .81 .81
127
128
To pinpoint the aspect of democracy that interacts with media freedom to
influence government respect for human rights, I ran a series of analyses with different
interactive terms. First I looked the interaction between executive constraints and media
freedom. The third models in both Tables 4 and 5 show that this interaction is
insignificant in both cases. The interaction between regulation of participation and media
freedom is included in the fourth models of both Tables 4 and 5. In both cases the
interaction is significant. Figure 20 depicts the contemporaneous effect of media freedom
across a range of regulation of participation. It shows that media Freedom has a
statistically significant and positive effect only when regulation of participation
approaches its highest level. The developments in Uruguay in the 1980s illustrate how
this might work. Although the media became mostly free in 1980, regulation of
participation was “restricted” (at level 4 of 5) in Uruguay during the first half of the
1980s. During this same time there was quite a bit of protest and government respect for
physical integrity rights was low (ranging from 4 to 5, with 8 being the highest possible
score). In 1985 when the military junta relinquished control of the country to newly
elected President Sanguinetti, the rules of participation were clearly delineated and
Uruguay moved into the “regulated” or highest level of regulation of participation. At the
same time, government respect for human rights improved dramatically in 1985 (rising to
7 and then in 1986 to 8), as the new government enjoyed popular support and faced little
or no protest (for more on the events in Uruguay, see Chapter 4).
The effect of media freedom five years later (depicted in Figure 21) is statistically
significant and negative as regulation of participation drops below its mid-range. Both
129
figures show the same trajectory of a negative effect at the lower levels of regulation of
participation and a positive effect in the upper levels. In summary, the findings here
indicate that while media freedom does have an effect on government respect for physical
integrity rights, this effect varies, depending on democracy. Furthermore, the aspect of
democracy that matters most in this interaction with media freedom is regulation of
participation. The next section presents findings that look at the effect of media freedom,
democracy and components of democracy on different types of physical integrity rights.
Figure 20: The Marginal Effect of Media Freedom on Human Rights
(Depending on Regulation of Participation)
-.4 -.2 0 .2 .4 .6
Marginal Effect of Media Freedom
.2 .4 .6 .8 1
Regulation of Participation
Marginal Effect of Media Freedom
95% Confidence Interval
Dependent Variable: Physical Integrity Rights Index
130
Figure 21: The Marginal Effect of Media Freedom on Human Rights Five Years
Later (Depending on Regulation of Participation)
Findings Part 4: Pulling Apart Physical Integrity Rights
From the analyses in the previous section, it is clear that media freedom and
different components of autocracy/democracy have different effects on physical integrity
rights. In this section I test whether or not these findings vary across types of physical
integrity rights. The CIRI physical integrity index comprises indicators of disappearance,
political imprisonment, killing and torture. In this series of analyses I consider each of
these in turn.
The analyses focusing on the influence on disappearance rights (shown in Tables
6 and 7) indicate that the effect of media freedom is for the most part negative and that
-1 -.5 0 .5 1
Marginal Effect of Media Freedom
.2 .4 .6 .8 1
Regulation of Participation
Marginal Effect of Media Freedom
95% Confidence Interval
Dependent Variable: Physical Integrity Rights Index (+5 years)
131
the effect depends on democracy as a whole rather than its components. Specifically, the
first models in both Tables 6 and 7 show that neither media freedom nor
autocracy/democracy have an independent effect, but the second models show that the
interaction between media freedom and autocracy/democracy does have a significant
effect both on contemporaneous disappearance rights and disappearance rights five years
later. As mentioned earlier, the results of interactions are best interpreted visually.
Figures 22 and 23 depict the marginal effects of media freedom as autocracy/democracy
changes. Figure 22 shows that states below the mid-range of autocracy/democracy with
media freedom are predicted to have a higher probability of having no government
respect for disappearance rights than those states with controlled media. Figure 23
illustrates that autocratic states with media freedom have a lower predicted probability of
having complete respect for disappearance rights than those with controlled media. A
case in point, the Philippines in 1984 was decidedly autocratic, yet the media was
functionally free (Van Belle 2000; Marshall et al. 2003). During this year, more than 50
people “disappeared” (Cingranelli and Richards 2006). Media freedom is not predicted to
have an effect that is statistically different from that of controlled media in states in the
middle to upper ranges of democracy.
33
The third models in both the contemporaneous and the five-years-later analyses
(Models 3 in Tables 18 and 19) show that neither media freedom nor executive
constraints have significant effects, but regulation of participation does have a
statistically significant and positive effect in both sets of analyses. Interestingly, in the
contemporaneous analyses, neither GDP/capita nor population had significant effects, but
33
The results (and figure) are substantively the same for the five-years-later model.
132
internal armed conflict consistently had a statistically significant and negative effect on
disappearance rights. In the five-years later analyses, as expected, both population and
internal armed conflict had statistically significant and negative effects; GDP/capita had a
statistically significant and positive effect only in Model 1 of Table 19.
Table 18: The Effect of Media Freedom and Components of Democracy on Disappearance N=1036
Dependent Variable:
Disappearance Rights (current)
Model 1
Coefficient
(Standard Error)
Model 2
Coefficient
(Standard Error)
Model 3
Coefficient
(Standard Error)
Model 4
Coefficient
(Standard Error)
Disappearance Rights (-1 year) 1.648 (.168)*** 1.583 (.158)*** 1.570 (.172)*** 1.566 (.172)***
Media Freedom -.430 (.367) -1.451 (.445)** -.393 (.290) -.646 (.496)
Autocracy/Democracy .159 (.138) -.063 (.107) --- ---
Media
Freedom*Autocracy/Democracy
--- .565 (.183)** --- ---
Executive Constraints --- --- .889 (.488) .803 (.491)
Regulation of Participation --- --- 1.368 (.523)** 1.014 (.627)
Media Freedom*Reg. of Participation --- --- --- .535 (.842)
GDP/capita (logged) .276 (.141) .201 (.127) .176 (.130) .161 (.133)
Population (logged) -.141 (.081) -.134 (.082) -.162 (.093) -.161 (.093)
Interstate Conflict .261 (.484) .147 (.484) .342 (.477) .345 (.478)
Intrastate Conflict -1.496 (.222)*** -1.519 (.214)*** -1.420 (.216)*** -1.409 (.214)***
Wald χ
2
155.62 158.96 153.73 158.29
McKelvey & Zavoina’s R
2
.39 .41 .42 .42
Ordered Logit Analyses *p<.05 ** p<.01 ***p<.001
133
Table 19: The Effect of Media Freedom and Components of Democracy on Disappearance Rights Five Years Later
N=1110
Dependent Variable:
Disappearance Rights (+5 years)
Model 1
Coefficient
(Standard Error)
Model 2
Coefficient
(Standard Error)
Model 3
Coefficient
(Standard Error)
Model 4
Coefficient
(Standard Error)
Disappearance Rights (current) 1.006 (.169)*** .925 (.158)*** .913 (.165)*** .912 (.169)***
Media Freedom -.666 (.459) -1.664 (.581)** -.499 (.363) -.565 (.652)
Autocracy/Democracy .202 (.187) -.043 (.166) --- ---
Media
Freedom*Autocracy/Democracy
--- .583 (.232)* --- ---
Executive Constraints --- --- .959 (.677) .938 (.633)
Regulation of Participation --- --- 1.927 (.672)** 1.836 (.672)**
Media Freedom*Reg. of Participation --- --- --- .139 (1.055)
GDP/capita (logged) .380 (.187)* .301 (.174) .251 (.168) .247 (.175)
Population (logged) -.257 (.087)** -.264 (.088)** -.297 (.094)** -.296 (.094)**
Interstate Conflict -.146 (.404) -.220 (.376) -.016 (.425) -.016 (.426)
Intrastate Conflict -1.268 (.319)*** -1.289 (.293)*** -1.148 (.272)*** -1.145 (.275)***
Wald χ
2
129.25 142.11 133.60 133.59
McKelvey & Zavoina’s R
2
.32 .35 .38 .38
Ordered Logit Analyses *p<.05 ** p<.01 ***p<.001
134
135
Figure 22: The Effect of Media Freedom on Disappearance Rights (Low)
Figure 23: The Effect of Media Freedom on Disappearance Rights (High)
0 .05 .1 .15
Pr. No Respect for Disappearance Rights
0 1 2 3 4
Autocracy/Democracy
Free Media Controlled Media
.4 .6 .8 1
Pr. Complete Respect for Disappearance Rights
0 1 2 3 4
Autocracy/Democracy
Free Media Controlled Media
136
The results of the analyses focusing on the right to be free from political killing
are mixed. In the contemporaneous analyses, neither media freedom, nor
autocracy/democracy, nor their interaction, have a significant effect (shown in Models 1
and 2 of Table 20). Yet in the analysis with the components of democracy, regulation of
participation has a statistically significant and positive effect (see Model 3 in Table 20).
Furthermore when the interaction between regulation of participation and media freedom
is included (Model 4 in Table 20), media freedom has a statistically significant and
negative effect, but only where regulation of participation is at its lowest level. Figure 24
shows that states with media freedom that lack regulation of participation have a higher
predicted probability of having frequent political killing than those states with controlled
media. An interesting case that exemplifies this effect is Columbia, which from 1981 to
1995 had functionally free media and was highly democratic (with a Polity score of 8, 10
being the highest), but had comparatively low regulation of participation (2 meaning
“multiple identity”), reflecting the country’s “intense and often violent factionalism”
(Van Belle 2000; Marshall and Jaggers 2007). During the same time period, there were
50 or more political killings each year (Cingranelli and Richards 2006). Similarly, in the
five-years-later model, neither media freedom nor autocracy/democracy have an
independent effect, but media freedom does have a statistically significant and negative
effect at mid to lower levels of autocracy/democracy (see Models 1 and 2 in Table 21
and Figure 25). In summary, in the analyses looking at the effects of the components of
democracy on political killing, only regulation of participation has a statistically
significant and positive effect (see Model 3 in Table 21). In all of these five-years-later
137
analyses, the control variables have statistically significant effects in the expected
directions with the exception of international armed conflict which remains insignificant
across all models.
Table 20: The Effect of Media Freedom and Components of Democracy on the Right to be Free from
Political Killing N=1036
Dependent Variable:
Kill (current)
Model 1
Coefficient
(Standard Error)
Model 2
Coefficient
(Standard Error)
Model 3
Coefficient
(Standard Error)
Model 4
Coefficient
(Standard Error)
Kill (-1 year) 1.781 (.160)*** 1.733 (.164)*** 1.654 (.166)*** 1.633 (.172)***
Media Freedom -.374 (.338) -1.126 (.615) -.344 (.302) -1.055 (.418)*
Autocracy/Democracy .098 (.121) -.038 (.103) --- ---
Media Freedom*Autocracy/Democracy --- .385 (.200) --- ---
Executive Constraints --- --- .711 (.411) .510 (.398)
Regulation of Participation --- --- 1.668 (.374)*** .737 (.494)
Media Freedom*Reg. of Participation --- --- --- 1.413 (.745)
GDP/capita (logged) .552 (.137)*** .494 (.137)*** .399 (.123)** .371 (.127)**
Population (logged) -.154 (.060)* -.154 (.059)** -.201 (.064)** -.204 (.064)**
Interstate Conflict .669 (.387) .597 (.386) .765 (.418) .781 (.404)
Intrastate Conflict -1.484 (.262)*** -1.497 (.259)*** -1.420 (.246)*** -1.401 (.245)***
Wald χ
2
294.74 311.91 321.87 319.38
McKelvey & Zavoina’s R
2
.54 .54 .57 .57
Ordered Logit Analyses *p<.05 ** p<.01 ***p<.001
138
Table 21: The Effect of Media Freedom and Components of Democracy on the Right to be Free from
Political Killing Five Years Later N=1110
Dependent Variable:
Kill (+5 years)
Model 1
Coefficient
(Standard
Error)
Model 2
Coefficient
(Standard
Error)
Model 3
Coefficient
(Standard
Error)
Model 4
Coefficient
(Standard
Error)
Kill (current) .952 (.170)*** .902 (.166)*** .792 (.172)*** .784 (.182)***
Media Freedom -.680 (.403) -1.665 (.779)* -.552 (.340) -.790 (.507)
Autocracy/Democracy .174 (.138) -.004 (.134) --- ---
Media
Freedom*Autocracy/Democracy
--- .498 (.263) --- ---
Executive Constraints --- --- .834 (.517) .758 (.496)
Regulation of Participation --- --- 2.185 (.559)*** 1.863 (.596)**
Media Freedom*Reg. of Participation --- --- --- .476 (.814)
GDP/capita (logged) .795 (.155)*** .723 (.154)*** .647 (.150)*** .639 (.151)***
Population (logged) -.242 (.064)*** -.242 (.063)*** -.320 (.073)*** -.321
(.073)***
Interstate Conflict .143 (.463) .075 (.447) .287 (.415) .292 (.417)
Intrastate Conflict -1.042 (.365)** -1.045 (.342)** -.941 (.295)** -.939 (.294)**
Wald χ
2
279.79 302.30 360.50 358.74
McKelvey & Zavoina’s R
2
.44 .45 .50 .50
Ordered Logit Analyses *p<.05 ** p<.01 ***p<.001
139
140
Figure 24: The Effect of Media Freedom on Probability of Frequent Political Killing
0 .05 .1 .15 .2
Pr. Frequent Political Killing
0 .2 .4 .6 .8 1
Regulation of Participation
Media Freedom Controlled Media
141
Figure 25: The Effect of Media Freedom on the Predicted Probability of Frequent
Political Killing Five Years Later
The results for torture are similar to those for the disappearance and political
killings variables. While media freedom does not have an independent effect in either of
the analyses, the interactions of media freedom with democracy/autocracy and media
freedom with regulation of participation are statistically significant in both the
contemporaneous and the five-years-later analyses (these results can be seen in Tables D
and E in the Appendix). Specifically, media freedom with low regulation of participation
is associated with an increased predicted probability of torture in both the
contemporaneous and five years later models (Figures 40 and 41 in the Appendix depict
these effects). Likewise, the analyses focusing on political imprisonment show that while
0 .2 .4 .6
Pr. Frequent Political Killings
0 1 2 3 4
Autocracy/Democracy
Free Media Controlled Media
142
media freedom does not have an independent effect, it does in certain cases have an
interactive effect (these results can be seen in Tables F and G in the Appendix).
34
In summary, while the significance varies minimally for each individual physical
integrity right, the trajectory of the effect of media freedom as regimes shift from extreme
autocracy to consolidated democracy remains reasonably consistent across all models.
Media freedom tends to have a negative effect or no effect at all in autocracies and states
that fall somewhere in the middle range of autocracy/democracy. Media freedom only
has a positive effect in the most consolidated democracies. Similarly, media freedom
consistently has a negative effect or no effect when participation is unregulated or
unorganized and only has a positive effect when participation is highly regulated and
organized. In the next section, I consider the implications of these findings in the context
of several cases, in particular Mexico and Uganda.
The Varying Effects of Free Media
Clearly, the presence or lack of media freedom has a different impact on
government respect for human rights in democracies than in autocracies. Likewise the
effect of media freedom changes, depending on the regulation of participation. Certainly,
some of the findings here are counterintuitive. It is expected that free media in a
democracy would serve to promote higher government respect for physical integrity
rights, and indeed, this is usually the case. For example, both the Netherlands and New
34
Because the analyses involving political imprisonment violated the parallel regression assumption,
ordered logit was not appropriate; therefore I used multinomial logit for these analyses. For more see the
technical report.
143
Zealand had free media, the highest possible Polity democracy score of 10 (including the
highest level of 5 for Regulation of Participation) and complete government respect for
physical integrity rights (the highest possible score of 8 on the CIRI Physical Integrity
Rights Scale) during the time span of this study.
It is somewhat surprising that free media in an autocracy would be associated
with lower respect for physical integrity rights. Still, it is important to consider that
leaders in autocratic regimes are less vulnerable to criticism than their counterparts in
democratic regimes because they do not have to worry about being voted out of office.
There is little or no incentive for elites in these countries to co-opt those who are
marginalized. It follows too, that media publicity of human rights violations in
autocracies might serve to keep citizens fearful and submissive, or might prompt protests
which in turn could lead to further repression. Additionally, in states with little or no
regulation of participation, there is more potential for competing political powers to
intervene or for social movements to gain momentum and highjack the political process.
This potential becomes greater when the media is free and can be used to mobilize public
support. It follows too that an autocratic regime would be threatened if it perceived (or
learned from media coverage) that a competing power was taking advantage of the lack
of regulation of participation and using the media to reach the public. In this scenario,
repression is likely because, as previous research has found, repression is more likely
when government is threatened (Poe and Tate 1994; Poe et al. 1999).
Two cases in point are Mexico and Uganda. Both fell into the category of
autocracies with free media for the years 1981 to 1995 and both had scores on the lower
144
end of the CIRI Physical Integrity Rights Scale for the same time period (Mexico’s scores
range from 0 to 4, with an average score of 2.2; Uganda’s CIRI scores range from 0 to 5,
with an average score of 2.2). These cases provide interesting views of the effect of
media that are free, or at least somewhat free, on an autocratic regime’s respect for
human rights. In each case both the media and the regime were undergoing transitions.
Part of the regime transitions are reflected in the changes in regulation of participation.
As mentioned earlier, during this time Mexico’s regulation of Participation score
decreased from a level 3 (“sectarian”) to a level 2 (“multiple identity), Whereas Uganda’s
increased from a level 3 (“sectarian”) to a level 4 (“restricted”). Figures 26 and 27 depict
the changes in government respect for physical integrity rights, regulation of
participation, autocracy/democracy and media freedom for Uganda and Mexico,
respectively.
35
In the 1990s the Mexican media, led by newspapers, became increasingly
independent and critical of the government. In particular the media began to aggressively
expose government abuse (Lawson 2002). One of the most dramatic was the 1995
massacre at Aguas Blancas when members of the state judicial police gunned down 17
Figure 26 shows that in Uganda in 1986 as regulation of participation
increased and media remained free, government respect for human rights improved, even
as Uganda became less democratic. In contrast, Figure 27 shows that in Mexico in 1988 a
decline in regulation of participation (while media remained free) was followed one year
later by a decline in government respect for human rights even as overall Mexico became
more democratic.
35
These figures were generated using the CIRI Physical Integrity Index, the Polity IV dataset’s Polity2 and
Regulation of Participation scores, the Global Press Freedom dataset. These scores have each been
transformed to a 0 to 1 scale for ease of interpretation.
145
unarmed peasants (Amnesty International Annual Report 1996; Lawson 2002). “Many of
Mexico’s media pursued the affair with assertiveness and vigor with newspapers like La
Jornada and Reforma giving saturation coverage to the massacre and subsequent cover-
up,” (Lawson 2002, 150). Yet, government violations of human rights persisted. News of
the attacks fueled public outrage. At a service commemorating the one year anniversary
of the killings, a group of armed rebels appeared and vowed to overthrow the
“illegitimate government,” (Fineman 1996). This threat prompted the government to send
in the Mexican army to establish a counterinsurgency base in Aguas Blancas, and while
many residents reportedly welcomed the security provided by the army, peasant-activists
claimed that the army served as a cover for state police to illegally detain and torture
activists (Fineman 1996). In addition, as the media became more aggressive in reporting
government abuses, journalists and human rights defenders were targeted (Amnesty
International Annual Report 1996). According to the Committee to Protect Journalists
(Simon 1997), in general as the Mexican media became more independent and
professional journalistic norms evolved, journalism became a more hazardous occupation
as “powerful figures, unaccustomed to public scrutiny, have lashed out violently.” In the
Mexican case, there is some indication that in the long run, media freedom can have a
positive impact. Lawson (2002) attributed persistent coverage of government scandals as
ultimately having a “delegitimizing effect” on Mexican’s one-party regime which
ultimately facilitated democratization. He found that the media was able to accomplish
this in spite of government efforts to restrict press freedom. “Independent publications
were typically met with official resistance and repression, establishing themselves only
146
after repeated struggles against the government,” (Lawson 2002, 183). It could be too that
the decrease in regulation of participation combined with the increasingly independent
media made it easier for the Mexican people to break the PRI’s hold on political power.
In the short term however, the Mexican case suggests that the effect of this combined
threat led to persistent government violations of physical integrity rights.
