International Policy Diffusion and Religious Freedom,

Transcription

1 Utah State University All Graduate Theses and Dissertations Graduate Studies International Policy Diffusion and Religious Freedom, Allison R. Hale Utah State University Follow this and additional works at: Part of the International Relations Commons Recommended Citation Hale, Allison R., "International Policy Diffusion and Religious Freedom, " (2017). All Graduate Theses and Dissertations This Thesis is brought to you for free and open access by the Graduate Studies at It has been accepted for inclusion in All Graduate Theses and Dissertations by an authorized administrator of For more information, please contact

2 INTERNATIONAL POLICY DIFFUSION AND RELIGIOUS FREEDOM, by Allison R. Hale A thesis submitted in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE in Political Science (International Security) Approved: Greg Goelzhauser, Ph.D. Major Professor Anna Pechenkina, Ph.D. Committee Member Laura Gamboa, Ph.D. Committee Member Norm Jones, Ph.D. Committee Member Mark R. McLellan, Ph.D. Vice President for Research and Dean of the School of Graduate Studies UTAH STATE UNIVERSITY Logan, Utah 2017

3 ii Copyright Allison R. Hale 2017 All Rights Reserved

4 ABSTRACT iii International Policy Diffusion and Religious Freedom, by Allison R. Hale, Master of Science Utah State University, 2017 Major Professor: Dr. Greg Goelzhauser Department: Political Science Why do governments restrict religious freedom? As more and more governments have adopted restrictive policies over the past thirty years, scholars have traditionally examined domestic factors that may influence government choices. I build on this literature by extending the discussion to external factors, proposing that mechanisms of international policy diffusion contribute to the adoption of restrictive government religion policy. Further developing the policy diffusion literature with an examination of restrictive policies, I advance the argument that several specific mechanisms such as geography, learning, imitation, or social construction are more closely interrelated than many previous studies assume. To test these theoretical assumptions, I compile a dataset that captures years of policy adoption for twenty types of restrictive government religion policy based on the information available from the Religion and State (RAS) Project for 175 countries between 1990 and Applying logistic regression models to conduct basic event history analysis, I find

5 moderate support for the assumption that all four of the policy mechanisms examined iv increase the odds of restrictive policy adoption. (74 pages)

6 PUBLIC ABSTRACT v International Policy Diffusion and Religious Freedom, Allison R. Hale Why do governments restrict religious freedom? As more and more governments have adopted restrictive policies over the past thirty years, scholars have traditionally examined internal domestic factors such as the role of democratic governance, economic growth, or internal competition that may influence government choices. I build on this literature by extending the discussion to external factors, arguing that some policies may also spread from one government to another. This process, identified by scholars as the idea of policy diffusion, may occur in several ways. While previous research has focused on the spread of policies that are generally considered positive (i.e. the spread of democracy), I extend the literature by specifically focusing on the spread of restrictive policies. I argue that these policies may spread across countries through several specific mechanisms: geographic neighbors may observe each other, policymakers may learn generally from the adoption of policies throughout the world, countries may imitate the examples of others they consider powerful, or the merits of a policy may be socially constructed within groups of countries that have similar cultures. To examine these theoretical assumptions, I first compile a dataset that captures years of policy adoption for twenty types of restrictive government religion policy based on the information available from the Religion and State (RAS) Project for 175 countries between 1990 and I then test the data with several statistical models that allow me to compare the extent to which the proposed mechanisms change the likelihood that a government will

7 adopt a restrictive policy. Through these tests, I find moderate statistical support for the vi assumption that all four of the policy mechanisms examined increase the odds of restrictive policy adoption.