Figure 26: Physical Integrity Rights, Media Freedom, Autocracy/Democracy and
Regulation of Political Participation in Uganda, 1981-1995
0 .2 .4 .6 .8 1
1980 1985 1990 1995
year
Physical Integrity Index Autocracy/Democracy
Regulation of Participation
Media Freedom
147
Figure 27: Physical Integrity Rights, Media Freedom, Autocracy/Democracy and
Regulation of Political Participation in Mexico, 1981-1995
In Uganda, after the fall of Idi Amin’s regime in 1979, the independent press
returned, but according to Ocitti (2005: 73), most of the newspapers were “outrageously
partisan” to the degree that “it spawned a near media war especially between those
supporting the government on the one hand and those aligned to the various opposing
political groups, on the other.” In this environment, Ugandan journalists sometimes
fabricated stories or failed to verify information from their sources (Ocitti 2005). The
government responded by arresting many journalists and expelling some foreign reporters
(Ocitti 2005). As the media war continued, opposition groups continued to fight against
the government and human rights organizations joined the media in accusing the
government of “massive human rights violations,” (Ocitti 2005: 86). When Yoweri
0 .2 .4 .6 .8 1
1980 1985 1990 1995
year
Physical Integrity Index Autocracy/Democracy
Regulation of Participation
Media Freedom
148
Moseveni’s National Resistance Army seized control of Uganda in 1986, the press was
mostly supportive of the new government (Ocitti 2005). However, when the press began
to report on human rights violations and rebellion in Northern Uganda, Moseveni’s
government responded by clamping down on journalists (Ocitti 2005). Yet, the news
media remained “vocal and independent” in spite of the government’s persistence in
arresting and charging journalists who publish negative information (Hostile to
Democracy: the Movement System and Political Repression in Uganda 1999). At the
same time, human rights organizations reported that both the government and armed
opposition groups were responsible for violating physical integrity rights, especially in
Northern Uganda (Amnesty International Annual Report 1995; Hostile to Democracy: the
Movement System and Political Repression in Uganda 1999). In 1995, the Press and
Journalists Law went into effect. Among other restrictions, this measure required that
journalists be registered and licensed. Yet this new law appeared to have little impact on
journalistic practice because most independent news organizations simply ignored it,
(Ocitti 2005). In spite of the continued challenges from opposition media, Moseveni’s
government has remained firmly in control. One reason for this could be the “no-party”
system which restricted political participation.
36
36
The “no-party” system was rejected in 2005, but the effects of this change are not discussed here because
this study focuses on the years 1981 to 1995.
Although Uganda experienced a decline
in democracy in 1985, government respect for physical integrity rights continued to
improve in the late 1980s and early 1990s, though it remained fairly low through the
1990s. In this scenario, the higher regulation of participation may have mitigated the
149
threats to the government from independent media and rebel groups and somewhat
reduced government repression of physical integrity rights.
In both the Mexican and the Ugandan cases, the media acted independently in
spite of government efforts to intimidate reporters. In each case, the professional
environment of the media played a key role in maintaining the media’s watchdog role. In
the case of Uganda, in the years immediately following Idi Amin’s regime, it was the
political environment that fueled the partisan nature of the media that prompted critical
coverage of the government. In Mexico, the economic environment, in particular the
commercial success of covering scandals, reinforced the emerging professionalism in the
news media. In both cases, initial government response to the coverage of human rights
violations was repressive, and often journalists were the targets of the repression. In the
long term though, at least in Mexico, the result may be more positive with independent
media facilitating the transition to a less repressive and more democratic regime; in
Uganda the struggle between the media and the government persists and human rights
violations continue (Human Rights Watch World Report 2007: Events of 2006 2007).
In contrast to free media, government-controlled media had little effect on
government respect for physical integrity rights. For the most part, democracies with
government-controlled media had less respect for physical integrity rights than
democracies with free media, but more respect for these rights than autocracies. For
example, Greece from 1981 to 1995 had controlled media, the highest score for
regulation of participation, and Polity scores that put it well within the consolidated
democracy range (8 to 10). Yet during this same time, Greece had an average score of
150
only 5.67 on the CIRI physical integrity rights index. Another example of a democracy
with controlled media is Turkey from 1983 to 1991. During this time Turkey’s Polity
score stayed in the 7 to 9 range, but its regulation of participation score was lower (2 to 3
out of 5) than that found in most consolidated democracies. During this time respect for
physical integrity rights as measured by the CIRI index ranged from 5 to 3 and then
dropped to 2 in 1991 and 1992 as the conflict with the Kurds escalated to civil war. As
the war continued, government respect for physical integrity rights deteriorated further
(for a CIRI score of 0 in 1994 and 1 in 1995). Thus, Turkey demonstrates the strong
negative effect of internal conflict and war on government respect for human rights.
Likewise government-controlled media appeared to have little impact on government
respect for physical integrity rights in autocracies. For example, Egypt from 1981 to 1995
had controlled media and a consistent Polity score of -6 and regulation of participation
Score of 4. Egypt’s average CIRI physical integrity rights score was 4.2 which was below
the mean score of 5.12, but higher than those for Mexico and Uganda. Indonesia which
had controlled media and a consistent Polity score of -7 and a regulation of participation
score of 4 from 1981 to 1995, had an average CIRI score of 2.47. This is not surprising
given extreme level of autocracy and the fallout from Indonesia’s invasion of East Timor
and the conflict with Gerakan Aceh Merdeka in Aceh.
Table 22: Summary of Findings of the Effect of Media Freedom and Autocracy/Democracy on Physical Integrity Rights
Physical
Integrity
Rights
Index
Physical
Integrity
Rights
Index
(+5 years)
Free
from
Dis-
appearance
Free from
Dis-
appearance
(+5 years)
Free
from
Political
Killing
Free from
Political
Killing
(+5 years)
Free
from
Torture
Free
from
Torture
(+5 years)
Free
from
Political
Imprison-
ment
Free
from
Political
Imprison-
ment
(+5 years)
Lag of the
Dependent
Variable
+ + + + + + + + + +
Media Freedom NS-* -* NS-* NS-* NS-* NS-* NS-* NS-* NS-* NS*
Autocracy/
Democracy (5 part)
+* +* NS* NS* NS NS* NS+* NS* NS+* NS+*
Media Freedom*
Autocracy/
Democracy (5 part)
+* +* +* +* NS NS* +* +* NS* NS
Autocracy/
Democracy (Polity)
NS+* --- --- --- --- --- --- --- --- ---
Media Freedom*
Autocracy/
Democracy (Polity)
+* --- --- --- --- --- --- --- --- ---
Executive
Constraints
+ + NS NS NS NS NS NS NS+ NS+
Media Freedom*
Executive
Constraints
NS NS --- --- --- --- --- --- --- ---
Regulation of
Participation
NS+* +* NS+* +* +* +* NS+* NS+* NS NS*
Media Freedom*
Reg. of
Participation
+* +* NS* NS* NS* NS* +* +* +* NS*
GDP/capita
(logged)
NS+ + NS NS + + + + NS NS+
Population (logged) - - NS - - - - - NS- -
Interstate Conflict NS NS NS NS NS NS NS NS NS NS
Internal Conflict - - - - - - - - - NS-
NS=not significant; “+”=significant and positive; “-“=significant and negative; mixed= positive and negative
*interpretation of interactive term needs to be considered in conjunction with components of the term (see figures)
151
152
Conclusion
All in all, these findings (summarized in Table 22) suggest that although the free
media is able to play a watchdog role over government behavior, this does not always
result in improved government treatment of citizens. Indeed, in the absence of democratic
characteristics, free media, at least in the short term, are associated with decreased
government respect for physical integrity rights. In particular, with most types of physical
integrity rights and physical intergrity rights in general, the effect of free media depends
on regulation of participation. This dimension of autocracy/democracy measures the
extent of the rules that govern political participation. If these rules are nonexistent or ill-
defined, independent media appear to have a negative effect on most physical integrity
rights. This could be because in this scenario, people have information, but no clearly
defined institutional mechanisms with which to act on the information. Thus, although
nongovernmental organizations that defend human rights have suggested that free news
media would help to improve government respect for such rights (Amnesty International
2006), this appears to be the case only in consolidated democracies with highly regulated
and organized political participation.
The cases of Mexico and Uganda indicate that while in the short term watchdog
media in an autocracy trigger a repressive response, the long term results of watchdog
media merit further study. Additionally, these cases highlight the importance of the
professional environment in shaping media freedom and point to the need for further
research into the different effects of partisan and objective media on government respect
for human rights. Moreover, an emerging area of human rights research looks at
international influences, in particular foreign aid and international law (Landman 2005;
153
Hathaway 2002; Hafner-Burton 2005; Barratt 2004), but the results of these studies are
mixed (Landman 2006). While this study focuses on the effects of domestic media, it
would be worthwhile to look at the effects of international media as well.
Finally, while I am not advocating against free media, it is imperative to
understand how the effects of independent media vary and are dependent on democratic
characteristics like regulation of political participation and executive constraints that
make governments more accountable and vulnerable to public opinion. Likewise, it is
important to consider that autocratic regimes that are facing a double threat from
independent media and a lack of regulation of political participation, are more likely to
resort to repression in order to maintain control. In the next chapter, I investigate whether
these findings are consistent across other types of human rights, namely women’s
political, social and economic rights.
Technical Report
To address concerns regarding multicollinearity between democracy and media
freedom, I ran a correlation matrix for the independent variables in the contemporaneous
study using the physical integrity index as the dependent variable. This matrix (Table 23)
shows a correlation of .74 between free media and autocracy/democracy and .70 between
GDP/capita and autocracy/democracy. Not surprisingly, the interaction variable is highly
correlated with both free media (.92) and autocracy/democracy (.84). Interestingly,
Variance Inflation Factors for the independent variables (shown in Table 24) indicate that
there are concerns regarding the GDP/capita and population variables, suggesting that the
154
variance for the parameter estimates for these variables are inflated by multicollinearity,
Yet, despite this multicollinearity, the effects of both of these variables are significant.
The VIF statistics for the autocracy/democracy, media freedom and interaction variables
are below the generally accepted threshold of 4. I then ran a model dropping the
GDP/capita variable and found that the effects of the remaining independent variables
were substantively the same as in the model with the GDP/capita variable. These results
(shown in Table 28) show that without GDP/capita, the direction and significance of the
parameters for the remaining independent variables remain about the same, with
autocracy/democracy becoming more significant. Because there are strong theoretical
reasons for including GDP/capita, I have left it in the main findings.
Table 23: Correlation between Independent Variables
Ciri PIR
(lagged)
Free
Media
Auto/
Dem
Free
Media*
Auto/Dem
GDP/
capita
(logged)
Pop
(logged)
Inter-
state
Conflict
Ciri PIR
(lagged)
1.00
Free Media .36*** 1.00
Auto/Dem .45*** .74*** 1.00
Free Media*
Auto/Dem
.47*** .92*** .84*** 1.00
GDP/capita
(logged)
.42*** .48*** .70*** .62*** 1.00
Population
(logged)
-.32*** -.02 -.01 -.02 .06* 1.00
Interstate
Conflict
-.10*** .02 .03 .03 -.01 .26*** 1.00
Internal
Conflict
-.51 -.11 -.11 -.14 -.12 .26*** .09**
*p<.05 ** p<.01 ***p<.001
155
Table 24: Variance Inflation Factors
Media Freedom 2.35
Autocracy/Democracy 1.79
Media Freedom *
Autocracy/Democracy
3.99
Government Respect for Physical
Integrity Rights (lagged)
1.54
GDP per capita (logged) 23.93
Population (logged) 6.88
Interstate Conflict 1.01
Internal Conflict 1.09
Table 25: The Effect of Media Freedom and Democracy
with and without Controlling for Development
Model 6
(from Table 5)
Without
GDP
Lag Physical
Integrity Rights
.650*** .655***
Free Media -.478** -.6**
Autocracy/Democracy
(five-part)
.08* .122***
Free Media
*Auto/Dem
.195** .239***
GDP (logged) .126** ---
Population (logged) -.106*** -.096***
Interstate Armed
Conflict
.097 .062
Internal Armed
Conflict
-1.122*** -1.13***
Constant 1.69*** 2.536***
R
2
.77 .76
*p<.05; **p<.01;*** p<.001
In order to ensure that the results presented here are not driven by a few
influential cases, I calculated the DFBETA scores for the independent variable of interest,
the interaction between media freedom and autocracy/democracy. Table 26 shows the
cases with the highest (absolute value) DFBETA scores. I then ran the model, dropping
156
all 15 of the observations for each of these four countries. The results (shown in Table
27) show that the interaction between free media and autocracy/democracy remains
highly significant, and the direction and significance of the other effects remain the same
with the exception of free media, which is no longer significant, and
autocracy/democracy, which becomes highly significant.
Because there is some debate about the inclusion of lagged dependent variables
on the right hand side of the equation, I also ran the analyses looking at the effects of
media freedom and autocracy/democracy on physical integrity rights without the lagged
dependent variable. The effects are substantively similar, though the magnitude of the
effects are greater. Figure 28 shows the marginal effect of media freedom on physical
integrity rights across a range of autocracy/democracy from the model without the lagged
dependent variable.
For the analyses using the components of democracy, the high R
2
combined with
the lack of significance of most of the variables indicated that there were problems with
collinearity. Collinearity diagnostics (shown in Table 28) indicated that there were
problems with the three executive recruitment variables. I tried substituting the Executive
Recruitment Concept variable which combines the three, but the collinearity problems
persisted (see Table 29). As mentioned in the findings section, because there were strong
theoretical reasons for including the Executive Constraints and because there were some
questions as to whether or not the Competitiveness of Participation variable included
consideration of censorship of opposition media, I decided to omit the Executive
Recruitment and Competitiveness of Participation variables.
157
Finally, because diagnostic testing revealed that the analyses involving Political
Imprisonment violated the parallel regression assumption, ordered logistical analysis was
not appropriate for this dependent variable; therefore I used multinomial logistic analysis
for these models.
Table 26: The Four Largest (absolute value) DFBETA Scores for Media
Freedom*Autocracy/Democracy
Country/Year DFBETA
Mozambique/1993 -.3023
Niger/1990 .2587
Kenya/1993 -.2271
Burkina Faso/1993 -.2118
Table 27: The Effect of Media Freedom and Autocracy/Democracy
on Government Respect for Human Rights
With All Cases Without Influential
Cases
Lag Physical
Integrity Rights
.642*** .673***
Free Media -.424* -.0751
Autocracy/Democracy
(five-part)
.074* .137***
Free Media
*Auto/Dem
.199** .137***
GDP (logged) .094** .175***
Population (logged) -.106*** -.101***
Interstate Armed
Conflict
.01 .107
Internal Armed
Conflict
-1.119*** -1.095***
Constant 1.974*** 1.086**
R
2
.87 .78
*p<.05; **p<.01;*** p<.001
158
Figure 28: The Marginal Effects of Media Freedom as the Level of
Autocracy/Democracy (without the lagged dependent variable)
-2 -1 0 1 2
Marginal Effect of Media Freedom
0 1 2 3 4
Regime Characteristics
Marginal Effect of Media Freedom
95% Confidence Interval
Dependent Variable: Government Respect for Physical Integrity Rights
159
Table 28: Collinearity Diagnostics
Table 29: More Collinearity Diagnostics
Variable VIF R
2
Physical Integrity Rights Index 2.26 .56
Media Freedom 2.38 .58
Executive Constraints (Decision Rules) 3.36 .70
Regulation of Participation 1.53 .35
GDP/capita (logged) 2.37 .58
Population (logged) 1.30 .23
Interstate Conflict 1.09 .09
Intrastate Conflict 1.38 .28
Mean VIF 1.96
Variable VIF R
2
Physical Integrity Rights Index (-1 year) 2.24 .55
Media Freedom 2.55 .60
Regulation of Chief Executive Recruitment 5.19 .81
Competitiveness of Executive Recruitment 14.42 .93
Openness of Executive Recruitment 2.09 .52
Executive Constraints (Decision Rules) 9.25 .89
Regulation of Participation 1.63 .39
Competitiveness of Participation 6.00 .83
GDP/capita (logged) 2.78 .64
Population (logged) 1.33 .25
Interstate Conflict 1.10 .09
Intrastate Conflict 1.35 .26
Mean VIF 4.16
160
Chapter 6
Media Freedom, Democracy and Women’s Rights
“I asked a Burmese why women, after centuries of following their men, now walk
ahead. He said there were many unexploded land mines since the war.”
Robert Mueller (quoted in Look 1957)(Shapiro and Epstein 2006)
“When you’ve got information, whether through the press, radio or television, you have
the power to inform, to speak and to have others speak, to pass on a message and to
communicate. The most convincing messenger is the image. I am part of the women’s
struggle.”
Franceline Oubda, Burkinabe filmmaker and television director (2000)
Introduction
In May of 2009, construction began in Bobo-Dioulasso, Burkino Faso on the
“pleasure hospital,” a clinic offering reconstructive surgery for victims of female genital
mutilation/cutting(FGM)(IRIN 2009).
37
37
The term “female genital mutilation” is used interchangeably here with “excision” because excision is
the type of genital mutilation practiced in Burkina Faso. The World Health Organization has identified four
types of female genital mutilation: “1. Clitoridectomy: partial or total removal of the clitoris… 2. Excision:
partial or total removal of the clitoris and the labia minora, with or without excision of the labia
majora…3. Infibulation: narrowing of the vaginal opening through the creation of a covering seal…4.
Other: all other harmful procedures to the female genitalia for non-medical purposes. (World Health
Organization 2008)
According to the Demographic and Health
Survey, about 75 percent of women in Burkina Faso have been excised (2003). To raise
awareness of the dangers of this practice, the National Anti-Excision Committee
(CNLPE) took to the airways in 1992, using graphic images on the evening news to
demonstrate the damage done by FGM (Stolz 1998). Research indicates that there has
been a decline of FGM in Burkina Faso, but that the problem persists despite laws
banning FGM and the mass media campaign to end it (Frontiers in Reproductive Health
161
2008). Indeed the problem of FGM, like many violations of women’s rights, is difficult to
eradicate because it “is deeply rooted in culture and cuts across class, religion and ethnic
groups,” (United States Department of State 2001). Unlike physical integrity rights
which are usually violated for political reasons, women’s rights tend to be violated for
reasons that go beyond the political.
The study presented in the previous chapter provided evidence that independent
news media are not always associated with improved human rights. Specifically in an
autocracy, free media is associated with lower government respect for physical integrity
rights. What implications does this finding have for the effect of free media on other
human rights? This chapter focuses on the effect of free media on women’s social,
economic and political rights. The study here is another test for the conditional theory of
media effects on human rights which is depicted in the cycles shown in Figures 5-8 of
Chapter 3. Specifically, I am testing the hypothesis that the effect of media freedom on
human rights (in this case women’s rights) depends on autocracy/democracy (H
3
). In the
first section I discuss the distinct characteristics of women’s rights and the additional
factors that may influence them. The findings are presented in the second section. In the
third section I consider these findings in the context of Burkina Faso.
38
In the concluding
section, I reflect on the implications of this chapter for the study as a whole.
38
Through the results of the statistical analysis, I was able to identify Burkina Faso as an important case
because it offered variation in the key independent variables: media freedom and regulation of participation
and executive constraints.
162
What Makes Rights Women’s?
Unlike government respect physical integrity rights which tends by definition to
be determined by political factors, government respect for women’s rights (or the lack
thereof) is frequently linked to social and economic issues (Bunch 1990). Additionally,
with violations physical integrity rights, the government is generally the perpetrator;
whereas violations of women’s rights are often carried out by non-government actors and
are tolerated at a societal level. With physical integrity rights the government is usually
directly involved; with women’s rights the government is culpable of a lack of
enforcement, or to put it plainly, looking the other way. Moreover, as the example of
FGM illustrates, gender disparities are often deeply rooted in cultural and religious
traditions (Coleman 2004). If free news media are truly able and willing to provide a
voice to the voiceless, it stands to reason that by drawing attention the plight of
marginalized women, the media might be able to make a difference. The question I seek
to answer here is, can a free press make a difference in government respect for women’s
rights, and if so, under what conditions are free media more or less effective?
Human rights are generally divided into two categories: negative and positive.
Physical integrity rights are negative in that they exist on a fundamental level and can be
maintained as long as they are not violated; whereas rights like women’s economic and
social rights are positive in that their establishment and maintenance require programs
and government investment. Of course, as Landman (2006) posited, there are positive
and negative dimensions to all human rights. For example, protection of civil liberties
and political rights requires government investment in legal systems and electoral
163
institutions; likewise, the protection of social and economic rights requires anti-
discrimination policies (Landman 2006).
In addition to looking at the effects of media freedom and autocracy/democracy
on women’s rights in general (using the additive index described in Chapter 3), I will
analyze how these factors influence women’s economic, social and political rights
individually. The Cingranelli-Richards (CIRI) Human Rights Data Project includes a
variable for women’s economic rights which is coded based on equality in compensation,
hiring and promotion practices, freedom to choose profession without male consent,
freedom from discrimination and sexual harassment, the right to work in the military, law
enforcement, and other dangerous occupations (as mentioned in Chapter 3 each of these
indicators ranges from 0=no rights to 3=full rights). The coding scheme is based on two
primary considerations, “one, the extensiveness of laws pertaining to women’s economic
rights; and two, government practices towards women or how effectively the government
enforces the laws,” (Cingranelli and Richards 2008: 71). As an example of how this
coding scheme is applied, for the year 2006, France received a code of 2 because,
although it is illegal for superiors to harass or discriminate based on gender against
subordinates, these laws do not apply to peers. Moreover, there was evidence of gender-
based discrepancies in compensation and hiring, in spite of laws against such practices
(Cingranelli and Richards 2008). These economic rights are positive in that they require
cultural recognition of women’s potential to contribute to the workforce, but they are
negative as well because they require the prohibition of discriminatory practices. I expect
development to play a particularly important role in shaping women’s economic rights
164
because in less developed countries where jobs are scarce, society may be less willing to
accept women competing with men for employment.