8 ACKNOWLEDGMENTS vii I am very grateful for the support of the entire Political Science department as I have taken classes, explored opportunities, asked questions, and decided what types of research I wanted to pursue. I would especially like to thank my committee members, Drs. Greg Goelzhauser, Anna Pechenkina, Laura Gamboa, and Norm Jones, for their support, patience, and assistance throughout the entire thesis process. I give special thanks to my family, friends, and colleagues for their encouragement, moral support, and patience as I worked my way from the initial proposal writing to this final document. I could not have done it without all of you. Allison R. Hale

9 CONTENTS viii Page ABSTRACT... iii PUBLIC ABSTRACT... v ACKNOWLEDGMENTS... vii LIST OF TABLES... ix LIST OF FIGURES... x INTRODUCTION... 1 PREVIOUS LITERATURE... 4 Studying Religion and State... 4 Origins of Government Religion Policy... 8 The Logic of Policy Diffusion INTERNATIONAL DIFFUSION MECHANISMS Geography Learning Imitation Social Construction EVALUATING DIFFUSION OF GOVERNMENT RELIGION POLICIES What Counts? Measuring Government Regulation of Religion Assessing Diffusion Mechanisms Controlling for Domestic Influences RESULTS CONCLUSION BIBLIOGRAPHY APPENDIX LOGIT TABLES AND FIGURES... 57

10 LIST OF TABLES ix Table Page 1 Types of Restrictive Policies Included in the Dataset Descriptive Statistics Correlation Matrix Individual Diffusion Mechanisms Accounting for Multiple Mechanisms Mechanisms Without Controls Individual Diffusion Mechanisms (Reporting Odds Ratios, Uncorrected for Rare Events) Accounting for Multiple Mechanisms (Reporting Coefficients, Uncorrected for Rare Events) Mechanisms Without Controls (Reporting Coefficients, Uncorrected for Rare Events)... 64

11 LIST OF FIGURES x Figure Page 1 Number of policies adopted annually Total policies adopted worldwide Percentage of policies adopted by politico-geographic region Predicted probabilities associated with geographic diffusion Predicted probabilities associated with general learning Predicted probabilities associated with the imitation of powerful countries Predicted probabilities associated with the imitation of OECD member countries Predicted probabilities associated with social construction in the same politico-geographic region Predicted probabilities associated with social construction across countries with the same majority religion... 62

12 INTRODUCTION In June 2016, Pew Research Center reported that up to 74 percent of the world s population lives in countries with high or very high levels of government regulation of religion ( Trends in Global Restrictions on Religion ). For example, Christian communities in China have seen numerous churches bulldozed and crosses torn down by the government (USCIRF 2016). Muslim women in at least twelve countries are subject to full or partial bans that prohibit the burka or other similar forms of religious dress (Sanghani 2016; USCIRF 2016b). Recent anti-terrorism legislation in Russia bans any form of preaching, praying, proselytizing, and dissemination of religious material outside of state-approved religious sites, and authorize[s] fines of up to $15,000 for these activities conducted in private residences or distributed through mass print, broadcast, or online media (USCIRF 2016c). Throughout several countries, prisoners of conscience people whom governments hold for reasons including those related to religion are regularly incarcerated and in some cases, subject to the death penalty (USCIRF 2016). Moreover, between 2007 and 2013, Pew observed that the number of countries in which governments specifically targeted, restricted, or harassed religious groups increased from 118 to 133 ( Latest Trends In Religious Restrictions and Hostilities ). Additional academic studies further confirm that the overall level of government involvement in religion (GIR) that is, government restrictions on religious practice, direct or indirect government support for religion, or other legislation that controls religious behavior in some way has largely increased over time, and religious legislation remains ubiquitous even in countries that constitutionally separate religion and state (Fox 2006, 2011a). Additional studies seem to confirm that a constitutional guarantee of religious freedom seems to only translate to minimal protection of individual rights in practice (Fox 2015, Fox and Flores 2009; Chilton

13 and Versteeg 2014). Yet even as individual governments increasingly regulate religion, 2 prominent intergovernmental organizations such as the United Nations continue to champion and uphold the right of individuals to freedom of thought, conscience and religion, including the right to change his religion or belief and to manifest his religion or belief in teaching, practice, worship, and observance (United Nations 1948). Why, then, do governments seem increasingly willing to adopt restrictive policies to regulate religious practice and expression? Since Berry and Berry (1990) first examined the adoption of a state lottery across the United States to understand whether or not public policies may predictably spread from one polity to another, a significant body of literature examining the mechanisms of policy diffusion has developed across several subfields of political science (Graham, Shipan and Volden 2013). This task is particularly difficult for scholars of international policy diffusion. As Simmons, Dobbin, and Garrett (2006, 789) summarized, The challenge facing theorists of international policy diffusion is to demonstrate that domestic political and economic factors cannot alone predict when governments adopt new policies, and to develop and test hypotheses that distinguish among the several possible mechanisms of diffusion. In short, the assumption that nations internally develop public policy independently from one another is juxtaposed against an assumption of policy interdependence, under which nations develop public policies as informed by some sense of the international community around them (Franzese and Hays 2008). I examine these competing arguments with respect to government regulation of religion, assessing the likelihood that a country will adopt a regulation in a given year by statistically testing data newly compiled from the Religion and State (RAS) project, V-Dem database, Correlates of War project, and World Bank. Using these data, I evaluate policy adoption in several different categories of regulation such as laws addressing religious

14 3 speech, observance, organizations, and proselyting across 175 countries between 1990 and I find moderate statistical support for the assumption that restrictive government religion policies are diffusing across countries through several possible mechanisms; future study, however, will be necessary to further clarify which specific mechanisms can best predict policy adoption in this case. These results expand the current political science literature on religion-state relationships by looking beyond purely domestic factors that may influence policy adoption, and extend discussions in the diffusion literature about the best ways to conceptualize and measure diffusion mechanisms. First, I situate this study within the general literature on the relationship between governments and religion, specifically exploring the implications of studying explicit government restrictions on religious activity. Then, I build on existing diffusion literature to develop hypotheses about specific mechanisms that may account for the international diffusion of government policies restricting religion. Within this theoretical framework, I outline my empirical approach; in addition to operationalizing my hypotheses on diffusion mechanisms, I draw on literature in comparative politics and international relations to account for relevant domestic factors that may contribute to policy adoption. After evaluating my results, I conclude by briefly discussing the implications of this study for future work in political science.

15 PREVIOUS LITERATURE 4 Studying Religion and State Studying religion in society is inherently difficult. In many fields, simply defining what constitutes religion, much less attempting to understand the overlap between religion and society, has been the subject of intense analysis and debate (Hulsether 2005; Nongbri 2008). Even narrowing the subject to focus just on the interaction of religion and politics is fraught with challenges. Recently warning scholars and policymakers of the instability and even incoherence of the category of religion as [a] basis of legal protection, Elizabeth Shakman Hurd argues that the adoption of religion as a legal and policy category helps to create the world that it purports to oversee and in so doing oversimplifies our view of social relations (2015, ). As such, she suggests that scholars ought to dethrone religion as a stable interpretive and policy category in their research (114). In short, Hurd suggests that because governments approach religion policy differently over time or apply such policies unevenly across different religious, spiritual, or non-religious communities in society, that the concept of religion policy cannot be systematically studied as such by political scientists. Acknowledging the complexity of the task, however, I disagree with Hurd s logic and concur with Ani Sarkissian s recent conclusion: By ignoring religion or incorporating it into analyses of other types of identities, we may risk missing the essential role it plays in the politics of many societies (2015, 6). Governments around the world have arguably sought to restrict, support, or control religion since at least the fourteenth century BCE and, as highlighted in the introduction to this paper, continue to do so today (Sarkissian 2015, 1-3). While government religion policies either codified in law or in consistent government practice are observable, however, many early

16 scholars failed to evaluate such policies systematically. 5 Admittedly, early scholarship often focused more generally on the role or influence of religion in society as a cultural force, and weighed in implicitly or explicitly to the normative debate over the appropriate balance of religion and state. Modernization and secularization theorists of the nineteenth and early twentieth century, for instance, largely assumed that as countries became more modernized, both the role of religion in the public sphere and private religiosity would decline (Almond 1960; Apter 1965; Beckford 1985; Deutsch 1953; Norris and Inglehart 2004; Smith 1970, 1974; Westhus 1976; Wilson 1966, 1982). The secularization school was a prominent influence on the social scientific study of religion through the 1990s and perhaps a bit later, as Jonathan Fox (2015) explains. He outlines at least two significant effects that this emphasis had on the literature: First, it arguably deterred research on the topic, as scholars assumed religion would become increasingly irrelevant; second, it forced much of the research that did exist to spend its efforts refuting secularization theory rather than building knowledge on the relationship between religion and politics (Fox 2015: 25). As Roger Finke observes, religion-state relationships had been virtually ignored in international studies. Prior to 2000, there were no systematic data collections on religious freedoms and few studies attempted to understand the origins or consequences of these freedoms (2013). By the late twentieth century, however, a few new theories began to emerge to explain how religion could become an increasingly salient force throughout the world (Antoun and Hegland 1987; Casanova 1994; Kepel 1994; Sahliyeh 1990; Westerlund 1996). For example, Huntington s prominent and controversial theory about a clash of civilizations which argues that conflict and competition between several religiously