In contrast, I expect cultural factors like religion to play much more of a role in
women’s social rights. The CIRI indicator for women’s social rights includes the right to
equal inheritance, the right to be an equal partner in a marriage and to initiate a divorce,
the right to obtain a passport, and travel to other states, property rights, education rights,
the freedom from FGM and forced sterilization (Cingranelli and Richards 2008). This
indicator does not include honor killings, dowry deaths, rape, domestic violence or
trafficking because the project directors plan to eventually establish separate indicators
for these issues (Cingranelli and Richards 2008). As with economic rights, the coding
scheme focuses not just on laws protecting women’s social rights, but also on
government enforcement of those laws and practices. There are also negative and
positive dimensions to these rights because they require policies that prohibit
discrimination as well as a more evolved society that accepts equality of women in most
aspects of life. As an example, Zambia in 2006 received a code of 0 because although
legally women were to be treated as equal to men, in practice compared to men, women
had poor education and employment opportunities (Cingranelli and Richards 2008).
Likewise women did not enjoy equal property, marriage and inheritance rights.
Additionally, there were traditional practices that were deemed harmful to women
including “’sexual cleansing,’ in which a widow had sex with her late husbands’ relatives
as part of cleansing ritual,” (Cingranelli and Richards 2008: 90).
165
The indicator for women’s political rights comprises voting rights, the rights to
petition government officials, join political parties, run for political office and hold
government positions, as well as representation (Cingranelli and Richards 2008). As with
the two other indicators, the coding scheme centers on the “extensiveness of laws”
pertaining to women’s political rights as well as enforcement of those laws and practice
(Cingranelli and Richards 2008). For example, for the year 2005, Switzerland received a
code of 2 because although it had universal suffrage and free and fair elections, women
held less than 30 percent of seats in the national legislature. Intuitively, I expect
autocracy/democracy to have a particularly strong effect on women’s political rights
because democratic institutions by definition would encourage a more equitable
distribution of representation.
While few would contest that women’s rights are human rights, there are many
reasons to consider the effect of free media on these rights independently from its impact
on other human rights. In particular, “many violations of women’s human rights are
distinctly connected to being female-that is, women are discriminated against and abused
on the basis of gender,” (Bunch 1990). Although the Western concept of human rights
tends to emphasize political and civil rights, those who study women’s rights consider
socio-economic rights as well because gender discrimination often has socio-economic
impacts. In fact, some have posited that advocating for women’s rights will have
economic benefits in developing countries. Coleman (2004) argued that though gender
discrimination exacts a heavy toll on women, “ultimately all of society pays a price for
them. Achieving gender equality is now deemed so critical to reducing poverty and
166
improving governance that it has become a development objective in its own right.”
However there remain significant gender disparities in many parts of the world, in part
because this discrimination is still accepted in certain cultures (Coleman 2004). Because
many, including journalists themselves, view providing a voice for the voiceless to be a
critical role of the media (Kovach and Rosenstiel 2001 ), the ability of a free news media
to improve government respect for women’s rights is a reasonable test of effective media.
Furthermore, because some cultures are more accepting of violations of women’s
rights compared to other human rights, it follows that this is a more rigorous test of the
media’s ability to affect government behavior. It is unlikely that the media would cover
violations of women’s rights in states where there is a cultural tolerance for such
violations. Thus, I expect that effect of free media on government respect for women’s
rights is dependent on whether or not the violations breech what I characterize as a
“cultural tipping point.” News media tend to cover stories that are out of the ordinary,
thus news coverage of violations of women’s rights will occur if and only if the
violations cross this cultural tipping point. Therefore cultural attitudes toward gender
equality are likely to influence not only whether or not women’s rights are respected, but
also the amount of attention and quality of coverage that the media provide. For
example, in 1975 the practice of FGM was common and widely accepted as an important
coming of age ritual for young women. Because most Burkinabe believed FGM was
appropriate, they were unreceptive to negative messages about it. Therefore, the
government radio campaign against excision was met with great hostility.
167
Certainly there are variables other than media freedom and the level of
democracy/autocracy and cultural tolerance that may increase or decrease government
respect for women’s rights. I expect the same factors that influence government behavior
regarding physical integrity rights will have similar effects on government respect for
women’s rights, although not always for the same reasons. I predict that violations of
women’s rights, as with physical integrity rights, are more likely to occur when a
government is facing a threat of either interstate or internal armed conflict (Poe et al.
1999). Although in the case of women’s rights, I expect that violations increase because
the government is too distracted by threat posed by the conflict to take measures to
enforce women’s rights. I expect the influence of the level of economic development to
be particularly important with women’s rights. Regarding physical integrity rights, the
rationale is that governments with strong economies enjoy more security from the threat
of domestic rebellion; therefore, the healthier the economy, the less likely a government
will feel the need to engage in repression (Poe et al. 1999). In the case of women’s rights
I expect that in less developed countries where there are fewer jobs to be had, men may
feel more threatened by the prospect of women entering the workforce. Indeed,
Inglehart, Norris and Welzel (2002) hypothesized that modernization drives
democratization which in turn encourages female participation in politics. Moreover,
“economic development also brings unforeseen cultural changes that transform gender
roles,” (Inglehart et al. 2002: 4-5). Of course, the argument has been made that the causal
arrow runs in the other direction, that a country’s level of development depends on
women’s economic rights (Coleman 2004). In either case, it is important to control for
168
economic development when studying women’s rights. As mentioned in Chapter 3, I am
using religion in lieu of a measure of cultural attitudes toward gender equality. My
justification for doing so is based on research indicating that there is a strong correlation
between religion and attitudes toward gender equality (Inglehart and Norris 2003).
Inglehart and Norris (2003)found that Orthodox and Muslim faiths were negatively
related with support for gender equality and that Catholic and Protestant faiths were
positively related to support for gender equality.
Therefore, while this study will focus on the relationship of the interaction
between the independent variables of media freedom (whether or not the media is free
from government control) and regime characteristics (the level of democracy or
autocracy) on the dependent variable of the government’s respect for women’s rights, I
will control for culture, economic development, involvement in international and internal
armed conflict. For the analyses focusing on the causes of government respect for
women’s rights in general, I use ordinary least squares regression with panel corrected
standard errors procedure. For the analyses of the women’s economic, social and political
rights, I use ordered logic and examine robust standard errors, controlling for the
unspecified effects of each individual country by clustering on country code.
39
39
Because these indicators are clearly ordered, the ordered logistic model was most appropriate; however it
was not possible to test the parallel regression assumption for these models because the Brant test for doing
so requires that all independent variables in the ordered model be retained in the implied binary models and
this was not possible.
In the
following section, I present the findings of these analyses, starting with an overview, then
focusing on women’s rights in general and then looking at the results for women’s
economic, social and political rights.
169
Findings Part One: Overview
The statistical analyses indicate that the effect of media freedom on women’s
rights varies depending on autocracy/democracy. While the findings vary depending on
the specific type of right, for the most part media freedom has a detrimental effect on
women’s rights in an autocratic setting and can sometimes have a positive effect in
consolidated democracies. In looking at the different types of rights, development and
religion make a difference in women’s economic and social rights, but apparently have
little effect on women’s political rights. Both internal and international conflict, for the
most part, had negative effects on women’s rights. An interesting exception to this is the
finding that sometimes internal conflict is associated with improved women’s rights five
years later.
Findings Part Two: Women’s Rights Overall
Although media freedom does not have an independent effect on women’s rights,
the interaction of media freedom with autocracy/democracy using either the five-part
measure (shown in Table 30) or the 21-point Polity score
40
40
These results are in Table 43 in the Technical Report at the end of this chapter.
does have an effect.
However, this effect is only present in the contemporaneous model. Media freedom does
not have any significant effect in the five-years-later model (see Table 31). In contrast,
autocracy /democracy had a significant and positive effect on women’s rights five years
later, but only had an interactive effect in the contemporaneous model. This suggests that
the interaction between media freedom and democracy has an immediate effect on
170
women’s rights, whereas autocracy/democracy has an independent medium term effect,
meaning that as a state becomes more democratic, government will eventually become
more respectful of women’s rights.
Table 30: Watchdog or Lapdog? The Effect of Media Freedom and Democracy (five-part measure)
on Women’s Rights N=938
Dependent Variable:
Women’s Rights
Index
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Women’s Rights
Index (-1year)
.894*** (.030) .864*** (.034) .854***
(.035)
.839*** (.039) .823***
(.041)
.822*** (.042)
Media Freedom .163 (.070) -.051 (.062) -.377**
(.126)
-.355** (.120) -.266* (.118) -.271* (.117)
Autocracy-
Democracy
.102*** (.028) .060* (.028) .048 (.035) .017 (.040) .019 (.039)
Media Freedom*
Autocracy/
Democracy
.133**(.050) .123* (.048) .094* (.046) .093* (.045)
Catholic .095 (.052) .090 (.056) .089 (.056)
Protestant .092 (.077) .076 (.079) .078 (.081)
Muslim -.108 (.081) -.109 (.084) -.109 (.084)
GDP/capita
(logged)
.103** (.035) .103** (035)
Population
(logged
-.003 (.019) .001 (.020)
Interstate
Conflict
-.080 (.104)
Internal
Conflict
-.013 (.063)
Constant .430*** (.124) .468*** (.123) .543***
(.134)
.610*** (.159) -.078 (.266) -.106 (.270)
R
2
.91 .91 .91 .92 .92 .92
*p<.05; **p<.01;*** p<.001
171
Table 31: The Effect of Media Freedom and Democracy on Women’s Rights Five Years Later N=670
Dependent Variable:
Women’s Rights Index
(+5 years)
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Women’s Rights Index .418(.090)*** .360(.090)*** .352(.091)*** .317(.095)** .280(.096)** .283 (.097) **
Media Freedom 1.296(.208)*** .101(.161) -.343(.369) -.388(.352) -.271(.328) -.311 (.332)
Autocracy-Democracy .421(.058)*** .349(.060)*** .349(.081)*** .223(.087)* .198 (.086) *
Media Freedom*
Autocracy/Democracy
.194(.122) .143(.117) .100(.115) .121 (.115)
Catholic .356(.079)*** .362(.085)*** .416 (.088) ***
Protestant .398(.183)* .460(.187)* .423 (.165) *
Muslim -.409(.212) -.319(.214) -.329 (.209)
GDP/capita (logged) .352(.076)*** .340 (.075) ***
Population (logged .066(.040) .043 (.031)
Interstate Conflict -.055 (.181)
Internal Conflict .022 (.106)
Constant 2.253(.358)*** 2.256(.339)**
*
2.315(.341)**
*
2.398(.390)*** -.747(.551) -.455 (.535)
R
2
.85 .86 .86 .86 .87 .87
*p<.05; **p<.01;*** p<.001
172
173
Figure 29: The Marginal Effect of Media Freedom on Women’s Rights across
Autocracy/Democracy
The effect of the interaction between media freedom and autocracy/democracy is
best interpreted graphically (Brambor, Clark and Golder 2006). Figure 29 depicts the
marginal effect of media freedom on women’s rights across autocracy/democracy. Here
media freedom is predicted to have a statistically significant and negative effect on
women’s rights in the most autocratic settings. Elsewhere it does not have a significant
effect on women’s rights. Niger in 1990 exemplifies this effect. In this particular year,
the media in Niger was functionally free although the country was extremely autocratic
(with a Polity score of -7, with -10 being the most autocratic) (Van Belle 2000; Marshall
et al. 2003). During this same year, there was very little respect for women’s rights in
Niger (the women’s rights index was 2, with 9 meaning full respect for these rights).
-.6 -.4 -.2 0 .2 .4
Marginal Effect of Media Freedom
0 1 2 3 4
Autocracy/Democracy
Marginal Effect of Media Freedom
95% Confidence Interval
Dependent Variable: Women's Rights Index
174
As expected, development (as measured by the log of GDP/capita) has a
significant and positive effect on women’s rights in both the contemporaneous model and
the five-years-later model (see Table 30 and Table 31). The dominant religion does not
have a significant effect in the contemporaneous model, but it does make a difference in
the five years later model. Specifically states that are predominantly Catholic or
Protestant are predicted to have greater government respect for women’s rights five years
later. Interestingly, neither population nor interstate and internal conflict have a
significant effect on women’s rights in general. The rather high R
2
s for these models (.90
to .92 for the contemporaneous models and .85 to .87 for the five-years later models)
indicate that models are a good fit; however I interpret this with caution because one of
the reasons the R
2
s are high is because I have allowed for a panel specific first-order
autocorrelation (PSAR1) and this procedure tends to yield higher R
2
s.
The results from models looking at the effects of the components of democracy
reveal that in the contemporaneous model, it is democracy as a whole that makes a
difference in women’s rights. Neither executive constraints nor regulation of
participation have an independent effect or an interactive effect on women’s rights (see
Table 32). However, five years later, the components of democracy interact with media
freedom in interesting and contradictory ways (see Table 33). In looking at the
components of democracy, media freedom has an independent, significant and negative
effect on women’s rights five years later. Both executive constraints and regulation of
participation have a significant and positive effect on women’s rights five years later.
175
However in looking at the interactions, these relationships become quite complicated, and
once again, they are best interpreted graphically.
Figure 30 depicts the marginal effect of media freedom on women’s rights five
years later as regulation of participation changes. Here we can see that media freedom is
predicted to have a significant and negative effect as regulation of participation
decreases. Specifically, holding all other factors constant, a state with media freedom,
but without regulation of participation is predicted to score 1 point lower on the 10-point
Women’s Rights Index than a state without media freedom. Figure 31 depicts the
marginal effect of media freedom on women’s rights as executive constraints change.
Here we see that media freedom does not have a statistically significant effect unless
executive constraints are at their highest level and then the effect is negative. Thus, a
state with media freedom and the highest level of executive constraints is expected to
score half a point lower on the Women’s Rights Index than a state without media
freedom, holding all other factors constant. Although it is counter-intuitive that a high
level of executive constraints combined with media freedom could detract from women’s
rights, it is plausible that a leader who has less decision-making power and faces potential
challenges from an independent media would have less ability to enforce any existing
laws protecting women’s rights. Likewise, free media are unlikely to take up the cause of
women’s rights unless there are pressures from civil society and/or societal support to do
so. Overall, the effect of media freedom on women’s rights five years later is decidedly
negative. An interesting case here is Ecuador from 1981 to 1995. During this time
Ecuador had free media and was highly democratic (with a Polity score ranging from 8 to
176
9, with 10 being most democratic). Executive constraints were at their highest level (7),
yet Ecuador’s regulation of participation was in the relatively low category of “multiple
identity” (2 out of 5), reflecting the state’s “deep geographic and ethnic cleavages” (Van
Belle 2000; Marshall and Jaggers 2007). During the same period, respect for women’s
rights remained quite low (2 out of 9).
Table 32: The Effect of Media Freedom and Components of Democracy on Women’s Rights
N=797
Dependent Variable:
Women’s Rights Index
Model 1 Model 2 Model 3
Women’s Rights Index (-1 year) .812 (.046)*** .812 (.046)*** .808 (.046)***
Media Freedom -.060 (.063) -.119 (.181) -.261 (.115)*
Executive Constraints .152 (.136) .139 (.147) .128 (.143)
Media Freedom*Executive Constraints --- .081 (.223) ---
Regulation of Participation .153 (.128) .145 (.132) -.008 (.170)
Regulation of Participation*Media Freedom --- --- .298 (.162)
Catholic .133 (.064)* .128 (.061)* .147 (.065)*
Protestant .187 (.098) .181 (.096) .191 (.096)*
Muslim -.047 (.087) -.048 (.088) -.045 (.086)
GDP/capita (logged) .156 (.053)** .153 (.051)** .151 (.052)**
Population (logged .006 (.019) .007 (.020) .004 (.019)
Interstate Conflict -.095 (.097) -.100 (.095) -.092 (.097)
Internal Conflict -.003 (.069) -.004 (.070) -.006 (.069)
Constant -.724 (.343)* -.696 (.328)* -.548 (310)
R
2
.92 .92 .92
*p<.05; **p<.01;*** p<.001
177
Table 33: The Effect of Media Freedom and Components of Democracy on Women’s Rights
5 Years Later N=570
Dependent Variable:
Women’s Rights Index
Model 1 Model 2 Model 3
Women’s Rights Index (-1 year) .316 (.094)** .312 (.095)** .297 (.094)**
Media Freedom -.278 (.135)* -.626 (.410) -1.012 (.392)*
Executive Constraints .745 (.253)** .928 (.258)*** .650 (.267)*
Media Freedom*Executive Constraints --- -1.143 (.482)* ---
Regulation of Participation .634 (.311)* .737 (.318)* .011 (.377)
Regulation of Participation*Media Freedom --- --- 1.061 (.461)*
Catholic .629 (.124)*** .686 (.124)*** .704 (.139)***
Protestant .796 (.228)*** .866 (.228)*** .856 (.223)***
Muslim -.037 (.194) -.012 (.196) .026 (.194)
GDP/capita (logged) .527 (.107)*** .549 (.104)*** .534 (.106)***
Population (logged .034 (.027) .025 (.027) .016 (.028)
Interstate Conflict .131 (.149) -.163 (.149) .143 (.150)
Internal Conflict .245 (.113) -.268 (.123)* .251 (.117)*
Constant -2.655 (.677)*** -2.878 (.658)*** -2.091 (.651)**
R
2
.90 .90 .89
*p<.05; **p<.01;*** p<.001
178
179
Figure 30: The Marginal Effect of Media Freedom on Women’s Rights 5 Years
Later as Regulation of Participation Changes
Surprisingly, when the interaction between media freedom and regulation of
participation is included (Model 3 in Table 33), internal conflict has a significant and
positive effect on women’s rights five years later. The case of Burkina Faso provides a
potential explanation for this. It is possible that a revolutionary movement will
incorporate women’s rights into its dogma. If the revolution proves to be successful, this
could lead to improvements in women’s rights down the road. This is precisely what
happened when Captain Thomas Sankara’s Conseil National de Revolution seized power
in 1983.
-2 -1.5 -1 -.5 0 .5
Marginal Effect of Media Freedom
0 .2 .4 .6 .8 1
Regulation of Participation
Marginal Effect of Media Freedom
95% Confidence Interval
Dependent Variable: Women's Rights Index (+5 years)
180
Figure 31: The Marginal Effect of Media Freedom on Women’s Rights Five Years
Later as Executive Constraints Change
Findings Part Three: Pulling Apart Women’s Rights
Media Freedom does not have an independent effect on women’s economic rights
in the contemporaneous or the five years later model (see Tables 34 and 35). The
interaction between media freedom and autocracy/democracy does have a significant
effect (see models 2 and 4 of Table 34); however, the graphic depictions in Figure 32 and
Figure 33 illustrate that the both the predicted probability of no respect for women’s
economic rights and full respect for women’s economic rights are, for the most part, not
statistically different for states with free media than for states with controlled media. In
fact, autocracy/democracy does not have a significant effect on women’s economic rights
in the contemporaneous model, but it does have a significant and positive effect five
-1 -.5 0 .5 1 1.5
Marginal Effect of Media Freedom
0 .2 .4 .6 .8 1
Executive Constraints
Marginal Effect of Media Freedom
95% Confidence Interval
Dependent Variable: Women's Rights Index (+5 years)
181
years later. Instead, when it comes to women’s economic rights, development, religion
and conflict make a difference. As predicted, GDP/capita had a significant and positive
effect on women’s economic rights in both the contemporaneous and the five years later
models. Being a predominantly Muslim state had a significant and negative effect on
women’s rights in the contemporaneous model, but was not significant in the five-years
later model. Interestingly, internal conflict had a significant and negative effect in the
contemporaneous model, but was insignificant five years later; whereas interstate conflict
had a significant and negative effect five years later, but was insignificant in the
contemporaneous model. In both models, neither executive constraints nor regulation of
participation had significant effects on women’s economic rights (these results are shown
in Tables 53 and 54 in the Appendix).
Table 34: The Effect of Media Freedom and Autocracy/Democracy on Women’s Economic Rights N=938
Dependent Variable:
Women’s Economic Rights
Model 1 Model 2 Model 3 Model 4
Women’s Economics Rights (-1 year) 2.846 (.278)*** 2.845 (.281)*** 2.859 (.277)*** 2.852 (.280)***
Media Freedom -.395 (.236) -1.030 (.401)* -.271 (.258) -.763 (.281)**
Autocracy/Democracy (5 part) .183(.118) .099 (.124) --- ---
Media Freedom*Autocracy/Democracy (5 part) --- .262 (.147) --- ---
Autocracy/Democracy (Polity) .024 (.022) .011 (.022)
Media Freedom*Autocracy/Democracy (Polity) --- .082 (.035)*
Catholic .077 (.321) .117 (.330) .058 (.314) .075 (.326)
Protestant -.050 (.336) -.094 (.348) -.057 (.336) -.103 (.344)
Muslim -.657 (.332)* -.690 (.339)* -.687 (.329)* -.716 (.335)*
GDP/capita (logged) .506 (.165)** .463 (.164)** .565 (.154)*** .513 (.151)**
Population (logged -.095 (.069) -.086 (.071) -.101 (.069) -.088 (.070)
Interstate Conflict -.676 (.407) -.697 (.413) -.652 (.405) -.687 (.413)
Internal Conflict -.562 (.227)* -.555 (.229)* -.554 (.233)* -.558 (.238)*
Wald χ
2
209.56 224.49 212.02 222.35
BIC -5260.589 -5256.279 -5258.611 -5255.658
McKelvey & Zavoina’s R
2
.65 .65 .65 .65
*p<.05; **p<.01;*** p<.001
182
Table 35: The Effect of Media Freedom and Autocracy/Democracy on Women’s Economic Rights
Five Years Later N=670
Dependent Variable:
Women’s Economic Rights (+5 years)
Model 1 Model 2 Model 3 Model 4
Women’s Economics Rights 1.550 (.287)*** 1.544 (.291)*** 1.569 (.284)*** 1.556 (.289)***
Media Freedom -.543 (.341) -1.068 (.679) -.407 (.352) -1.005 (.413)*
Autocracy/Democracy (5 part) .436 (.168)** .372 (.176) --- ---
Media Freedom*Autocracy/Democracy (5 part) --- .204 (.220) --- ---
Autocracy/Democracy (Polity) --- --- .077 (.030)* .063 (.030)*
Media Freedom*Autocracy/Democracy (Polity) --- --- --- .090 (.050)
Catholic .204 (.461) .261 (.458) .123 (.442) .185 (.452)
Protestant .253 (.456) .223 (.473) .267 (.454) .225 (.470)
Muslim -.605 (.497) -.615 (.503) -.675 (.484) -.680 (.490)
GDP/capita (logged) .509 (.234)* .494 (.233)* .607 (.219)** .581 (.220)**
Population (logged -.052 (.094) -.050 (.096) -.061 (.093) -.048 (.096)
Interstate Conflict -1.480 (.738)* -1.500 (.748)* -1.494 (.754)* -1.541 (.765)*
Internal Conflict -.626 (.322) -.610 (.316) -.676 (.347) -.675 (.344)
Wald χ
2
137.60 138.08 131.24 131.92
BIC -3407.528 -3401.987 -3404.251 -3400.406
McKelvey & Zavoina’s R
2
.57 .57 .57 .57
*p<.05; **p<.01;*** p<.001
183
184
Figure 32: The Predicted Probability of No Respect for Women’s Economic Rights
Figure 33: The Predicted Probability of Full Respect for Women’s Economic Rights
0 .005 .01 .015 .02
Pr. Full Respect for Women's Econ Rights
-10 -5 0 5 10
Autocracy/Democracy (Polity)
Free Media Controlled Media
0 .02 .04 .06 .08
Pr. No Respect for Women's Econ Rights
-10 -5 0 5 10
Autocracy/Democracy (Polity)
Free Media Controlled Media
185
The findings for women’s social rights were similar to those for women’s
economic rights. In particular, media freedom did not have an independent effect, but did
have an interactive effect (see Table 36). However, the graphic depictions of the
interactive effect show that statistically there is little difference in the predicted
probability of having no respect for women’s social rights in states that have free media
and those that have controlled media (see Figure 34). Likewise, the overlapping lines in
Figure 35 indicate there is also virtually no difference in the predicted probability of
having full respect for women’s social rights. Free media had no effect on women’s
social rights five years later (see Table 37). Autocracy/Democracy had a significant and
positive effect on women’s social rights in the analysis using the Polity score (see model
3 in Table 37), but was otherwise insignificant. Being a predominantly Muslim state was
associated with lower government respect for women’s social rights in the
contemporaneous model. Interstate conflict had a significant and negative effect on
women’s social rights five years later. In all models, development, as measured by
GDP/per capita, was associated with higher government respect for women’s social
rights. The components of democracy did not have a significant effect on women’s
social rights or women’s social rights five years later (these results are shown in Tables
55 and 56 in the Appendix).
Table 36: The Effect of Media Freedom and Autocracy/Democracy on Women’s Social Rights
N=938
Dependent Variable:
Women’s Social Rights
Model 1 Model 2 Model 3 Model 4
Women’s Social Rights (-1 year) 3.669 (.225)*** 3.670 (.225)*** 3.681 (.223)*** 3.677 (.223)***
Media Freedom -.543 (.297) -1.008 (.513)* -.406 (.300) -.822 (.378)*
Autocracy/Democracy (5 part) .222 (.105)* .163 (.124) --- ---
Media
Freedom*Autocracy/Democracy
(5 part)
--- .190 (.169) --- ---
Autocracy/Democracy (Polity) --- --- .031 (.020) .021 (.022)
Media
Freedom*Autocracy/Democracy
(Polity)
--- --- --- .068 (.042)
Catholic .279 (.290) .300 (.295) .256 (.283) .261 (.290)
Protestant .305 (.300) .267 (.308) .297 (.300) .253 (.302)
Muslim -.627 (.313)* -.656 (.321)* -.655 (.309)* -.688 (.314)*
GDP/capita (logged) .470 (.132)*** .439 (.127)** .532 (.126)*** .493 (.120)***
Population (logged -.055 (.060) -.050 (.062) -.063 (.060) -.054 (.061)
Interstate Conflict -.400 (.539) -.419 (.554) -.367 (.537) -.400 (.553)
Internal Conflict -.240 (.201) -.232 (.204) -.228 (.203) -.225 (.207)
Wald χ
2
420.28 413.86 413.91 408.82
BIC -5360.146 -5354.470 -5357.913 -5353.412
McKelvey & Zavoina’s R
2
.80 .80 .80 .80
*p<.05; **p<.01;*** p<.001
186
Table 37: The Effect of Media Freedom and Autocracy/Democracy on Women’s Social Rights Five Years Later
N=670
Dependent Variable:
Women’s Social Rights (+5 years)
Model 1 Model 2 Model 3 Model 4
Women’s Social Rights 2.091 (.249)*** 2.099 (.251)*** 2.100 (.249)*** 2.097 (.251)***
Media Freedom -.523 (.519) -.016 (.724) -.550 (.510) -.650 (.520)
Autocracy/Democracy (5 part) .316 (.180) .381 (.209) --- ---
Media Freedom*Autocracy/Democracy
(5 part)
--- -.198 (.259) --- ---
Autocracy/Democracy (Polity) --- --- .070 (.034)* .068 (.038)
Media Freedom*Autocracy/Democracy
(Polity)
--- --- --- .015 (.059)
Catholic .508 (.478) .462 (.447) .456 (.465) .464 (.458)
Protestant .592 (.448) .631 (.464) .632 (.451) .623 (.466)
Muslim -.827 (.498) -.812 (.500) -.844 (.488) -.847 (.491)
GDP/capita (logged) .628 (.224)** .643 (.221)** .665 (.204)** .661 (.203)**
Population (logged .046 (.108) -.042 (.107) .040 (.106) .043 (.107)
Interstate Conflict -1.416 (.636)* -1.396 (.642)* -1.424 (.639)* -1.432 (.648)*
Internal Conflict -.090 (.375) -.100 (.373) -.144 (.380) -.146 (.380)
Wald χ
2
212.12 215.81 213.36 214.30
BIC -3337.995 -3332.529 -3339.614 -3333.189
McKelvey & Zavoina’s R
2
.69 .70 .70 .70
*p<.05; **p<.01;*** p<.001
187
188
Figure 34: The Predicted Probability of No Respect for Women’s Social Rights
Figure 35: The Predicted Probability of Full Respect for Women’s Social Rights
0 .05 .1 .15
Pr. of No Respect for Women's Social Rights
-10 -5 0 5 10
Autocracy/Democracy (Polity)
Free Media Controlled Media
0 .005 .01 .015
Pr. of Full Respect for Women's Social Rights
-10 -5 0 5 10
Autocracy/Democracy (Polity)
Free Media Controlled Media
189
The analyses of the influences on women’s political rights revealed that neither
media freedom, autocracy/democracy, nor the components of democracy had significant
effects in the contemporaneous models (these results are shown in Table 57 in the
Appendix). Autocracy/Democracy and Regulation of Participation did have significant
and positive effects on women’s political rights five years later (see Model 1 of Table 38
and Model 1 of Table 39). The interaction of Regulation of Participation and Media
Freedom did have a significant effect. As depicted in Figure 36, the predicted effect of
media freedom on the probability of having full respect for women’s political rights is
significant and positive only when participation is highly regulated, but this effect is
small and (because the confidence margins are wide) not all that different from the effect
of controlled media. When it comes to women’s political rights, the most influential
factors seem to be previous respect for women’s political rights (which had a consistently
significant and positive effect) and interstate conflict (which had a significant and
negative effect in all models except the those analyzing the effects of democracy
components five years later). Interesting, development as measured by GDP/capita did
not have a significant effect on women’s political rights in any of the analyses.
190
Figure 36: The Predicted Probability of Full Respect for Women’s Political Rights
Overall these findings indicate that as with physical integrity rights, the effect of
media freedom on women’s rights depends on autocracy/democracy. Additionally these
results suggest that when it comes to women’s rights in general and women’s economic
and social rights in particular, development and religion matter. To put these findings in
context, the next section focuses on changes and developments in women’s rights in
Burkina Faso. This West African state witnessed tumultuous regime changes and
interesting developments in media freedom in the 1980s and 1990s. I selected Burkina
Faso using the results of the regression analyses. In order to examine how the effects
identified in the statistical analysis apply to specific cases, I needed to select a case that
was a good fit to the model, meaning that the residuals from the regression analysis were
very low, specifically that there was very little difference between the predicted value and
0 .05 .1 .15 .2
Pr. Full Respect for Political Rights
0 .2 .4 .6 .8 1
Regulation of Participation
Free Media Controlled Media
191
the actual value of the dependent variable (government respect for women’s rights).
Additionally, because I am particularly interested in the effect of media freedom in non-
democracies, I needed to select a case with these characteristics. Through examining the
residuals, I identified Burkina Faso as the non-democracy with media freedom that was
the best fit for the model.
Table 38: The Effect of Media Freedom and Democracy on Women’s Political Rights Five Years Later
N=670
Dependent Variable:
Women’s Political Rights (+5 years)
Model 1 Model 2 Model 3 Model 4
Women’s Political Rights 2.094 (.268)*** 2.077 (.368)*** 2.107 (.371)*** 2.099 (.371)***
Media Freedom .560 (.525) -.582 (.819) .864 (.510) .579 (.484)
Autocracy/Democracy (5 part) .308 (.149)* .170 (.178) --- ---
Media Freedom*Autocracy/Democracy
(5 part)
--- .465 (.247) --- ---
Autocracy/Democracy (Polity) --- --- .032 (.031) .025 (.035)
Media Freedom*Autocracy/Democracy
(Polity)
--- --- --- .046 (.047)
Catholic -.129 (.397) -.053 (.407) -.187 (.393) -.170 (.395)
Protestant .521 (.510) .426 (.520) .489 (.525) .459 (.528)
Muslim -.626 (.434) -.682 (.424) -.714 (.428) -.727 (.426)
GDP/capita (logged) -.030 (.241) -.059 (.237) .081 (.234) .070 (.235)
Population (logged) .078 (.114) .090 (.114) .069 (.114) .077 (.114)
Interstate Conflict -1.397 (.706)* -1.485 (.712)* -1.391 (.720) -1.424 (.734)*
Internal Conflict .298 (.391) .333 (.386) .311 (.407) .311 (.407)
Wald χ
2
65.48 81.12 59.54 63.73
BIC -3424.654 -3422.627 -3419.835 -3414.018
McKelvey & Zavoina’s R
2
.54 .55 .53 .53
*p<.05; **p<.01;*** p<.001
192
Table 39: The Effect of Media Freedom and Democratic Components on Women’s Political Rights
5 Years Later N=661
*p<.05; **p<.01;*** p<.001
Dependent Variable:
Women’s Political Rights (+5 years)
Model 1 Model 2 Model 3
Women’s Political Rights 2.151 (.357)*** 2.152 (.357)*** 2.104 (.360)***
Media Freedom .990 (.539) 1.080 (.828) -1.046 (1.213)
Executive Constraints .189 (.696) .212 (.769) -.170 (.759)
Media Freedom*Executive Constraints --- -.119 (1.366) ---
Regulation of Participation 1.603 (.723)* 1.620 (.827) -.122 (1.072)
Regulation of Participation*Media Freedom --- --- 3.352 (1.644)*
Catholic -.120 (.408) -.122 (.402) -.022 (.462)
Protestant .352 (.508) .355 (.505) .330 (.536)
Muslim -.814 (.417) -.814 (.417) -.813 (.421)
GDP/capita (logged) .034 (.219) .035 (.220) -.057 (.216)
Population (logged .077 (.110) .076 (.111) .067 (.116)
Interstate Conflict -1.385 (.731) -1.381 (.751) -1.314 (.722)
Internal Conflict .588 (.425) .591 (.424) .596 (.412)
Wald χ
2
79.25 79.82 85.53
BIC -3371.724 -3365.242 -3379.996
McKelvey & Zavoina’s R
2
.55 .55 .58
193
194
Media Freedom and Women’s Rights in an Autocracy: The Campaign to End FGM
in Burkina Faso
In spite of domestic and international efforts to prevent it, FGM remains a serious
threat to Burkinabe girls and women. A young girl recalled how when she was nine, her
grandmother took her on a seemingly innocent trip to visit friends:
When I arrived in the yard, I noticed several other old women and some girls
about my age. When I asked my grandmother why there were so many people
there, she told me to behave and everything would be alright. She left me there. I
panicked and tried to run away, but two women caught me and took me to an old
woman who circumcised me. I bled a lot after the excision. Every morning for a
week my grandmother washed the wound with a preparation made from roots and
dressed it with a black powder. (Unicef 2005)
Her story is all too common especially in Burkina Faso where it is estimated that 75
percent of all women have been excised (2003).
One Country, Many Coups
Burkina Faso (formerly Upper Volta), a small landlocked country in West Africa,
is one of the world’s most impoverished states. An estimated 45 percent of Burkina
Faso’s population of 15.2 million live on less than one U.S. dollar a day (IRIN 2007;
2009). The majority (55 percent) of Burkinabe are Muslim, but most people also practice
traditional African religions (2009). Since Burkina Faso gained independence from
France it has had a series of coups and dictators (Marshall and Jaggers 2007). Between
the years 1981 to 1995 (the primary focus of this study) there were three coups in
Burkina Faso. In 1982 Colonel Zerbo, who had seized control of the country in a
bloodless coup in 1980, was overthrown in a military coup (Marshall and Jaggers 2007).
Major Jean-Baptiste Ouedraogo led this military regime briefly until he was usurped by
195
Captain Thomas Sankara in 1983. Sankara maintained control until he was killed in a
coup led by Captain Blaise Compaore. In the 1990s, responding to both domestic and
international pressures, Compaore reformed the political system and introduced a
multiparty constitution(Marshall and Jaggers 2007). Compaore was elected president in
1991 in an election that was marred by an opposition boycott (2009) and remains in that
office as of this writing.
Although Burkina Faso has taken some steps toward democracy under
Compaore’s leadership, it remains quite autocratic (Marshall and Jaggers 2007). As
illustrated in Figure 37
41
, from 1981 to 1995, the Polity score remains quite low (-7 from
1981 to 1990, rising to -5 in 1991) (Marshall et al. 2003). Likewise executive constraints
remain slight (“unlimited authority” for 1981 to 1990, rising one notch in 1991). In
comparison, participation was highly regulated and categorized as “restricted” from 1981
to 1995. Meanwhile, women’s rights fluctuated, but remained consistently in the lower
range of this index. During the same time span there were major shifts in media freedom.
41
This figure was generated with STATA 10 using the Women’s Rights Index taken from the CIRI dataset
(Cingranelli and Richards 2006), the Polity Score, Executive Constraints, and Regulation of Participation
variables from the Polity IV dataset (Polity IV 2005) and the Media Freedom Index from the Global Press
Freedom dataset (Van Belle 2000) To make is possible to compare these variables, the Media Freedom
Index has been transposed (so that the lower the value the more controlled the media and the higher, the
more free) and each variable has been normalized (rescaled so that all values fall between 0 and 1).
196
Figure 37: Media Freedom, Democracy and Women’s Rights in Burkina Faso
Media Freedom, Government Control: a Mixed Message
Under Zerbo’s reign the media was party free, but the military coup of 1982
ushered in period of tight government control of the news media. Throughout this period
the state-owned media largely reflected the views of the government. However, it was
during this time that Franceline Oubda launched a program focusing on women’s issues.
I entered the field of television in 1985, through a recruitment campaign for
journalists to enhance national television programming. Once there, I observed
that there was no programming for women. Thus, I decided to create a program
called “Women and Development”, which actually focused on the participation of
women in the development process (Ellerson 2000: 231).
When Compaore acquiesced to pressures to reform in 1990, one of the first
measures he took was to make some provisions for independent media. Africa’s first
0 .2 .4 .6 .8 1
1980 1985 1990 1995
year
Reg. of Participation Exec. Constraints
Polity Women's Rights
Media Freedom
197
independent radio station, Horizon FM began broadcasting in Burkina Faso on December
31, 1990, following an eight year struggle by Horizon founder Moustapha Thiombiano
(Geekie 1992). The 1990 Information Code made it possible for private citizens like
Thiombiano to own a radio station, but these reforms were limited at best. On the one
hand the 1990 Information Code granted press freedom, but on the other, newspapers and
journalists were not allowed to defend themselves against defamation charges which in
effect gave the government broad powers to intimidate reporters and news organizations
alike (1995; 1996). Still, these reforms opened the door for independent media. Media
observers noted that the emergence of independent media also had an effect on the state-
owned media. In particular, the independent newspaper, L’Observateur, which started
publishing in February of 1991 and published articles by members of opposition groups
as well as articles by its own reporters, soon became the country’s most popular
newspaper, usurping the state-owned Sidwaya. “The new competition forced Sidwaya to
begin to develop its own editorial line, which was no longer necessarily that of the
government,” (Lardner 1991: 10). These new developments also cultivated journalistic
professionalism. In June of 1991, the Association of Journalists of Burkina held a
seminar which prompted the journalists working at Sidwaya to demand more
independence from the government and following the seminar, members of the
opposition enjoyed more coverage by the state-owned paper (Lardner 1991). In 1993
journalists facing libel charges were allowed to defend themselves in court, and this
prompted the news media to act more independently (1995). However, the media’s
relationship with Compaore became contentious when Norbert Zongo, the editor of
198
L’Independent, and three other people were murdered. An independent investigation
named six members of the presidential guard as suspects and concluded that the killings
were connected to Zongo’s investigative reporting regarding the death of the president’s
brother’s chauffeur (Reporters Sans Frontieres 2009). The public was outraged and
responded with protests and strikes. In December of 2008, on the 10
th
anniversary of the
killings, thousands demonstrated calling for the killers to be punished (Reporters Sans
Frontieres 2009). Although the Zongo murder underscored the dangers of reporting
critically about the Compaore regime, as of this writing, the media remains “relatively
free and criticism tolerated,”(Reporters Sans Frontieres 2009).
Burkinabe Women’s Economic, Social and Political Rights
There are a number of traditional practices that have marginalized Burkinabe
women. In addition to FGM, underage marriage and polygamy are common, and
although the 1991 Constitution is non-discriminatory, “Traditional law does not
recognize inheritance rights for women and regards the woman as property that can be
inherited upon her husband’s death,” (IRIN 2007). The tradition called pogsyure is
particularly degrading of women. Pogsyure is basically an institutionalized exchange of
women.
According to pogsyure, a man from one lineage gives his daughter in marriage to
a man of another lineage in exchange for the first daughter to come out of this
marriage, who will then become part of the former household and lineage.
Pogsyure is thus a mechanism of circulation and exchange of women, (Englebert
1999: 138).
199
Although this practice has subsided over time due to changes brought by colonization, the
Catholic missions and modernization, remnants of this tradition, especially the
persistence of polygamy, remained evident in the 1980s (Englebert 1999).
Because most Burkinabe (about 90 percent) are subsistence farmers, the labor of
women is critical (Englebert 1999; Central Intelligence Agency 2009). In addition to
preparing meals, women in rural areas must often walk for miles for water. Moreover,
Burkinabe women have very little access to land. Although the national laws regarding
land are gender neutral, in practice traditional systems are prejudiced against women.
The majority of Burkinabe women can only hope to receive a small plot of land
from their husband—otherwise they have few ways to claim land. The women’s
use rights are temporary and they are responsible for growing crops for their
families’ use, such as cereals, legumes, and peanuts. The land they are given is
often of poor quality and far away (Hatcher, Meggiolaro and Ferrer 2005: 24).
In urban areas, women enjoy more economic autonomy; however, in the 1990s
they still were “underrepresented in the formal economy and public administration, where
they make up less than 10 percent of the salaried private sector and less than 20 percent
of the public service,” (Englebert 1999). Additionally, there is a great disparity between
the education of women and men, especially in rural areas. About 15 percent of women
age 15 and over are able to read and write compared to 29 percent of men over 15
(Central Intelligence Agency 2009). In some rural areas 95 percent of the women are
illiterate (IRIN 2007). The extreme poverty and comparative lack of education of women
make it difficult for them to assert their rights. It also limits their access to information
from the news media.
200
The Campaign to Stop FGM in Burkina Faso
In Burkina Faso, excision is often part of a rite of passage and is not restricted to a
particular class or religion; it is widespread (United States Department of State 2001).