17 homogeneous civilizations will dominate international politics has been subject to 6 intensive debate as scholars have sought to better understand the potential role of religion and religious identities in driving international conflict in the post-cold War world (Fox 2005; Huntington 1996; Henderson and Tucker 2001; Mungiu-Pippidi and Mindruta 2002; Russett, Oneal, and Cox 2000). Yet while many have continued to evaluate a culturally-based assumption that religion engenders conflict and cooperation (as evidenced in part by the publication of extensive surveys such as the Routledge Handbook of Religion and Security (Seiple, Hoover and Otis 2012) or the Oxford Handbook of Religion, Conflict and Peacebuilding (Omer, Appleby and Little 2015)), relatively few scholars have methodically explored or empirically tested concrete ways in which religion and governments interact or the ways that different forms of institutionalized religion-state relationships may influence (or be influenced by) conflict behavior or other outcomes. In short, a new body of literature on government religion policy has been slowly developing since roughly 2000 (Philpott 2009). Several important studies focus primarily on the potential causal impact of government policy on other outcomes. One important step forward in this endeavor is Grim and Finke s The Price of Freedom Denied: Religious Persecution and Conflict in the Twenty-First Century (2011). With a combination of large-n empirical analysis and a series of case studies, Grim and Finke find that when governments restrict religious freedom, the likelihood of violent persecution, conflict, and terrorism within a country increases. Specifically, they argue that government and social restrictions of religion are a mechanism through which social, political, economic, and religious differences make a difference in predicting levels of social and political stability and conflict (2011: 86).

18 Interestingly, measures of government favoritism of religion were not statistically 7 significant when included; although correlated with higher restrictions of religious freedom, observing government favoritism of religion was less directly effective in predicting conflict (2011: ). Citing a separate Hudson Institute study, Grim and Finke note wherever the level of religious freedom is high, there tends to be fewer incidents of armed conflict, better health outcomes, higher levels of earned income, prolonged democracy, and better educational opportunities for women (2011: 206). Reflecting on religious freedom as part of a bundle of freedoms, Grim and Finke conclude that significant research remains to disentangle causal relationships as religious freedom is increasingly associated with many positive outcomes (2011: 206). Additional literature has begun to explore the potential impact of religious cultures and government regulation of religion on economic outcomes. Early studies attempted to link economic development and the rise of capitalism to the dominance of Protestantism in a country, although this assertion has been questioned (Guiso et al. 2003; Iannccone 1998; Weber 1930 (2001); Landes 1998). In terms of the religion-state relationship, Barro and McCleary (2005) analyze the economic implications of whether or not a country has an established state religion at three specific points in the last century (1900, 1970, 2000), but find little relationship between established religion and per capita GDP. Gill (2013) explores several potential mechanisms that would link positive economic outcomes to the provision of religious freedom; Grim, Clark, and Snyder (2014) argue specifically that religious freedom contributes to good business and economic outcomes even after controlling for other political or social factors. While these types of studies are interesting, they still emphasize religion policy as the independent variable. Only a few recent studies, in fact, have considered reversing the causal direction: can