Recent efforts to stop excision have employed a variety of tactics, including the use of
radio, television and film to disseminate information about the disastrous effects of this
practice. The primary justification provided for FGM is that it will discourage female
infidelity. Other reasons include beliefs that excision “cleanses” women and improves
fertility and/or that contact with the clitoris during childbirth is fatal to infants (Kouba
and Muasher 1985; United States Department of State 2001). There is little consensus as
to when and where the practice of FGM originated, but there is evidence that this practice
has been going on in many parts of the world for centuries.
Circumcised females have been discovered among the mummies of ancient
Egyptians. Herodotus, the Greek historian, found the Egyptians practicing male
and female circumcision when he visited their country around the middle of the
fifth century B.C. A Greek papyrus in the British Museum, dated 163 B. C., refers
to operations performed on girls in Memphis at the age when they received their
dowries. (Kouba and Muasher 1985: 95)
In Burkina Faso, efforts to eradicate this practice date back to the early 1900s when
Christian missionaries threatened to excommunicate families of excised girls (Lamizana
1997). These threats proved ineffective because most men refused to marry women who
had not undergone FGM. After the independence of Burkina Faso in 1960, the new
government launched awareness campaigns against FGM, but these failed to stop the
practice (Lamizana 1997). When the United Nations declared 1975 to be International
Women’s Year, activists in Burkina Faso responded with a mass radio campaign against
FGM that reached even “the most remote villages” (Assaad 1982: 394). These
201
broadcasts featured a gynecologist who detailed the harmful effects of excision and
disclaimed the “unfounded beliefs regarding morality, sexuality, and childbirth,”
(Triendregeogon 1982: 362). This message was not well received even by women in
healthcare and education because it went against practices that were widely accepted
(Triendregeogon 1982).
When Sankara seized power, he emphasized the importance of women’s
liberation as a way to enhance development, and under his watch a number of steps
aimed at improving the lot of women were taken (Englebert 1999). In 1985, Sankara’s
government established the Union of Burkinabe Women and created a ministry of family
development which led to a family law plan (this was not passed until 1988 and still
permitted polygamy). In 1985, galvanized by “National Women’s Week,” activists began
to urge the government to ban FGM. In the end though, Sankara’s Conseil National de
Revolution (CNR) “brought more hope than concrete results with respect to women’s
plight,” especially for those living in rural areas where the campaign against excision met
with a hostile reception (Englebert 1999: 141).
In 1988, after the coup with brought Compaore into power, Burkina Faso’s
Ministry of Social Welfare held a seminar on traditional practices that brought together
more than 300 people from a range of backgrounds including traditional leaders,
women’s activists and professionals from the legal and health care sectors (Lamizana
1997). This seminar led to the formation of a provisional committee on excision. In 1990
a presidential decree formally established the National Committee against Excision
202
(CNLPE). The first lady of Burkino Faso, Chantal Compaore remains the honorary
chairperson of the committee and is an active supporter of efforts to end excision.
The committee employed a variety of strategies to raise awareness about the
harmful effects of FGM, including encouraging graphic news coverage. In 1992 an
evening news report on excision featured a plastic model of female genitals. In 1996 the
committee also showed part of a documentary on television that shows an excision being
performed:
Jacob Ouedraogo, the high commissioner for Yatenga province, recalls: "An
enormous number of people phoned me demanding that I stop the broadcast
straight away. The following day they told me they would smash up the television
equipment if we dared show it again." After this shock treatment, "things did
change, but it can still be an emotive issue. When you see these horrible things
happening, you think of your own child. But then you come up against the
traditional belief that, to be a proper woman, you’ve got to have had an
excision."(Stolz 1998)
In addition to using television to show people what really happened during
excision, the committee established the SOS Circumcision, a telephone hotline, for
people to use to report cases of girls who were in danger of excision or who had been
excised. The committee also offered training sessions targeting people in key sectors,
including media professionals. Part of this training session included the showing of the
documentary (mentioned above) that ends with a young girl dying from complications
following excision (United States Department of State 2001). There is evidence too that
international pressures gave President Compaore incentive to craft policy aimed at
eradicating FGM. In 1994 Burkina Faso’s loan agreement with the World Bank and
International Monetary Fund included a promise that Compaore’s government would step
up the campaign against excision (1994).
203
The results of this campaign are mixed. In 1996 a law was passed that
criminalized FGM and imposed punishments including fines and imprisonment.
According to government records, since the law was passed 400 people have been
sentenced for excision and the percentage of girls undergoing FGM has decreased by 40
percent (IRIN 2007). However, there are also indications that girls are now being
circumcised at a much younger age, sometimes as infants, in order to avoid detection
(IRIN 2005). This clearly indicates that practices that are deeply rooted in culture and
tradition are particularly difficult to eradicate even if government, media and civil society
work in tandem to do so. One positive outcome is that Burkinabe are no longer afraid to
talk about FGM.
Information drives and outreach programmes at national, regional and even
individual level have had an impact—genital mutilation is no longer a taboo
subject for discussion. One campaigner recalls how, on a visit to a school on the
outskirts of Ouagadougou, the girls all knew the SOS Circumcision number by
heart, (IRIN 2005).
Pulling the Past into the Future: Trends in Burkinabe Women’s Rights
Since 1981, Burkina Faso has seen a number of changes in its political and media
systems. Likewise there have been some changes in women’s rights, and some areas have
improved more than others. Figure 38 illustrates the shifts in women’s economic, social
and political rights from 1981 to 2007.
42
42
This figure was generated using STATA 10 and the CIRI dataset (Cingranelli and Richards 2006).
Women’s political rights improved with
Sankara’s revolution in 1983 and again with the political reforms and new constitution in
1991. Yet, during this same period women’s social and economic rights remained poor
(between “no rights” and “some rights, but poor enforcement” for the most part). In
204
particular, it is interesting that as media gained freedoms in 1991, women’s social rights
decreased and remained at the “no rights” level for a decade.
43
Likewise, women’s
economic rights decreased in 1993 to the “no rights” level and did not increase again
until 2003 when it was coded at “some rights and effective enforcement” and then
decreased to “some rights, but poor enforcement.” This figure shows that in the case of
Burkina Faso, women’s political rights improved with the introduction of some political
liberalization and media freedom, but these changes were also associated by decreased
women’s social rights which were followed by decreased women’s economic rights.
Perhaps in the face of political change, there is a tendency to push back at the societal
level which encourages those who are not comfortable with the political changes to
adhere more strongly to their traditional practices which, in the case of Burkina Faso,
tend to be practices which marginalize women and girls. It follows too that as respect for
women’s social rights decreases there will be a subsequent decrease in support of
women’s economic rights, as Figure 9 depicts. The case of Burkina Faso suggests that it
is easier to institute reforms that improve women’s political rights than it is to improve
social rights. The trajectory depicted for economic rights in Figure 37 suggests that over
time improved political rights may lead to improved economic rights, but in the case of
Burkina Faso it is not yet possible to tell if these trends will be sustained.
43
Data was not available for Women’s Social Rights for 2005 and 2006 but the coding was at 0 for 2004
and 2007.
205
Figure 38: Women’s Economic, Social and Political Rights in Burkina Faso
0 .5 1 1.5 2 3
1980 1990 2000 2010
year
social rights economic rights
political rights
Table 40: Summary of Findings of the Effect of Media Freedom and Autocracy/Democracy on Women’s Rights
NS=not significant; “+”=significant and positive; “-“=significant and negative; mixed= positive and negative
*interpretation of interactive term needs to be considered in conjunction with components of the term (see figures)
Women’s
Rights
Index
Women’s
Rights
Index
(+5 years)
Women’s
Economic
Rights
Women’s
Economic
Rights
(+5 years)
Women’s
Social
Rights
Women’s
Social
Rights
(+5 years)
Women’s
Political
Rights
Women’s
Political
Rights
(+5 years)
Lag of the Dependent Variable + + + + + + + +
Media Freedom NS/-* NS/-* NS/-* NS/-* NS/-* NS NS NS*
Autocracy/Democracy (5 part) NS/+* + NS* NS/+ NS/+* NS NS NS/+
Media Freedom*Autocracy/
Democracy (5 part)
+* NS NS* NS NS NS NS NS
Autocracy/Democracy (Polity) NS/+ NS NS* + NS NS/+* NS NS
Media Freedom*Autocracy/
Democracy (Polity)
+* NS +* NS* NS NS NS NS
Executive Constraints NS +* NS NS NS NS/+ NS NS
Media Freedom* Executive
Constraints
NS -* NS NS NS NS NS NS
Regulation of Participation NS NS/+* NS NS NS NS NS NS/+*
Media Freedom*
Regulation of Participation
NS +* NS NS NS NS NS +*
Catholic NS/+ + NS NS NS NS NS NS
Protestant NS/+ + NS NS NS NS NS NS
Muslim NS NS - NS/- - NS NS NS
GDP/capita (logged) + + + + + + NS/+ NS
Population (logged NS NS NS NS NS NS NS NS
Interstate Conflict NS NS NS NS/- NS - - NS/-
Internal Conflict NS mixed - NS/- NS/- NS/- NS NS
206
207
Conclusion
Overall these findings demonstrate that media freedom and democracy have
similar effects on women’s rights as they do on physical integrity rights. Table 40 shows
the summarized findings. Media freedom does not have an independent effect, but it
does have an effect that depends on democracy on women’s rights in general and on
women’s economic rights, providing support for H
3,
that effect of media freedom on
government respect for women’s rights varies depending on the level of
autocracy/democracy. Democracy, when it is significant, has a positive effect on
women’s rights and women’s economic, social and political rights. The components of
democracy, executive constraints and regulation of participation were, for the most part,
not significant. Being a predominantly Catholic or Protestant state was associated with
higher government respect for women’s rights in general in both the contemporaneous
and the five-years-later models. Being a predominantly Muslim state was associated with
lower respect for women’s economic and social rights. So, as predicted, religion did
make a difference for women’s social rights, but it was also an important influence on
women’s economic rights. As expected, development (as measured by a log of the gross
domestic product per capital) was associated with improved government respect for
women’s rights in general, as well as women’s economic and social rights. Findings for
the effect of GDP/per capita on women’s political rights were mixed. Unlike the findings
in the analyses of physical integrity rights, population did not have a significant effect on
women’s rights. The findings for the effects of conflict on women’s rights were
interesting in that they were different from those for physical integrity rights. In the
208
previous chapter, the findings revealed that interstate conflict does not have a significant
effect on physical integrity rights. Yet, interstate conflict did have a significant and
negative effect in some of the models looking at women’s economic, social and political
rights, supporting the premise that interstate conflict distracts and detracts from
government attention to domestic issues. Specifically, a government dealing with
interstate conflict might be less inclined or able to expend resources enforcing any laws
protecting women’s rights. Internal conflict, which had a consistent significant and
negative effect on physical integrity rights, did not have a significant effect in the
contemporaneous model of women’s rights or on women’s political rights. Across all
models, the lags of the dependent variable are statistically significant and positive,
indicating that the level of respect for women’s rights is persistent and slow to change.
44
The case of Burkina Faso in the 1980s and 1990s shows that improvements in
women’s political rights do not always translate into improvements in women’s social
and economic rights. In fact while media freedom and political liberalization greatly
improved women’s political rights, women’s social and economic rights actually
declined. The struggle to eradicate FGM illustrates the difficulty in improving women’s
social rights in societies where the practices that marginalize women are deeply rooted in
tradition. Similarly, the subjugation of Burkinabe women through a lack of education, as
well as inheritance and property rights that are constrained by traditional law and
practice, underscores the difficulty of bringing about improvements in women’s
44
Achen (2000)posited that inclusion of a lagged dependent variable on the right side of the equation
masks the effect of the independent variables. I suspect this is the case here, but included it because there
are both theoretical and methodological reasons for including the lagged dependent variable, to model both
persistence of respect for women’s rights and address autocorrelation issues.
209
economic rights. While media freedom and political liberalization are associated with
improvements in women’s political rights, thus far they have not been effective in
improving women’s economic rights. The case of the fight against FGM suggests that
media campaigns aimed at bringing about change will not succeed unless there is an
element in society that is receptive to this change. Yet, the success of the campaign in
breaking the taboo surrounding FMG indicates that overtime, persistent media exposure
can bring about a shift in public opinion.
Technical Report
While using a pooled cross-sectional-time-series allows me to test my hypotheses
across countries and over time, doing so presents some methodological challenges.
Specifically I have to take into account panel-specific effects as well as time series issues.
With the Women’s Rights Index, I had a choice between estimating models using
ordinary least squares or ordered logistic regression. Because I determined it is important
to account for time, and because of ease of interpretation, I chose to use OLS. Because
media freedom, autocracy/democracy and women’s rights are usually slow to change
which means that the fixed effects method is not appropriate; therefore I chose to employ
the panel corrected standard errors procedure.
In the analyses of women’s economic, social and political rights, I chose to use
ordered logistic regression because these variables are limited, but clearly ordered.
Although this approach did not permit me to consider the effects of time, I was able to
control for unspecified effects of each individual country by clustering on country code.
210
This procedure does make it more difficult to obtain significant results. It was not
possible to test the parallel regression assumption for these models because the Brant test
for doing so requires that all independent variables in the ordered model be retained in
the implied binary models and this was not possible.
In all models, I included a lagged dependent variable on the right side of the
equation because there are both theoretical (to model persistence) and methodological (to
address autocorrelation) reasons for doing so. I have some concerns that including the lag
masks the effects of the other independent variables(Achen 2000); however since
including the lag was a more difficult test of my hypotheses and therefore decreases my
chances of making a Type I error (a false positive; rejecting the null hypothesis when the
null hypothesis is true), I have opted to retain the lagged dependent variable in the model.
To address concerns about multicollinearity, especially between media freedom
and autocracy/democracy, I report the variance inflation index in Table 41. These results
indicate that there is not a problem with media Freedom or with autocracy/democracy
because the variance inflation factors for the variables are below the generally accepted
threshold of 4. When the interactive term is added, there are some issues; however, as
Achen (1982) argued, while multicollinearity makes it difficult to get significant results,
it does not violate any regression assumptions.
211
Table 41: Collinearity Diagnostics/Women’s Rights
VIF VIF
Women’s Rights Index 1.74 1.74
Media Freedom 2.47 7.64
Autocracy-Democracy 3.76 12.43
Media Freedom*
Autocracy/Democracy
--- 1.72
Catholic 1.86 1.87
Protestant 1.69 1.72
Muslim 1.84 1.86
GDP/capita (logged) 2.38 2.51
Population (logged 1.19 1.19
Interstate Conflict 1.09 1.09
Internal Conflict 1.16 1.16
Mean VIF 1.92 3.47
Multicollinearity is also an issue with the models including the components of
democracy. Preliminary analyses indicated that there were problems with the three
indicators for executive recruitment, so I elected to use the executive recruitment concept
variable that incorporates those indicators. Even so, these results displayed classic signs
of multicollinearity (a high R
2
without any significant coefficients). Diagnostics (shown
in Table 42) confirmed there were still substantial problems with multicollinearity,
especially with the executive recruitment, executive constraints and competitiveness of
participation variables. Because previous research has indicated the executive constraints
variable is critical to the Polity score (Gleditsch and Ward 1997), I opted to keep it and
drop executive recruitments. I chose to drop the competitiveness of participation variable
because in addition to the collinearity issue, there is some ambiguity in the latest version
codebook as to whether or not the measure incorporates censorship of opposition media
212
(Marshall and Jaggers 2009). Excluding these two variables took care of the
multicollinearity issue and allowed me to test my hypotheses.
Table 42: Collinearity Diagnostics with Democracy
Components/Women’s Rights
VIF VIF
Women’s Rights Index (-1
year)
1.79 1.73
Media Freedom 2.48 2.44
Executive Recruitment 4.75 ---
Executive Constraints 8.43 3.01
Regulation of Participation 1.50 1.47
Competitiveness of
Participation
7.48 ---
Catholic 1.89 1.89
Protestant 1.73 1.70
Muslim 1.99 1.83
GDP/capita (logged) 2.86 2.62
Population (logged 1.23 1.18
Interstate Conflict 1.09 1.09
Internal Conflict 1.20 1.17
Mean VIF 2.96 1.83
In the previous chapter, preliminary analyses revealed that autocracy/democracy as
measured by the Polity score does not have a linear effect on physical integrity rights. For
these analyses looking at women’s rights I use both the rescaled five-part
autocracy/democracy and the Polity score. For the most part the results (shown in Table
43) are substantively the same as those with the five-part autocracy/democracy measure.
Table 43: Watchdog or Lapdog? The Effect of Media Freedom and Democracy (Polity2)
on Women’s Rights N=938
Dependent Variable:
Women’s Rights Index
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Women’s Rights Index (-
1year)
.894***
(.030)
.880***
(.034)
.866***
(.032)
.849***
(.036)
.825***
(.041)
.824***
(.041)
Media Freedom .163 (.070) .025 (.069) -.232*
(.090)
-.215**
(.089)
-.155
(.090)
-.162
(.091)
Autocracy/Democracy .015**
(.005)
.008
(.005)
.006
(.007)
.001
(.007)
.001 (.007)
Media Freedom*
Autocracy/Democracy
.041**
(.012)
.037**
(.011)
.026*
(.011)
.026*
(.001)
Catholic .093
(.054)
.086
(.058)
.084 (.057)
Protestant .094
(.079)
.076
(.081)
.077 (.082)
Muslim -.120
(.082)
-.116
(.086)
-.116
(.085)
GDP/capita (logged) .117**
(.036)
.117**
(036)
Population (logged -.002
(.019)
.002 (.020)
Interstate Conflict -.073
(.105)
Internal Conflict -.017
(.063)
Constant .430***
(.124)
.529***
(.133)
.564***
(.133)
.632***
(.148)
-.191
(.281)
-.215
(.284)
R
2
.91 .91 .91 .91 .92 .91
*p<.05; **p<.01;*** p<.001
213
214
Chapter 7
Journalists on the Frontline without Weapons
“News is what somebody somewhere wants to suppress; all the rest is advertising.”
Alfred Harmsworth, 1
st
Viscount Northcliffe, British publisher 1865-1922
In 2007 a group of 47 journalism students met twice a month in Azerbaijan to
interview newsmakers. They were working on a production of a United States Agency of
International Development (USAID) media project aimed at teaching them to ask tough
questions because “normal press conferences consist of a government spokesperson
reading a statement without being questioned,” (USAID 2007). Also in Azerbaijan in
2007, USAID sponsored a television production called “Problems/Solutions” in which
local television stations were “tasked with exploring instances of corruption, abuse of
power, lack of social services and then recording those infractions,” (USAID 2007).
As the project continues, the quality standards have gradually increased, forcing
station staff to improve both reportage and technical skills. The
“Problems/Solutions” program producer remains in constant contact with local
stations, offering constructive criticism to support the continual need to improve.
And there is strong financial incentive to succeed: this “learn as you earn”
program compensates stations only for packages that are accepted (USAID 2007).
These are just two of many media assistance projects designed to foster
independent media in developing countries. What these reports, or as USAID termed
them, “Success Stories,” did not specify is that it is quite risky to ask tough questions and
report on corruption and abuse of power in non-democracies like Azerbaijan. At the same
time these USAID projects were underway, Azerbaijan was identified by the Committee
to Protect Journalists (2007) as “one of the world’s leading jailors of journalists.” Those
215
imprisoned included editor Eynulla Fatullayev, who was arrested after he published a
story on “an official cover-up” of the 2005 murder of his colleague, Elmar Huseynov.
Journalists told CPJ that they saw no incentive to cover sensitive subjects—the
most dangerous of which is reporting on the [President] Aliyev family (the
president’s family)—when colleagues have been beaten, harassed, or jailed after
taking that risk (Committee to Protect Journalists 2009).
On the face of it, promoting media freedom seems like a good idea. Having
independent media should help to hold government accountable and improve government
respect for human rights, and it probably does in consolidated democracies. The findings
presented here suggest free media has a different and often detrimental effect in non-
democracies. Free media are by definition able and willing to criticize bad government
behavior, but in the absence of democracy, the dissemination of negative information
about the government can promote dissent which in turn can lead to repression of
journalists and protesters alike.
In this concluding chapter, I first summarize the statistical findings from the three
studies which focused on the effect of media freedom and autocracy/democracy on
protest, physical integrity rights and women’s rights. In the second section I compare the
four non-democracies with free media that were explored in this project and consider the
implications of these cases. The results presented here raise new questions to be explored
in future research. Thus, I consider the avenues for future research in the third section. In
conclusion, I summarize the theoretical and polity implications of this project.
Table 44: Summary of Findings of the Effect of Media Freedom and /Democracy on Protest and Human Rights
Media
Freedom
Autocracy/
Democracy
Media
Freedom*
Autocracy/D
emocracy
Regulation
of
Participation
Media
Freedom*
Regulation of
Participation
Executive
Constraints
Summary
General
Strikes
+* +* -* --- --- --- General strikes are more likely in
non-democracies with media
freedom.
Anti-
Government
Demonstra-
tions
NS NS NS --- --- --- Demonstrations are more likely
as repression increased
Riots NS NS NS --- --- --- Riots are more likely as
repression increased
Physical
Integrity
Rights
NS-* +* +* NS+* +* + Physical integrity rights increase
as democracy and democratic
components increase. The effect
of media freedom varies,
depending on democracy and
regulation of participation
Free from
Disap-
pearance
NS-* NS* +* NS+* NS* NS Disappearance is more likely in
autocracies with media freedom
Free from
Political
Killing
NS-* NS* NS* +* NS* NS Political killing is more likely in
states with low regulation of
participation and media freedom
Free from
Torture
NS-* NS+* +* NS+* +* NS Torture is more likely in states
with media freedom and low
regulation of participation or
autocracy
Free from
Political
Imprisonment
NS-* NS+* NS* NS* +* NS+ Results were mixed for political
imprisonment
Women’s
Rights
NS-* NS+* NS+* NS+* NS+* NS+* Women’s rights increase as
democracy and democratic
components increase. The
effect of media freedom
depends on democracy and
regulation of participation
216
Table 44, Continued
Media
Freedom
Autocracy/
Democracy
Media
Freedom*
Autocracy/
Democracy
Regulation of
Participation
Media
Freedom*
Regulation of
Participation
Executive
Constraints
Summary
Women’s
Economic
Rights
NS-* NS+* NS* NS NS NS Women’s economic rights are
more influenced by development,
religion and conflict
Women’s
Social
Rights
NS-* NS+* NS NS NS NS+ Women’s social rights are more
influenced by development,
religion and conflict
Women’s
Political
Rights
NS* NS+ NS NS+* NS+* NS Women’s political rights are more
likely to be fully respected in
states with the highest level of
regulation of participation and
media freedom
Summary
The effect of
media
freedom
depends on
democracy
and
regulation of
participation
Autocracy/
democracy
has a
positive
effect on
large protests
and most
human rights
Media
freedom &
autocracy
has a
positive
effect on
large
protests and
a negative
effect on
most human
rights
Regulation of
participation
has a positive
effect on most
human rights
Media freedom
& low
regulation has a
negative effect
on most human
rights
Executive
constraints
have a
positive
effect on
most human
rights
Summary Statement:
Democracy, executive constraints
and regulation of participation
have a positive effect on the most
human rights in this study. Media
freedom & autocracy has a
negative effect on most human
rights; whereas media freedom &
the highest level of democracy
have a positive effect on some of
the rights studied here.
NS=not significant; “+”=significant and positive; “-“=significant and negative; mixed= positive and negative
*interpretation of interactive term needs to be considered in conjunction with components of the term (see figures)
217
218
Summary of Findings
The findings support a conditional theory of the effect of free media on protest
and human rights. In particular, I hypothesized that the effect of free media would not
have the same effect in non-democracies as they do in democracies. In considering the
potential role of media in the repression-protest nexus, I posited that in the absence of
democratic characteristics like executive constraints, political competition and
participation, watchdog media in a repressive regime would serve to promote protest,
which in turn might lead to more repression. The rationale is that news coverage of
government abuses will outrage citizens, but in a non-democracy the people have no
institutional mechanisms to push for reform. Consequently, if the repression is severe, the
need for reform intensifies and protest becomes more likely. Moreover, free media can
help facilitate the mobilization necessary for large scale protest. Thus, the combination of
the threat from repression and the information and opportunity for mobilization offered
by free media increase the likelihood of protest and more repression. I tested these
hypotheses across 121 countries from 1981 to 1995. Overall, the results, which are
summarized in Table 44, support these hypotheses; media freedom in non-democracies
was associated with an increased likelihood of protest and repression.
In the first set of analyses, which focused on the effects of media freedom,
autocracy/democracy and repression on different types of protest, the results indicate that
the effect of free media on protest and government respect for human rights varies
depending on the presence of democratic institutions. In particular, autocratic states with
free media were more likely to have a general strike than those without free media.
219
Interestingly, neither free media nor autocracy/democracy had a significant effect on the
likelihood of an anti-government demonstration or riot. However, according to the
definition used by the Cross-National Time-Series Data Archive, general strikes require
ten times the participation of anti-government demonstrations and riots (one thousand as
opposed to one hundred) (Banks 2008). Therefore, a general strike in this analysis is a
large scale protest, whereas the demonstrations and riots may be much smaller events. It
follows then that independent news media would make a difference with large scale
protests because they can be used to mobilize large numbers of people to act collectively.
Regardless of the level of autocracy/democracy or media freedom, repression was
associated with an increased likelihood of a general strike, anti-government
demonstration and/or riot. Thus, there is support here for the argument that both the threat
presented by repression and the opportunity presented by independent media can fuel
protest movements.
The results of the analyses of the effect of media freedom and
autocracy/democracy on physical integrity rights show that media freedom only has a
positive effect in the most consolidated democracies, has no significant effect on
countries that are somewhat democratic (in the middle of the autocracy/democracy
spectrum), and has a negative effect the most autocratic states. Interestingly, in looking at
the different components of democracy and media freedom, it is the interaction between
media freedom and regulation of participation that make a difference. Regulation of
participation is defined as the extent to which there are rules that govern political
participation (Marshall and Jaggers 2009). Thus, the findings here indicate that media
220
freedom has a positive effect on human rights in states that have clearly defined rules of
the game for political participation. In the absence of these rules media freedom has no
effect or it has a negative effect. This suggests that people need both information and a
structure that allows them to act upon that information in order to bring about
improvements in physical integrity rights. These effects are, for the most part, consistent
across difference types of physical integrity rights.
45
The results of the analyses focusing on women’s rights were similar to those for
physical integrity rights in that the effect of media freedom in autocracies was significant
and negative. Even more striking, the effect of media freedom on women’s rights five
years later was negative and significant across low to middle levels of regulation of
participation. In contrast, media freedom, autocracy/democracy and regulation of
participation had little or no significant effect on women’s social, economic and political
rights. Although the interaction of media freedom and regulation of participation did have
a significant effect on women’s political rights five years later, this effect was slight.
Instead, the results indicate that women’s social and economic rights are far more
influenced by economic development, religion and conflict. In particular, states with a
higher GDP/capita were more likely to respect women’s economic and social rights.
States that were involved in internal or interstate conflict and states that were
Disappearance was more likely in
autocracies with media freedom. Political killing and torture were more likely in states
with media freedom and low levels of regulation of participation.
45
The results for political imprisonment were less straightforward because I used multinomial logistic
regression. Neither media freedom nor democracy had an independent effect, but the interaction of media
freedom with regulation of participation was significant in some cases.
221
predominantly Muslim states were less likely to respect women’s social and economic
rights.
These analyses indicate that media freedom and autocracy/democracy do make a
difference with physical integrity rights and some women’s rights, but not in the way
many human rights advocates and policy makers would expect. The findings here suggest
the effect of media freedom is dependent on democratic characteristics, in particular
regulation of participation. In the next section I compare how the combination of media
freedom and non-democracy has influenced protest movements and human rights in the
four countries explored in this project: Uruguay, Mexico, Uganda and Burkina Faso.
Media Freedom in Non-Democracies
To put the statistical findings in context, I explored the relationship between free
media, non-democracy, repression and protest in Uruguay, Mexico, Uganda and Burkina
Faso. Table 45 provides a summary of some of the characteristics of these countries. The
protest movement in Uruguay in the first half of the 1980s provides an example of the
interactive nature of the repression and protest and the media’s intervening role in this
process. After years of repression, the Uruguayan junta began to relax its vice grip on the
media and gave citizens the opportunity to vote on a constitution which would have
institutionalized the military’s control of the country. This set in motion a cycle of protest
and repression which was in part fueled by an emerging independent media. In the end,
Uruguay became a consolidated democracy.
222
Like Uruguay, Mexico eventually democratized, but from 1981 to 1995, Mexico
remained autocratic and government violation of physical integrity rights persisted in
spite of an increasingly independent and critical news media. At the same time the media
became more influenced by the Western professional model of journalism, and they
became increasingly critical of government in the early 1990s. During this same period,
there was an increase in anti-government demonstrations, though not nearly as many as
were reported in Uruguay.
Table 45: Summary of Characteristics of Four Non-Democracies with Media Freedom
Beginning
Regime
Type
Beginning
Repression
Media Protest Ending
Regime
Type
Ending
Repression
Summary
Uruguay
1980-1985
Military
Dictator-ship
1980 Polity
score=-7
Widespread
political
imprisonment &
torture
Controlled, but
emerging
independent &
partisan media; free
after 1984
1982-1984
General strike, riots
and anti-
government
demonstrations
Democratic
1985 Polity
score=9
2007 Polity
score=10
Full respect for
all physical
integrity rights
except torture
Increased media freedom and
protest preceded
democratization in Uruguay;
with democratization respect for
human rights improved
Mexico
1981-1995
Autocratic
1981 Polity
score=-3
Widespread
political killing,
disappearance,
political
imprisonment
and some torture
Somewhat free;
increasingly critical;
heavily influenced by
Western profess-
sionalism
General strike 1989;
Anti-government
demonstrations in
1993-1995
Becoming
Democratic
1995 Polity
score=4
2007 Polity
score=8
Widespread
political killing
and torture, some
disappearance
and political
imprisonment
Repression and some protest
persisted as media became more
critical of government abuse;
Mexico did eventually
democratize, but government is
undermined by drug cartels
Uganda
1981-1995
Autocratic
1981 Polity
score=3
Widespread
political killing,
disappearance,
political
imprisonment
and torture
Somewhat free;
intensely partisan
No reported strikes,
riots or
demonstrations
Autocratic
1995 Polity
score=-4
2007 Polity
score=-1
Widespread
torture, some
political killing
and political
imprisonment; no
disappearance
Intensely partisan media
remained mostly free, repression
remained widespread, but little
protest was reported; Uganda
remains autocratic
Burkina
Faso
1981-1995
Autocratic
1981 Polity
score=-7
Some political
killing and
political
imprisonment;
no torture or
disappearance
Media became
mostly free in 1991;
mix of state owned
and private
General strike in
1994;
few reported anti-
government
demonstrations
Autocratic
1995 Polity
score=-5
2007 Polity
score =0
Some political
killing, political
imprisonment and
torture; no
disappearance
Media became mostly free,
repression persisted but was not
widespread, some protest was
reported, but was not
widespread; Burkina Faso
remains autocratic, though less
so than in 1981
223
224
In contrast, there were no protests reported in Uganda from 1981 to 1995 (Banks
2008).
46
This study has focused on the short term effects of media freedom and
autocracy/democracy on human rights. Although these results indicate that media
freedom does not make life better, at least in terms of human rights in autocratic states in
the short run, we do not yet know the long term effects of media freedom on
democratization and corruption. Clearly, the interactions between free media and
government in these non-democracies indicate that the long term effects of media
freedom merit further investigation. It could be that the short time effects identified here
are part of a longer transition. Perhaps the cycle of protest and repression that is
associated with independent media in non-democracies is essential to bring about positive
political change. Maybe the dissent and deterioration in human rights will eventually lead
Yet the media, which was strongly partisan, persisted in criticizing the
government and repression remained widespread. One reason for the persistent repression
in Uganda is the on-going internal conflict from rebel groups. In fact, much of the
repression in Uganda is carried out by rebel groups, though government abuses are
common too. Unlike Mexico and Uruguay, Uganda has not democratized. Similarly
Burkina Faso has not democratized in spite of having a functionally free media. Like
Uganda, Burkina Faso has experienced several coups, but its government has more
respect for physical integrity rights. There has been some protest reported in Burkina
Faso, but not nearly as much as was reported in Uruguay and Mexico.
46
This may be due to this dataset’s reliance on New York Times coverage to code for domestic conflict.
However during the same time period the same dataset did record protests in Burkina Faso, which is also in
Sub-Saharan Africa (Banks 2008).
225
to democratization and improved human rights. Yet, the cases of Uganda and Burkina
Faso suggest that some states are not on a trajectory toward democracy.
New Directions for Research
This project has focused on the more immediate effects of media freedom and
autocracy/democracy on protest and human rights. As more data become available, it will
be important to consider the long term effects of media freedom in the repression-protest
nexus. It could be that over time, the opportunity that media freedom provides to
mobilize protest could help to bring about reform and improved government respect for
human rights. Similarly, it will be important to investigate the long term effects of media
freedom on democracy. Policy makers have argued that media freedom could have a
democratizing effect, but this has yet to be tested. We do not yet know the long term
effects of media freedom in a non-democratic setting.
Former World Bank President James D. Wolfensohn (1999) argued that free
media are “at the absolute core of equitable development, because if you cannot
enfranchise poor people, if they do not have a right to expression, if there is no
searchlight on corruption and inequitable practices, you cannot build the public consensus
needed to bring about change.” Yet, we know very little about the effect of free media on
corruption in developing, democratizing and non-democratic countries. Therefore, it
would be worthwhile to study the effect of media freedom on government corruption and
good governance across regime types.
226
Although this study focused on the effect of media free and democracy on protest
and repression, the argument that protest movements are shaped by a combination of
threats and opportunities and the conceptualization of repression as a threat and media
freedom as an opportunity suggest that there might be an important interactive effect of
media freedom and repression on dissent. Thus, this is another avenue for future research.
In general, the role of media in protest movements remains underexplored. The potential
of media freedom as an opportunity should certainly be considered.
The contrast between the “professionally” motivated media in Mexico and the
partisan motivated media in Uganda points to the need to consider how different models
of journalism might shape media influence. Do media that follow the professional model
and aspire to be objective enjoy more credibility? Or do partisan media that freely
espouse their affiliations seem more honest? Do audiences care? Certainly a comparative
study of the effects of different types of media systems and their credibility would help to
shed some light on media influence.
While broad statistical analyses help to identify trends and effects across countries
and over time, in-depth case studies would help to identify the connections between
independent media, protest movements and repression. The brief case studies included in
this project point to the need to consider the role of journalistic agency. Why is it that
journalists are sometimes willing to risk their lives to perform a watchdog role in
environments that are decidedly dangerous and at other times not willing to go out on a
limb in environments that are relatively safe? These are questions that are best addressed
with qualitative analysis.
227
Finally, what effect will new media have on media freedom and media influence
on government behavior? Will new media thwart government efforts to control political
communication or will the tried and true methods of repression work with new media as
they have with traditional media? Certainly the Internet has facilitated all types of
communication, and much has been made of the effect that this might have on
transnational activism (Keck and Sikkink 1998). Yet, we do not know if this will have an
effect on how governments treat citizens within their borders. According to the
Committee to Protect Journalists, “governments are quickly learning how to turn
technology against bloggers by censoring and filtering the Internet, restricting online
access and mining personal data. When all else fails, the authorities simply jail a few
bloggers to intimidate the rest of the online community,” (Committee to Protect
Journalists 2009). Thus, studies comparing the effects of media freedom before and after
the Internet gained popularity would help us to better understand the influence of new
media.
Conclusion
This project offers both theoretical implications and pragmatic application. First
of all in introducing a conditional theory of the effect of media freedom on protest and
human rights, I argue and provide evidence that the effect of media freedom varies
depending of the presence or absence of democratic characteristics. The general
assumption has been that free media would have the same effect, regardless of regime
characteristics. The findings of this study suggest that in non-democracies free media can
228
actually make life worse. This is counterintuitive until we consider how the media work
in conjunction with democratic institutions because the role of the media is always
intervening.
When the independent media shine a spotlight on government bad behavior, it is
then up to the people and the government to decide what to do about it. In a consolidated
democracy, the rules of the game are clearly drawn. People know how they can
participate in and influence government. They can complain, they can petition, they can
vote. Also, the executive’s decision-making power is constrained. If he or she makes a
mistake, there are usually any number of competing political elites who will use that error
to their advantage. In short, democratic governments are vulnerable to public opinion. In
the absence of the constraints of democratic institutions, when the independent media
draw attention to government misdeeds, there are no clearly defined mechanisms for
people to use to bring the government in line; there are no rules of the game. The people
are armed with information, but their target seems impenetrable. Overtime, this bottled
up discontent can explode into protest.
This project also has policy implications. Specifically, the findings here suggest
that advocating media freedom in non-democracies may not have the intended effects. In
fact, doing so could lead to increased protest and repression. Instead, the results of this
study suggest that media freedom needs to be introduced in conjunction with democratic
institutions, especially the establishment of laws regarding political participation. The
long term effects of media freedom in non-democracies are unknown. It may in fact have
a democratizing effect; however, this study indicates that in the short term, media
229
freedom in autocratic countries can fuel dissent and, consequently, government abuse of
citizens. Finally, policy makers should consider the potential repercussions of
encouraging journalists in non-democracies to follow journalistic practices that work in
democracies. The consequences could be dire for the journalists. In fact, foisting media
freedom on an autocratic state puts journalists on the frontline without weapons. They are
trained to report critically about government behavior, but if government is not
vulnerable to public opinion, their criticism is far more likely to provoke retaliation than
it is to bring about reform.
230
References
(1982). Uruguay's First Public Political Demonstration since 1973. BBC summary of world
broadcasts.
(1984). Uruguay Breaks up Plant Sit-In. New York Times.
(1984). Workers Shut Down Capital with First General Strike in 11 Years. Associated Press.
(1994). "Update." Africa report 39(3): 12.
(1995). Burkina Faso Human Rights Practices, 1994, United States Department of State.
(1996). World Press Freedom Review 1996. IPI Report. Zurich, International Press Institute.
(2000). Spore: Information for Agricultural Development in Acp Countries. Wageningen, The
Netherlands, Technical Centre for Agricultural and Rural Cooperation (CTA).
(2003). Demographic and Health Surveys, United States Agency for International Development.
(2009). Background Note: Burkina Faso, U.S. Department of State Bureau of African Affairs.
Achen, C. H. (1982). Interpreting and Using Regression. Quantitative applications in the social
sciences. Beverly Hills, CA, Sage.
Achen, C. H. (2000). Why Lagged Dependent Variables Can Suppress the Explanatory Power of
Other Independent Variables. Annual Meeting of the Political Methodology Society. Los
Angeles.
Althaus, S. L. (2003). "When News Norms Collide, Follow the Lead: New Evidence for Press
Independence." Political Communication 20: 381-414.
Amnesty International (2006). Press Freedom: Journalists in Need of Protection, Amnesty
International.
Araujo, G. (1984). "How I Run Radio Cx-30." Index on censorship 13(5): 14-15.
Assaad, F. (1982). "The Sexual Mutilation of Women." Worlth health forum 3(4): 391-394.
Bagdikian, B. H. (1987). The Media Monopoly. Boston, Beacon Press.
Banks, A. S. (2008). Cross-National Time-Series Data Archive, Databanks International.
Beck, N. and J. N. Katz (1995). "What to Do (and Not to Do) with Times-Series-Cross-Section
Data." American Political Science Review 89: 634-647.
231
Benford, R. and D. A. Snow (2000). "Framing Processes and Social Movements: An Overview and
Assessment." Annual Review of Sociology 26: 611-639.
Bennett, W. L. (1990). "Toward a Theory of Press-State Relations in the United States." Journal
of Communication 40(2): 103-125.
Bennett, W. L., R. G. Lawrence, et al. (2007). When the Press Fails. Chicago and London, The
University of Chicago Press.
Brambor, T., W. R. Clark, et al. (2006). "Understanding Interaction Models: Improving Empirical
Analyses." Political Analysis 14: 63-82.
Buckley, S., K. Duer, et al. (2008). Broadcasting, Voice, and Accountability: A Public Interest
Approach to Policy, Law, and Regulation. Ann Arbor, Michigan, University of Michigan
Press.
Bueno de Mesquita, B., G. W. Downs, et al. (2005). "Thinking inside the Box: A Closer Look at
Democracy and Human Rights." International Studies Quarterly 49(3): 439-457.
Bueno de Mesquita, B., A. Smith, et al. (2003). Logic of Political Survival Cambridge, Mass., MIT
Press.
Bunch, C. (1990). "Women's Rights as Human Rights: Toward a Re-Vision of Human Rights."
Human Rights Quarterly 12(4): 486-498.
Burns, J. (1982). Uruguay: Voters Cast Ballots for 'End to Military Dictatorship'. Christian Science
monitor. Boston.
Central Intelligence Agency (2009). World Factbook, Central Intelligence Agency.
Christian, S. (1990). Freer Economy Resisted in Uruguay. New York Times. New York.
CIMA (2008). Empowering Independent Media: U.S. Efforts to Foster Free and Independent
News around the World. Washington, Center for international media
assistance/National endowment for democracy.
Cingranelli, D. L. and D. L. Richards (2006). The Cingranelli-Richards (Ciri) Human Rights Dataset.
Cingranelli, D. L. and D. L. Richards (2008). Cingranelli-Richards Human Rights Data Project
Coding Manual.
Coad, M. (1984). Uruguay Queues for Poll/First Election for 11 Years to End Military
Dictatorship. Guardian. London.
Coleman, I. (2004). "The Payoff Women's Rights." Foreign Affairs.
232
Collelo, T. (1987). Syria: A Country Study. Country studies. L. R. Mortimer. Washington, DC,
Library of Congress Federal Research Division.
Committee to Protect Journalists (2007). Azerbaijani President Pardons Five Journalists; Three
Left Behind Bars. New York.
Committee to Protect Journalists (2009). 10 Worst Countries to Be a Blogger, Committee to
Protect Journalists.
Committee to Protect Journalists (2009). Attacks on the Press in 2008: Azerbaijan. Attacks on
the press.
Committee to Protect Journalists. (2009, April 3, 2009). "Journalists Killed." Retrieved June 5,
2009, 2009, from www.cpj.org/deadly.
Dahl, R. A. (1998). On Democracy. New Haven & London, Yale University Press.