19 economic indicators actually predict the extent to which governments regulate religion? 8 Theoretically, this claim is just beginning to emerge. Fox (2006), for example, found that high economic development might actually be correlated with an increased likelihood of government involvement in religion. Buckley and Mantilla (2013) similarly argue that high levels of economic development may enable increased regulation of religion through associated growth in state capacity, finding a consistently positive and substantively significant relationship between development and the level of government regulation targeting religion. How then, can scholars begin to disentangle the relationship between government religion policies and other outcomes? I suggest that to better understand the impact of policies providing religious freedom (relative to more restrictive government policies), scholars must first understand the conditions that result in high or low levels of government restriction of religion in the first place. If, as Grim and Finke assert, government religion policy moderates or amplifies existing differences in a way that can shape conflict behavior or promote positive economic and social outcomes, it becomes even more important to understand how and why governments adopt policies that restrict religion. Origins of Government Religion Policy Recent studies have begun to explore the domestic origins of religious restrictions or the deregulation of religion. Building on rational choice theory, the religious economies argument developed primarily by Finke (1990, 2013), Stark and Finke (2000), and Gill (2008) holds that a government s regulation or deregulation of religion is largely a result of competing interests held by politicians and religious leaders. Suggesting that religious groups compete for resources from the state and that politicians seek to maximize the chances of

20 9 their political survival, Gill (2008) argues that politicians will court religious leaders to gain legitimacy, particularly in countries with majority religious groups, but will ultimately favor deregulation as their political position becomes more secure. Despite the prevalence of this theory throughout new literature, however, as Finke and Martin (2014) acknowledge, little empirical work has directly tested these arguments. In fact, while a number of studies have begun to consider the relationship between domestic institutions, economy, or demography and government religion policy, the literature has failed to arrive at consistent conclusions. Based in part on Gill s assumptions, Grim and Finke (2011) extensively discuss potential religious and social motivations for restricting religious freedom: from the perspective of a dominant religion, to preserve and protect the culture and society as a whole, or to limit competition between religious groups for political or cultural power (46-50). From the perspective of states, controlling religious activity by forming alliances with select religious groups or by restricting the activities of all allows the state to secure political stability and survival by controlling any potential threats from religion (50). In short, they theorize that when politicians view alliances with established religions as necessary for survival or they perceive religious movements as a potential threat to the state, religious freedoms will be denied (51). Gill s work and other studies built on the religious economies framework, however, have more commonly focused on forms of political competition most prominently found in democracies. Democracies are recognized to generally provide greater protection for human rights than non-democracies, and therefore may be less likely to adopt restrictions on religious freedom (Apodaca, 2001; Bueno de Mesquita et al. 2003, 2005; Davenport, 1995, 1999; Hofferbert and Cingranelli, 1996; Keith, 2002; Poe and Tate, 1994; Poe, Tate, and Keith, 1999). At the institutional level, however, recent studies suggest that a democratic

21 political system or does not seem to necessarily imply a strong separation of religion and 10 state (Driessen 2010; Fox 2007). More nuanced work suggests that certain features of democratic governance may matter more than others: Finke and Martin (2014) find that while free elections or measures of government effectiveness were insignificant in their models, the absence of an independent judiciary was a strong predictor of religious restrictions. What about non-democracies? In Varieties of Religious Repression, Ani Sarkissian (2015) builds on Gill s (2008) religious economies framework and insights from Grim and Finke (2011) to develop a theory of religious repression in nondemocratic countries, acknowledging that the policies, regimes and incentives in nondemocratic states will differ from those in democracies. She argues that levels of religious repression in a country are a function of both political competition and religious divisions in society, based on preliminary quantitative analysis and a number of case studies. In addition to extending the religious economies framework beyond democracies, Sarkissian s work notably begins to address another shortfall of previous work on the subject: the fact that Gill and others primarily frame their work with a focus on conditions leading to deregulation of religion, rather than spending much time explaining the conditions that lead to repression of religion, which studies show has been increasing (2015: 11). A recent, detailed analysis of different types of government religion policy across nearly 200 countries found that between 1990 and 2008, 55 percent of states instituted new/additional religion policies while only 12 percent removed them (about 16 percent did some combination of both and 17 percent did not significantly change) (Fox 2015). In fact, while 90 percent of all countries offer some legal assurance of religious freedom, roughly 86 percent still have laws regulating religious practice (Finke and Martin 2014). As such, I suggest with Sarkissian that scholars must continue to develop theoretical explanations to