Davenport, C. and D. A. Armstrong (2004). "Democracy and the Violation of Human Rights: A
Statistical Analysis from 1976 to 1996." American Journal of Political Science 48(3): 538-
554.
Diamond, L., Ed. (1992). The Democratic Revolution: Struggles for Freedom and Pluralism in the
Developing World. New York, Freedom House.
Diehl, J. (1983). Uruguay's Junta Split on Surrendering Power. Washington post. Washington.
Earl, J. (2003). "Tanks, Tear Gas, and Taxes: Toward a Theory of Movement Repression."
Sociological Theory 21(1): 44-68.
Egorov, G., S. M. Guriev, et al. (2008). Media Freedom in Dictatorships, SSRN.
Ellerson, B., Ed. (2000). Sisters of the Screen: Women of Africa on Film Video and Television.
Trenton, NJ, Africa World Press.
Englebert, P. (1999). Burkina Faso: Unsteady Statehood in West Africa. Boulder, Colorado,
Westview Press.
Entman, R. M. (2004). Projections of Power. Chicago, University of Chicago Press.
Fein, H. (1995). "More Murder in the Middle: Life-Integrity Violations and Democracy in the
World, 1987." Human Rights Quarterly 17(1): 170-191.
Freedom House (2002). Freedom of the Press 2002 Survey, Freedom House.
Freedom House (2007). Freedom in the World, Freedom House.
Freedom House (2007). Freedom of the Press 2007: A Global Survey of Media Independence,
Freedom House.
233
Freedom House (2009). Freedom of the Press 2009 Survey, Freedom House.
Frontiers in Reproductive Health (2008). Burkina Faso: Political Will, Law Enforcement, and
Educational Campaigns Appear to Be Reducing Fgm/C. Washington DC, Frontiers in
Reproductive Health/Population Council.
Gamson, W. A. and G. Wolfsfeld (1993). "Movements and Media as Interacting Systems." Annals
of the American Academy of Political and Social Science 526: 114-127.
Gans, H. J. (1979). Deciding What's News: A Study of Cbs Evening News, Nbc Nightly News,
Newsweek and Time. New York, Random House.
Geekie, R. (1992). "Interview: Moustapha Thiombiano, African Radio Pioneer." Africa report
37(1): 10.
Gillespie, C. G. (1985). "Uruguay's Return to Democracy." Bulletin of Latin American research
4(2): 99-107.
Gleditsch, K. S. and M. D. Ward (1997). "Double Take: A Reexamination of Democracy and
Autocracy in Modern Politics." The journal of conflict resolution 41(3): 361-383.
Gleditsch, N. P., M. Eriksson, et al. (2002). "Armed Conflict 1946-2001: A New Dataset." Journal
of Peace Research 39(5): 615–637(5): 615-637.
Goldstone, J. A. and C. Tilly (2001). Threat (and Opportunity): Popular Action and State Response
in the Dynamics of Contentious Action. Silence and Voice in the Study of Contentious
Politics. R. R. Aminzade, J. A. Goldstone, D. McAdamet al. Cambridge, Cambridge
University Press.
Gomez, J. (2005). Freedom of Expression and the Media in Singapore. London, Article 19.
Goodsell, J. N. (1980). Uruguayan Vote: A Slap in Face to Military Rulers. Christian Science
monitor. Boston.
Goodsell, J. N. (1983). Once-Idyllic Uruguay Beset by Economic, Political Troubles. Christian
Science monitor. Boston.
Gorney, C. (1981). In Uruguay, the Junta Is Puzzled and Its Opponents Pleased; Plebiscite Tally
Jolts Uruguay's Ruling Military. Washington Post. Washington.
Gurevitch, M. and J. G. Blumler (1990). Political Communication Systems and Democratic Values.
Democracy and the Mass Media. J. Lichtenberg. Cambridge, Cambridge University Press.
Hallin, D. C. (1986). The "Uncensored War" the Media and Vietnam. Berkeley, University of
California Press.
234
Hallin, D. C. and P. Mancini (2004). Comparing Media Systems: Three Models of Media and
Politics. Cambridge, Cambridge University Press.
Hamilton, J. T. (2004). All the News That's Fit to Sell: How the Market Transforms Information
into News. Princeton, NJ, Princeton University Press.
Hatcher, J., L. Meggiolaro, et al. (2005). Cultivating Women's Rights for Access to Land. Rome,
ActionAid International, International Food Security Network.
Heinz, W. S. and H. Fruhling (1999). Determinants of Gross Human Rights Violations by State and
State-Sponsored Actors in Brazil, Uruguay, Chile and Argentina, 1960-1990. The Hague,
Martinus Nijhoff Publishers.
Herman, E. S. and N. Chomsky (1988). Manufacturing Consent: The Political Economy of the
Mass Media. New York, Pantheon.
Heston, A., R. Summers, et al. (2002). Penn World Table Version 6.1, Center for International
Comparisons at the University of Pennsylvania (CICUP).
Hicks, E. S. (1999). As Fragile as a Crystal Glass: Press Freedom in Iran, Human Rights Watch.
Hudson, R. A. and S. W. Meditz (1990). Uruguay: A Country Study. Country studies/area
handbook program. L. R. Mortimer, Federal Research Division Library of Congress.
Hughes, S. (2006). Newsrooms in Conflict: Journalism and the Democratization of Mexico.
Pittsburgh, University of Pittsburgh Press.
Human Rights Watch (1998). Shielded from Justice: Police Brutality and Accountability in the
United States, Human Rights Watch.
Human Rights Watch (1999). Human Rights Watch World Report 1999, Human Rights Watch.
Inglehart, R. and P. Norris (2003). Rising Tide: Gender Equality and Cultural Change around the
World. Cambridge, Cambridge University Press.
Inglehart, R., P. Norris, et al. (2002). "Gender Equality and Democracy." Comparative Sociology
1(2-3): 321-345.
IRIN (2005). In-Depth: Razor's Edge--the Controversy of Female Genital Mutilation, United
Nations Office for the Coordination of Humanitarian Affairs.
IRIN (2007). Burkino Faso: Humanitarian Country Profile, United Nations office for the
coordination of humanitarian affairs.
IRIN. (2009). ""Sri Lanka: "Attacks on Media Must Stop"." Retrieved 1/24/09, 2009, from
http://www.irinnews.org/Report.aspx?ReportId=82356
235
IRIN (2009). West Africa: "Pleasure Hospital" under Construction for Fgm/C Victims, IRIN.
Jacobs, R. N. (1996). "Civil Society and Crisis: Culture, Discourse, and the Rodney King Beating."
The American journal of sociology 101(5): 1238-1272.
Jefferson, T. (1900, 2006). The Jeffersonian Cyclopedia, Funk & Wagnalls Company.
Jenkins, J. C. (1983). "Resource Mobilization Theory and the Study of Social Movements." Annual
Review of Sociology 9: 527-553.
Johnson, P. (2003). Amanpour: Cnn Practiced Self-Censorship. USA Today.
Karlekar, K. D. (2009). Press Freedom in 2008: Restrictive Laws and Physical Attacks Fuel Further
Declines. Freedom of the press, Freedom House.
Keck, M. E. and K. Sikkink (1998). Activists Beyond Borders. Ithaca, Cornell University Press.
Keech, W. (forthcoming). Economic Politics in Latin America: Rethinking Democracy and
Dictatorship. Cambridge, Cambridge University Press.
King, G., M. Tomz, et al. (2000). "Making the Most of Statistical Analyses: Improving
Interpretation and Presentation." American Journal of Political Science 44(2): 341-355.
King, G. and L. Zeng (1999). Logistic Regression in Rare Events Data, Harvard University,
Department of Government.
Kouba, L. J. and J. Muasher (1985). "Female Circumcision in Africa: An Overview." African studies
review 28(1): 95-110.
Kovach, B. and T. Rosenstiel (2001 ). The Elements of Journalism: What Newspeople Should
Know and the Public Should Expect. New York, Random House.
Lamizana, M. (1997). The Campaign against Excision: Burkina Faso's Experience. Washington DC
and New York John's Hopkin's University/Center for Communication Programs.
Landman, T. (2006). Studying Human Rights. London & New York, Routledge.
Lardner, T. (1991). "The Press Is Free, but Will Burkinabe Buy It?" Africa report 36(5): 10.
Lawson, C. (2002). Building the Fourth Estate: Democratization and the Rise of a Free Press in
Mexico. Ewing, NJ, University of California Press.
Linz, J. and A. Stepan (1996). Problems of Democratic Transition and Consolidation. Baltimore,
Johns Hopkins University Press.
Livingston, S. (1997). Clarifying the Cnn Effect: An Examination of Media Effects According to the
Type of Military Intervention, Joan Shorenstein Center for Press, Politics and Public
Policy.
236
Macur Olson, J. (1965). The Logic of Collective Action: Public Goods and the Theory of Groups.
Cambridge, Harvard University Press.
Marshall, M. G. and K. Jaggers (2002). Polity Iv Project Political Regime Characteristics and
Transitions, 1800-2002 Dataset Users' Manual Integrated Network for Societal Conflict
Research Program
Center for International Development and Conflict Management.
Marshall, M. G. and K. Jaggers (2007). Polity Iv Country Reports 2007, Center for Systemic Peace.
Marshall, M. G. and K. Jaggers (2009). Polity Iv Project: Political Regime Characteristics and
Transitions, 1800-2007 Dataset Users' Manual, Center for Systematic Peace.
Marshall, M. G., K. Jaggers, et al. (2003). Polity Iv Project, Center for International Development
and Conflict Management at the University of Maryland.
McAdam, D. (1982). Political Process and the Development of Black Insurgency, 1930-1970.
Chicago, University of Chicago Press.
McAdam, D., S. Tarrow, et al. (2001). Dynamics of Contention. Cambridge, Cambridge University
Press.
McCarthy, J. D. and M. N. Zald (1977). "Resource Mobilization and Social Movements: A Partial
Theory." American Journal of Sociology 82(6): 1212-1241.
McCormick, J. M. and N. J. Mitchell (1997). "Human Rights Violations, Umbrella Concepts, and
Empirical Analysis." World Politics 49(4): 510-525.
McQuail, D., D. A. Graber, et al. (2008). Conclusion: Contemporary Challenges in Journalism and
Democracy. The Politics of News the News of Politics. D. A. Graber, D. McQuail and P.
Norris. Washington DC, CQ Press.
Moe, T. M. (2006). "Political Control and the Power of the Agent." Journal of law, economics,
and organization 22(1): 1-29.
Most, B. A. and H. Starr (1993). Inquiry, Logic and International Politics. Columbia, SC, University
of South Carolina Press.
Mueller, J. (1992). "Democracy and Ralph's Pretty Good Grocery: Elections, Equality, and the
Minimal Human Being." American Journal of Political Science 36(4): 983-1003.
Munck, G. L. and J. Verkuilen (2002). "Conceptualizing and Measuring Democracy: Evaluating
Alternative Indices." Comparative Political Studies 31(1): 5-34.
Nam, T. (2006). "What You Use Matters: Coding Protest Data." PS: Political science & politics
39(2): 281-287.
237
Norris, P. (2005). Stm103 Global Indicators Shared Dataset V2.0.
O'Connor, M. (2009). "And Then They Came for Me: Journalism under the Gun in Sri Lanka."
Columbia Journalism Review.
O'Heffernan, P. (1994). A Mutual Exploitation Model of Media Influence in U.S. Foreign Policy.
Taken by Storm: The Media, Public Opinion, and U.S. Foreign Policy in the Gulf. W. L.
Bennett and D. L. Paletz. Chicago, University of Chicago Press.
O'Neill, K. (2004). "Transnational Protest: States, Circuses and Conflict at the Frontline of Global
Politics." International Studies Review 6: 233-251.
Ocitti, J. (2005). Press Politics and Public Policy in Uganda. Lewiston, NY, Edwin Mellen Press.
OpenNet Initiative (2007). Country Reports, OpenNet Initiative.
Parenti, M. (1986). Inventing Reality: The Politics of Mass Media. New York, St. Martin's Press.
Pew Research Center for the People and the Press (2000). Self Censorship: How Often and Why.
Pew Research Center for the People and the Press web site, Pew Research Center for
the People and the Press. 2005: A Survey of Journalists in Association with Columbia
Journalism Review.
Poe, S. C. and C. N. Tate (1994). "Repression of Human Rights to Personal Integrity in the 1980s:
A Global Analysis." American Political Science Review 88(4): 853-872.
Poe, S. C., C. N. Tate, et al. (1999). "Repression of the Human Right to Personal Integrity
Revisited: A Global Cross-National Study Covering the Years 1976-1993." International
Studies Quarterly 43(2): 291-313.
Polity IV (2005). Polity Iv Project, Integrated Network for Societal Conflict Research (INSCR)
Program, Center for International Development and Conflict Management (CIDCM)
Randall, V. (1993). "Media and Democratisation in the Third World." Third World quarterly
14(3): 625-646.
Rasler, K. (1996). "Concessions, Repression, and Political Protest in the Iranian Revolution."
American Sociological Review 61(1): 132-152.
Reporters Sans Frontieres (2009). World Report 2009, Reporters Sans Frontieres.
Reporters Without Borders (2008). Press Freedom Index 2008, Reporters Without Borders.
Reporters Without Borders. (2008, December 30, 2008). "Press Freedom Round-up 2008."
Retrieved June 5, 2009, 2009, from http://www.rsf.org/article.php3?id_article=29797.
238
Riding, A. (1984). Uruguay's New Leaders Inherit a Bad Cash Flow. New York Times. New York
Schock, K. (1999). "People Power and Political Opportunities: Social Movement Mobilization and
Outcomes in the Philippines and Burma." Social Problems 46(3): 355-375.
Schumacher, E. (1980). Uruguay's Vote: Exception to the Rule. New York Times. New York.
Schumacher, E. (1984). Uruguayan Public Discontent Spurs Crackdown on Dissent. New York
Times. New York.
Scott, D. C. (1994). Mexican Indians Seize Towns and Demand Land in Challenge to Salinas.
Christian Science Monitor. Boston, MA.
Seib, P. (2008). The Al Jazeera Effect: How the New Global Media Are Reshaping World Politics,
Potomac Books, Inc. .
Shapiro, F. R. and J. Epstein (2006). The Yale Book of Quotations. New Haven, CT, Yale University
Press.
Small, M. and J. D. Singer (1982). Resort to Arms: International and Civil Wars, 1816-1980.
Beverly Hills, Sage.
Steele, J. (1984). From Dictarship to Democracy (Part 3): Flight of the Generals/Uruguay.
Guardian. London.
Steele, J. (2007). "The Voice of East Timor: Journalism, Ideology and the Struggle for
Independence." East Asian Studies Review 31(3): 261-282.
Stolz, J. (1998). Putting a Stop to the Excision in Burkina Faso. Le Monde diplomatique. October
1998.
Strand, H., J. Carlsen, et al. (2005). Armed Conflict Dataset Codebook Version 3-2005,
International Peace Research Institute, Oslo, Department of Peace and Conflict
Research, Uppsala University, Department of Sociology and Political
Science/Department of Geography Norwegian University of Science and Technology.
Tilly, C. (2004). "Contentious Choices." Theory and society 33(3/4): 473-481.
Tilly, C. (2004). Social Movements: 1768-2004. Boulder, Colorado, Paradigm Publishers.
Tomz, M., G. King, et al. (1999 ). Relogit: Rare Events Logistic Regression, Version 1.1.
Cambridge, MA, Harvard University.
Tomz, M., J. Wittenberg, et al. (2003). Clarify: Software for Interpreting and Presenting
Statistical Results.
239
Triendregeogon, A. (1982). Female Circumcision in Upper Volta. Traditional Practices Affecting
the Health of Women and Children. T. Baasher, R. H. Bannerman, H. Rushwan and I.
Sharaf. Alexandria, Egypt, World Health Organization, Regional Office for the Eastern
Mediterranean. Technical Publication No. 2 Vol. 2.
Unicef (2005). K.S, 14, Student, Victim of Female Genital Mutilation at the Age of 9. Real Lives.
United States Department of State (2001). Burkina Faso: Report on Female Genital Mutilation
(Fgm) or Female Genital Cutting (Fgc), United States Department of State.
USAID (2007). Azerbaijani Media Students Learn to Ask Tough Questions. Success stories, United
States Agency of International Development.
USAID (2007). Program Boosts Tv News Production, Educates Viewers on Local Issues. Success
Stories, United States Agency for International Development.
Van Belle, D. A. (1997). "Press Freedom and the Democratic Peace." Journal of Peace Research
34(4): 405-414.
Van Belle, D. A. (2000). Press Freedom and Global Politics. Westport, CT, Praeger Publishers.
Van Belle, D. A. (2002). Press Freedom and Peace: Theory and Findings. Power and Conflict in
the Age of Transparency. B. I. Finel and K. M. Lord. New York, Palgrave Macmillan.
VanBelle, D. A. (2000). Press Freedom and Global Politics. Westport, CT, Praeger Publishers.
Vanderwerff, M. (2008). New Media and International Media Development: A Resource Guide
for Europe and Eurasia, United State Agency for International Development.
Waisbord, S. (2000). Watchdog Journalism in South America. New York, Columbia University
Press.
Wickrematunge, L. (2009). And Then They Came for Me. The Sunday Leader. Colombo.
Wolfensohn, J. D. (1999). Voices of the Poor. Washington Post. Washington.
World Bank (2009). Freedom of Information Necessary for Sri Lanka's Development. Colombo,
Sri Lanka, World Bank.
World Health Organization (2008). Female Genital Mutilation Fact Sheet. Geneva, Switzerland,
World Health Organization.
Zakaria, F. (1997). "The Rise of Illiberal Democracy." Foreign Affairs 76(6): 22-43.
Zakaria, F. (2003). The Future of Freedom: Illiberal Democracy at Home and Abroad. New York,
W.W. Norton & Company.