22 account for these documented increases in the regulation of religion across the world. As 11 described above, several authors have begun to explore potential domestic determinants of state religion policy. I thus expand upon existing literature on religion and state relationships by looking beyond internal determinants of policy, attempting to evaluate the ways in which restrictive religion policy may spread across countries. The Logic of Policy Diffusion While domestic characteristics are likely to partially predict the adoption of restrictions, governments in individual countries also observe and interact with one another in ways that may condition their policy choices. I suggest that theories of policy diffusion may provide additional insight into a country s willingness to adopt restrictive government religion policy. Although the diffusion literature in IR and comparative politics is comparatively underdeveloped relative to that in the American politics subfield, diffusion/contagion processes fit logically in that set of questions which are traditionally asked by foreign policy or comparative politics analysts (Most and Starr 1990, 392). Scholars across subfields largely agree about the generic meaning of policy diffusion, despite slight differences in the way diffusion literature has developed across the different specialties. At a basic level, Shipan and Volden (2012) define diffusion as one government s policy choices being influenced by the choices of other governments. Karch (2007) clarifies, Diffusion is about the movement of a policy across jurisdictional boundaries. In contrast, adoption is the decision to establish a policy in an individual jurisdiction. Other scholars further emphasize diffusion as a process, as opposed to an outcome. That is, diffusion is the interdependent process conducive to the spread of policies, not the extent of convergence that can result from it (Gilardi 2012).

23 What have scholars learned about the process of policy diffusion thus far? Early 12 studies of diffusion in the American politics subfield, particularly those that developed theoretical and methodological approaches allowing for internal and external policy determinants to be tested together, laid the groundwork for an extensive literature to emerge during the 1990s and 2000s (Walker 1969, Gray1973, Berry and Berry 1990). Several authors have previously provided detailed reviews of this literature and draw out key theoretical lessons about policy diffusion based on previous results (Shipan and Volden 2012, Karch 2007, Gilardi 2012). These reviews each suggest that the literature has developed broadly under three questions succinctly articulated by Karch (2007): first, why does diffusion occur that is, what driving forces (referred to as mechanisms ) result in diffusion? Second, which political actors or forces (such as policy entrepreneurs or crossborder organizations) facilitate diffusion? Finally, what specific policy content is being diffused? In an effort to address these questions, the diffusion literature has grown from basic studies simply attempting to gauge whether or not diffusion has occurred to increasingly sophisticated analyses developing detailed theoretical claims about different potential diffusion mechanisms and actors. For instance, recent work has pioneered new techniques for evaluating the role of interest groups or advocacy networks as political actors facilitating diffusion (Garrett and Jansa 2015, Donno 2010). In terms of policy content, diffusion theory is typically used to explain the spread of policy innovations defined broadly as programs or policies that are new to the governments adopting them (Walker 1969). Scholars of American politics have typically focused their studies on the diffusion of specific types of statutes across state and local governments, such as the adoption of a state lottery (Berry and Berry 1990), or anti-smoking

24 legislation (Shipan and Volden 2008). In international relations and comparative politics, 13 diffusion scholarship has more broadly attempted to explain the emergence of international norms of behavior such as the tendency of countries to invite international election observers (Hyde 2009) or agree to weapons bans (Carpenter 2011), revolutionary waves such as the Arab Spring (Weyland 2012), and the spread of democracy and economic liberalism (Bunce and Wolchik 2011; Huntington 1991; O Loughlin et al. 1998; Simmons and Elkins 2004; Starr 1991; Wotipka and Ramirez 2008; Weyland 2010). As Maske and Volden (2011) note, however, there has been limited attention to the attributes of policies that may limit the generalizability of previous diffusion studies. Given that many studies find policy diffusion while many others do not, they assert, the lack of an overarching emphasis on differences across these policies has limited the potential for cumulative building of knowledge across studies. With a few exceptions (such as Horowitz s (2010) examination of the spread of suicide bombing tactics across terrorist organizations), the IR and comparative literature has overwhelmingly focused on the diffusion of innovations in terms of policies or techniques that are typically associated with the normative ideals of protecting human rights, strengthening democratic institutions, or improving economic outcomes. As such, this study further contributes to the diffusion literature by considering the potential diffusion of policies clearly designed to restrict personal freedoms.