240
Appendix: Tables and Figures from Preliminary Analyses
Table 46: Media Freedom and Levels of Democracy/Autocracy 1948-1995, N=5575*
Polity Score Government-Controlled Media Free Media Total
-10 187 (93%) 15 (7%) 202
-9 588 (99%) 8 (1%) 596
-8 321 (100%) 1 (0%) 322
-7 1220 (96%) 48 (4%) 1268
-6 182 (74%) 64 (26%) 246
-5 145 (90%) 17 (10%) 162
-4 84 (97%) 3 (3%) 87
-3 99 (72%) 38 (28%) 137
-2 66 (87%) 10 (13%) 76
-1 81 (71%) 33 (29%) 114
0 117 (82%) 26 (18%) 143
1 44 (96%) 2 (4%) 46
2 38 (46%) 44 (54%) 82
3 35 (78%) 10 (22%) 45
4 117 (82%) 26 (18%) 143
5 52 (40%) 78 (60%) 130
6 76 (61%) 49 (39%) 125
7 96 (55%) 80 (40%) 176
8 60 (23%) 203 (77%) 263
9 41 (20%) 161 (80%) 202
10 40 (4%) 970 (96%) 1010
Total 3689 (66%) 1886 (34%) 5575
*This Table is based on data from the Global Press Freedom Dataset (Van Belle
2000) and the Polity IV Dataset (Marshall et al. 2003)
241
Table 47: Media Freedom and Levels of Democracy/Autocracy 1981-1995, N=1395*
Polity Score Government-Controlled Media Free Media Total
-10 3 (100%) 0 (0%) 3
-9 71 (100%) 0 (0%) 71
-8 69 (100%) 0 (0%) 69
-7 234 (98%) 5 (2%) 239
-6 75 (89%) 9 (11%) 84
-5 33 (79%) 9 (21%) 42
-4 17 (100%) 0 (0%) 17
-3 9 (56%) 7 (44%) 16
-2 27 (75%) 9 (25%) 36
-1 32 (100%) 0 (0%) 32
0 3 (33%) 6 (67%) 9
1 8 (80%) 2 (20%) 10
2 9 (75%) 3 (25%) 12
3 14 (100%) 0 (0%) 14
4 26 (93%) 2 (7%) 28
5 38 (72%) 15 (28%) 53
6 47 (71%) 19 (29%) 66
7 24 (46%) 28 (54%) 52
8 20 (22%) 73 (78%) 93
9 12 (12%) 90 (88%) 102
10 30 (9%) 317 (91%) 347
Total 801 (57%) 594 (43%) 1395
*This Table is based on data from Douglas Van Belle’s Press Freedom Dataset
(2000) and the Polity IV Dataset (Marshall 2002)
242
Table 48: Binary Decomposition Results
Variables Coefficients
Autocracy/Democracy=-9 -.003
Autocracy/Democracy=-8 -.492
Autocracy/Democracy=-7 -.093
Autocracy/Democracy=-6 -.934
Autocracy/Democracy=-5 -.4644
Autocracy/Democracy=-4 -.322
Autocracy/Democracy=-3 -2.257**
Autocracy/Democracy=-2 .305
Autocracy/Democracy=-1 -.021
Autocracy/Democracy=0 -2.606**
Autocracy/Democracy=1 -.964
Autocracy/Democracy=2 .008
Autocracy/Democracy=3 -.906
Autocracy/Democracy=4 -.621
Autocracy/Democracy=5 -.680
Autocracy/Democracy=6 -.280
Autocracy/Democracy=7 .692
Autocracy/Democracy=8 -.302
Autocracy/Democracy=9 .174
Autocracy/Democracy=10 1.909**
Media Freedom .111
GDP/Capita (logged) .211***
Population (logged) -.287***
International Conflict -.068
Intrastate Conflict -2.33***
Constant 6.138***
*p < 0.05; **p < .01; ***p < .001
Table 49: The Effect of Media Freedom and Components of Democracy on the Right to be Free from Torture
N=1036
Dependent Variable:
Torture (current)
Model
Coefficient
(Standard Error)
Model 1
Coefficient
(Standard Error)
Model
Coefficient
(Standard Error)
Model 2
Coefficient
(Standard Error)
Torture (-1 year) 1.918 (.134)*** 1.890 (.134)*** 1.818 (.133)*** 1.808 (.134)***
Media Freedom -.094 (.212) -.865 (.446) .035 (.209) -.765 (.329)*
Autocracy/Democracy .177 (.086)* .075 (.076) --- ---
Media
Freedom*Autocracy/Democracy
--- .338 (.144)* --- ---
Executive Constraints --- --- .567 (.343) .440 (.323)
Regulation of Participation --- --- 1.346 (.337)*** .424 (.363)
Media Freedom*Reg. of
Participation
--- --- --- 1.347 (.429)**
GDP/capita (logged) .368 (.115)** .309 (.118)** .287 (.114)* .263 (113)*
Population (logged) -.185 (.055)** -.185 (.055)** -.234 (.057)*** -.241 (.056)***
Interstate Conflict -.300 (.424) -.326 (.422) -.213 (.411) -.220 (.410)
Intrastate Conflict -.917 (.286)** -.916 (.268)** -.798 (.257)** -.789 (255)**
Wald χ
2
407.77 433.05 451.81 499.12
McKelvey & Zavoina’s R
2
.57 .58 .58 .59
Ordered Logit Analyses *p<.05 ** p<.01 ***p<.001
243
Table 50: The Effect of Media Freedom and Democracy Components on the Right to be Free from
Torture Five Years Later N=1110
Dependent Variable:
Torture (+5 years)
Model
Coefficient
(Standard Error)
Model 1
Coefficient
(Standard Error)
Model
Coefficient
(Standard Error)
Model 2
Coefficient
(Standard Error)
Torture (current) 1.118 (.141)*** 1.077 (.135)*** 1.035 (.140)*** 1.007 (.143)***
Media Freedom -.238 (.287) -1.417 (.704)* -.091 (.266) -1.268 (.516)*
Autocracy/Democracy .218 (.123) .069 (.111) --- ---
Media Freedom*
Autocracy/Democracy
--- .504 (.210)* --- ---
Executive Constraints --- --- .653 (.428) .485 (.395)
Regulation of Participation --- --- 1.202 (.483)* -.145 (.526)
Media Freedom*Reg. of
Participation
--- --- --- 1.927(.657)**
GDP/capita (logged) .623 (.160)*** .535 (.161)** .570 (.158)*** .553 (.158)***
Population (logged) -.284 (.063)*** -.284 (.062)*** -.324 (.068)*** -.341 (.068)***
Interstate Conflict -.561 (.324) -.598 (.321) -.532 (.342) -.523 (.329)
Intrastate Conflict -.940 (.327)** -.925 (.300)** -.832 (.294)** -.825 (.288)**
Wald χ
2
228.27 268.79 256.60 275.67
McKelvey & Zavoina’s R
2
.50 .51 .51 .52
Ordered Logit Analyses *p<.05 ** p<.01 ***p<.001
244
Table 51: The Effect of Media Freedom and Democracy Components on the Right to be Free from
Political Imprisonment (Massive Political Imprisonment is the Base Outcome ) N=1036
Dependent Variable:
Political Imprisonment
Model
Coefficient (St.Error)
Model 1
Coefficient (St.Error)
Model
Coefficient (St.Error)
Model 2
Coefficient (St.Error)
Some Political Imprisonment
Political Imprisonment (-1 year) 1.475 (.176)*** 1.464 (.185)*** 1.498 (.179)*** 1.422 (.172)***
Media Freedom .216 (.307) -.052 (.460) .321 (.269) -.825 (.579)
Autocracy/Democracy .147 (.130) .116 (.149) --- ---
Media Freedom*Autocracy/Democracy --- .141 (.236) --- ---
Executive Constraints --- --- .266 (.392) .153 (.373)
Regulation of Participation --- --- -.047 (.586) -1.008 (.791)
Media Freedom*Reg. of Participation --- --- --- 2.294 (.928)*
GDP/capita (logged) -.130 (.151) -.136 (.152) -.069 (.158) -.109 (.162)
Population (logged) -.157 (.075)* -.156 (.075)* -.154 (.076)* -.177 (.078)*
Interstate Conflict -1.281 (.705) -1.307 (.723) -1.196 (.708) -1.240 (.691)
Intrastate Conflict -.858 (.308)** -.870 (.303)** -.840 (307)** -.828 (.333)*
Constant 1.703 (1.199) 1.769 (1.209) 1.239 (1.163) 2.446 (1.289)
No Political Imprisonment
Political Imprisonment (-1 year) 3.195 (.253)*** 3.157 (.273)*** 3.240 (.268)*** 3.100 (.292)***
Media Freedom .217 (.517) -.541 (1.222) .407 (.461) -1.994 (.711)**
Autocracy/Democracy .786 (.218)*** .694 (.226)** --- ---
Media Freedom*Autocracy/Democracy --- .342 (.443) --- ---
Executive Constraints --- --- 2.479 (.763)** 2.324 (.692)**
Regulation of Participation --- --- .900 (.656) -1.549 (.860)
Media Freedom*Reg. of Participation --- --- --- 4.383 (1.104)***
GDP/capita (logged) .190 (.174) .135 (.177) .264 (.185) .208 (.177)
Population (logged) -.378 (133)** -.376 (.133) -.391 (.138)** -.469 (.150)**
Interstate Conflict -.806 (.975) -.835 (.987) -.550 (1.057) -.585 (1.020)
Intrastate Conflict -2.477 (.494)*** -2.538 (.488)*** -2.450 (.512)*** -2.440 (.483)***
Constant -2.542 (1.693) -1.971 (1.864) -3.732 (1.666)* -.878 (2.031)
Wald χ
2
399.66 491.19 400.83 363.15
McFadden’s R
2
.46 .46 .46 .47
Multinomial Logit Analyses *p<.05 ** p<.01 ***p<.001
245
Table 52: The Effect of Media Freedom and Democracy Components on the Right to be Free from
Political Imprisonment Five Years Later (Massive Political Imprisonment is the Base Outcome) N=1110
Dependent Variable:
Political Imprisonment (+5 years)
Model
Coefficient (St. Error)
Model 1
Coefficient (St. Error)
Model
Coefficient (St. Error)
Model 2
Coefficient (St. Error)
Some Political Imprisonment
Political Imprisonment (current) .579 (.148)*** .585 (.154)*** .600 (.177)** .571 (.160)***
Media Freedom .169 (.382) .349 (.552) .136 (.360) -.184 (.827)
Autocracy/Democracy .220 (.144) .254 (.170) --- ---
Media Freedom*Autocracy/Democracy --- -.097 (.261) --- ---
Executive Constraints --- --- .946 (.572) .906 (.576)
Regulation of Participation --- --- .211 (.713) -.052 (1.095)
Media Freedom*Reg. of Participation --- --- --- .659 (1.268)
GDP/capita (logged) -.075 (.180) -.060 (.184) -.056 (.183) -.074 (.186)
Population (logged) -.344 (.096)*** -.342 (.096)*** -.361 (.106)** -.367 (.107)**
Interstate Conflict -.171 (.506) -.162 (.516) -.115 (.521) -.117 (.514)
Intrastate Conflict -.586 (.263)* -.588 (.267)* -.582 (.2270)* -.585 (.271)*
Constant 3.380 (1.412) 3.222 (1.494) 3.111 (1.378)* 3.505 (1.596)*
No Political Imprisonment
Political Imprisonment (current) 1.171 (.229)*** 1.130 (.230)*** 1.218 (.223)*** 1.108 (.212)***
Media Freedom .164 (.465) -.563 (.802) .249 (.469) -.972 (1.029)
Autocracy/Democracy .620 (.211)** .554 (.227)* --- ---
Media Freedom*Autocracy/Democracy --- .286 (.300) --- ---
Executive Constraints --- --- 2.276 (.711)** 2.178 (.689)**
Regulation of Participation --- --- 1.097 (.856) -.240 (1.186)
Media Freedom*Reg. of Participation --- --- --- 2.194 (1.611)
GDP/capita (logged) .556 (.252)** .487 (.256) .546 (.266)* .523 (.264)*
Population (logged) -.620 (.144)*** -.615 (.144)*** -.653 (.148)*** -.697 (.150)***
Interstate Conflict -.078 (.462) -.122 (.453) .073 (.492) .036 (.487)
Intrastate Conflict -1.222 (.639) -1.228 (.623)* -1.128 (.594) -1.143 (.558)*
Constant -.634 (1.031) .011 (2.124) -1.191 (1.993) .356 (2.481)
Wald χ
2
177.03 174.12 184.21 170.24
McFadden’s R
2
.32 .33 .32 .33
Multinomial Logit Analyses *p<.05 ** p<.01 ***p<.001
246
Table 53: The Effect of Media Freedom and Components of Democracy on Women’s Economic Rights
N=927
*p<.05; **p<.01;*** p<.001
Dependent Variable:
Women’s Economic Rights
Model 1 Model 2 Model 3
Women’s Economics Rights (-1 year) 2.848 (.277)*** 2.849 (.278)*** 2.834 (.281)***
Media Freedom -.342 (.252) -.954 (.579) -.882 (.512)
Executive Constraints .443 (.420) .243 (.444) .336 (.413)
Media Freedom*Executive Constraints --- .826 (.732) ---
Regulation of Participation .481 (.456) .357 (.471) -.039 (.606)
Regulation of Participation*Media Freedom --- --- .884 (.721)
Catholic .157 (.327) .167 (.332) .167 (.334)
Protestant -.004 (.331) -.029 (.338) -.020 (.338)
Muslim -.672 (.326)* -.678 (.327)* -.687 (.322)*
GDP/capita (logged) .527 (.173)** .523 (.171)** .506 (.173)**
Population (logged -.109 (.069) -.101 (.070) -.105 (.070)
Interstate Conflict -.663 (.417) -.691 (.422) -.656 (.405)
Internal Conflict -.514 (.233)* -.526 (.239)* 531 (.228)*
Wald χ
2
203.76 213.41 220.54
BIC -5176.281 -5170.482 -5171.043
McKelvey & Zavoina’s R
2
.65 .65 .66
247
Table 54: The Effect of Media Freedom and Democracy Components on Women’s Economic Rights
5 Years Later N=661
Dependent Variable:
Women’s Economic Rights (+5 years)
Model 1 Model 2 Model 3
Women’s Economics Rights 2.848 (.277)*** 2.849 (.278)*** 2.834 (.281)***
Media Freedom -.342 (.252) -.954 (.579) -.882 (.512)
Executive Constraints .443 (.420) .243 (.444) .336 (.413)
Media Freedom*Executive Constraints --- .826 (.732) ---
Regulation of Participation .481 (.456) .357 (.471) -.039 (.606)
Regulation of Participation*Media Freedom --- --- .884 (.721)
Catholic .157 (.327) .167 (.332) .167 (.334)
Protestant -.004 (.331) -.029 (.338) -.020 (.338)
Muslim -.672 (.326)* -.678 (.327)* -.687 (.322)*
GDP/capita (logged) .527 (.173)** .523 (.171)** .506 (.173)**
Population (logged -.109 (.069) -.101 (.070) -.105 (.070)
Interstate Conflict -.663 (.417) -.691 (.422) -.656 (.405)
Internal Conflict -.514 (.233)* -.526 (.239)* -.531 (.228)*
Wald χ
2
203.76 213.41 220.54
BIC -5176.281 -5170.482 -5171.043
McKelvey & Zavoina’s R
2
.65 .65 .66
*p<.05; **p<.01;*** p<.001
248
Table 55: The Effect of Media Freedom and Democracy Components on Women’s Social Rights
N=927
*p<.05; **p<.01;*** p<.001
Dependent Variable:
Women’s Social Rights
Model 1 Model 2 Model 3
Women’s Social Rights (-1 year) 3.719 (.226)*** 3.721 (.226)*** 3.709 (.227)***
Media Freedom -.529 (.286) -.969 (.777) -.983 (.523)
Executive Constraints .566 (.341) .426 (.414) .484 (.368)
Media Freedom*Executive Constraints --- .590 (.920) ---
Regulation of Participation .568 (.405) .482 (.443) .148 (.628)
Regulation of Participation*Media Freedom --- --- .735 (.730)
Catholic .380 (.295) .383 (.299) .385 (.303)
Protestant .375 (.294) .355 (.298) .356 (.299)
Muslim -.661 (.324)* -.668 (.325)* -.683 (.319)*
GDP/capita (logged) .506 (.121)*** .503 (.118)*** .485 (.119)***
Population (logged -.074 (.059) -.069 (.061) -.071 (.060)
Interstate Conflict -.362 (.552) -.382 (.567) -.355 (.553)
Internal Conflict -.205 (.203) -.213 (.208) -.219 (.205)
Wald χ
2
392.20 390.30 397.57
BIC -5290.740 -5284.350 -5284.883
McKelvey & Zavoina’s R
2
.81 .81 .81
249
Table 56: The Effect of Media Freedom and Democracy Components on Women’s Social Rights
Five Years Later N=661
Dependent Variable:
Women’s Social Rights (+5 years)
Model 1 Model 2 Model 3
Women’s Social Rights 2.128 (.251)*** 2.131 (.253)*** 2.165 (.256)***
Media Freedom -.448 (.564) .177 (1.039) .336 (.829)
Executive Constraints 1.173 (.623) .1.349 (.730) 1.269 (.612)*
Media Freedom*Executive Constraints --- -.817 (1.326) ---
Regulation of Participation .040 (.496) .178 (.532) .762 (.957)
Regulation of Participation*Media Freedom --- --- -1.181 (1.133)
Catholic .421 (.488) .396 (.472) .404 (.467)
Protestant .565 (.444) -.591 (.454) .580 (.439)
Muslim -.868 (.488) -.874 (.490) -.859 (.499)
GDP/capita (logged) .674 (.205)** .675 (.205)** .698 (.201)**
Population (logged .037 (.108) .026 (.110) .042 (.106)
Interstate Conflict -1.438 (.633)* -1.410 (.648)* -1.451 (.635)*
Internal Conflict -.145 (.434) -.117 (.439) -.120 (.423)
Wald χ
2
225.86 227.59 233.86
BIC -3275.379 -3269.542 -3271.042
McKelvey & Zavoina’s R
2
.70 .70 .70
*p<.05; **p<.01;*** p<.001
250
Table 57: The Effect of Media Freedom and Autocracy/Democracy on Women’s Political Rights N=938
Dependent Variable:
Women’s Political Rights
Model 1 Model 2 Model 3 Model 4
Women’s Political Rights (-1 year) 5.226 (.325)*** 5.225 (.324)*** 5.225 (.323)*** 5.225 (.322)***
Media Freedom .416 (.363) -.290 (.757) .569 (.358) .220 (.665)
Autocracy/Democracy (5 part) .113 (.134) .032 (.139) --- ---
Media Freedom*Autocracy/Democracy (5 part) --- .284 (.230) --- ---
Autocracy/Democracy (Polity) --- --- .007 (.026) -.001 (.026)
Media Freedom*Autocracy/Democracy (Polity) --- --- --- .056 (.074)
Catholic -.263 (.251) -.241 (.261) -.281 (.247) -.283 (.251)
Protestant .207 (.321) .157 (.321) .194 (.323) .163 (.321)
Muslim -.371 (.317) -.413 (.319) -.414 (.316) -.432 (.320)
GDP/capita (logged) .227 (.152) .189 (.152) .282 (.142)* .252 (.143)
Population (logged .049 (.068) .057 (.066) .045 (.068) .053 (.067)
Interstate Conflict -.836 (.318)** -.864 (.316)** -.813 (.318)* -.836 (.316)**
Internal Conflict -.234 (.249) -.232 (.256) -.216 (.252) -.222 (.253)
Wald χ
2
303.43 297.95 312.68 310.62
BIC -5731.862 -5726.353 -5731.263 -5725.212
McKelvey & Zavoina’s R
2
.80 .80 .80 .80
*p<.05; **p<.01;*** p<.001
251
Table 58: The Effect of Media Freedom Democratic Components on Women’s Political Rights N=927
*p<.05; **p<.01;*** p<.001
Dependent Variable:
Women’s Political Rights
Model 1 Model 2 Model 3
Women’s Political Rights (-1 year) 5.205 (.324)*** 5.205 (.324)*** 5.191 (.329)***
Media Freedom .518 (.331) .250 (1.252) -.385 (.632)
Executive Constraints .150 (.448) .074 (.465) -.057 (.465)
Media Freedom*Executive Constraints --- .353 ( 1.446) ---
Regulation of Participation .742 (.438) .696 (.511) -.119 (.803)
Regulation of Participation*Media Freedom --- --- 1.508 (.967)
Catholic -.208 (.253) -.208 (.255) -.202 (.262)
Protestant .230 (.311) .220 (.312) .203 (.316)
Muslim -.391 (.310) -.393 (.310) -.402 (.308)
GDP/capita (logged) .207 (.153) .205 (.153) .173 (.154)
Population (logged .053 (.064) .056 (.063) .057 (.066)
Interstate Conflict -.795 (.318)* -.809 (.314)* -.789 (.298)**
Internal Conflict -.231 (.259) -.238 (.256) -.248 (.275)
Wald χ
2
305.27 306.37 305.07
BIC -5642.753 -5636.006 -5638.122
McKelvey & Zavoina’s R
2
.80 .80 .80
252
253
Figure 39: Media Freedom and Levels of Democracy/Autocracy 1948-1995
0 .1 .2 .3
kernel density
-10 -5 0 5 10
Polity IV Regime Score
free media government controlled media
254
Figure 40: The Effect of Media Freedom on the Predicted Probability
of Frequent Political Torture
0 .2 .4 .6 .8
Pr. Frequent Political Torture
0 .2 .4 .6 .8 1
Regulation of Participation
Free Media Controlled Media
255
Figure 41: The Effect of Media Freedom on the Predicted Probability of No Political
Torture Five Years Later
0 .1 .2 .3
Pr. No Political Torture
0 .2 .4 .6 .8 1
Regulation of Participation
Free Media Controlled Media
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Power to the masses? The rise of direct democracy in Latin America
PDF
The erosion of democracy in the post–Cold War world: understanding the phenomenon and its causes
PDF
Property and democracy: authority in four American property-rights regimes
PDF
The partial Good Samaritan states: China and Japan in the international relations of autocracy and democracy
PDF
The politics of transformation: mass media and the Northern Ireland peace process
PDF
The political psychology of democracy promotion, or how democracies democratically promote democracy promotion
PDF
Egalitarianism begins at home and in cohorts: egalitarianism in United States public opinion, 1952-2000
PDF
Institutional variance of the democratic peace, 1816-2002: electoral, executive, and federal institutions in time and space
PDF
Partisan media bias vs. structural media bias: which one dominates in a polarized environment?
PDF
Use Signal, use Tor? The political economy of digital rights technology
PDF
Health care for all? Anti-Latino and anti-immigrant attitudes, health care policy, and the Latino community
PDF
Democracy in the 21st century: social media and politics - global village or cyber-Balkans?
PDF
Authoritarian religion: explaining when and why authoritarian governments instrumentalize religion
PDF
Enabling freedom: Richard Rorty and education
PDF
Shaking the foundations of violent civil conflict: institutions, disasters, and the political economies of state-rebel interaction
PDF
Selling virtue: how human rights NGOs and their donors work together to create a better world ... for themselves
PDF
Political determinants and economic effects of HIV/AIDS: a push for the multisectoral approach
PDF
Developing countries in the digital era: state and business interactions for industrial upgrading
PDF
Carnival in the birdcage: a study of the film and media industries' reforms in China after 1978 and its soft power
PDF
Decisions to ratify the Kyoto Protocol: a Latin American perspective on poliheuristic theory
Asset Metadata
Creator
Whitten-Woodring, Jenifer
(author)
Core Title
The fabled fourth estate: media freedom, democracy and human rights
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Politics
Publication Date
08/05/2010
Defense Date
06/25/2009
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
autocracy,censorship,democracy,human rights,media freedom,OAI-PMH Harvest,press freedom,Protest
Place Name
Mexico
(countries),
North America
(countries),
Uganda
(countries),
Uruguay
(countries)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
James, Patrick (
committee chair
), Crigler, Ann N. (
committee member
), Seib, Philip (
committee member
), Wise, Carol (
committee member
)
Creator Email
jenifer.woodring@gmail.com,whittenw@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m3277
Unique identifier
UC1233760
Identifier
etd-WhittenWoodring-3773 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-373942 (legacy record id),usctheses-m3277 (legacy record id)
Legacy Identifier
etd-WhittenWoodring-3773.pdf
Dmrecord
373942
Document Type
Dissertation
Rights
Whitten-Woodring, Jenifer
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
autocracy
media freedom
press freedom