25 INTERNATIONAL DIFFUSION MECHANISMS 14 I attempt to assess whether or not restrictions on religious practice are spreading by evaluating different mechanisms that have been highlighted in previous literature. The following subsections develop hypotheses about the spread of restrictive government religion policy based on four potential diffusion mechanisms: geography, learning, imitation, and social construction. As outlined below, several scholars have attempted to build on early analyses based on geography by distinguishing between the ideas of learning, imitation, and social construction as separate or independent diffusion mechanisms. I argue, however, that imitation and social construction should just be considered as modified possible forms of learning. Geography Diffusion scholars across disciplines have long considered the potential role of regional affinity as a mechanism of policy diffusion (Karch 2007). For many, the very definition and measurement of diffusion is linked with idea of geographic proximity. One example of a typical definition, provided by Bunce and Wolchik (2011), suggests: we understand diffusion to be a process whereby new ideas, institutions, policies, models, or repertoires of behavior spread geographically from a core site to new sites, whether within a given state or across states (17). Early, prominent studies in both the American and IR/comparative literature largely focused on the diffusion of policies between neighboring units or within regional bounds (Berry and Berry 1990; Lutz 1987; Mooney and Lee 1995; Weyland 2005; Collier and Messick 1975). Karch (2007) notes that most published research assumes that the existence of a

26 public policy in nearby states provides a model upon which state officials can draw. He 15 then outlines three of the most prominent theoretical explanations regularly offered to support the idea of geographic diffusion: first, the assumption that geographic proximity can facilitate the development of the communications networks among policymakers through which information about public policies spreads; second, the idea that overlapping media markets may alert citizens and government officials to the existence of political forms and policies in nearby states; and third, the assumption that public officials might be most inclined to use nearby states as policy models because they are likely to be culturally and demographically similar to their own states (Karch 2007). While some recent papers have revisited the basic idea of spatial/geographic diffusion (Brinks and Coppedge, 2006; Maske and Volden 2011), others have criticized the idea that geographic proximity, in and of itself, matters. For instance, Shipan and Volden (2012) argue that while offering a good starting point, the classic view of policy diffusion as geographic clustering is often overly limiting, sometimes misleading (or even wrong), and increasingly outdated. Most critics emphasize that while regional clustering may have shaped diffusion in the past, today s low barriers to communication and travel weaken the assumption that countries primarily obtain information about public policies from their neighbors (Shipan and Volden 2012, Karch 2007). Moreover, only limited empirical support exists to support the idea of a regional diffusion effect (Mooney 2001, Karch 2007). As such, scholars have largely begun to develop and test more specific potential mechanisms of diffusion often based in part on assumptions previously associated with geography such as those highlighted above on their own terms. Although I agree with the assertion that the role of geographic proximity in the diffusion process is more complicated than most scholars have heretofore suggested, I

27 believe there are still reasons to include and test for the idea of spatial diffusion. First, the 16 prominence of geographic explanations in the literature suggests that simply ignoring the possibility without some form of empirical test would be insufficient. Second, although critics argue that the role of geography should be less prominent as communication and transportation networks has increased over time and many of the assumptions that justify studies of geographic diffusion have since been tested through more specific theoretical mechanisms, these arguments do not fully rule out the basic possibility that geography matters to some extent. Finally, it seems imprudent to assume that the impact of geographic proximity on diffusion would necessarily be the same 1) in both subnational and international studies, or 2) for all types of policy. As such, I briefly address both general and specific approaches to policy diffusion. Particularly because few studies evaluate the spread of restrictive or illiberal policies, I begin with a general, geography-focused hypothesis of spatial diffusion simply to assess the extent to which the diffusion of religious restrictions may actually occur across countries. Ways to parse out any potential impact of geography as opposed to other, more specific mechanisms are developed and discussed in the sections that follow. H1: The likelihood of adopting religious restrictions increases as the number of policies adopted by countries within the same geographic region rises. Learning Policy diffusion scholars often consider learning as a diffusion mechanism. Although the basic idea that policymakers learn from experience, examples, and new information and act accordingly makes intuitive sense, precisely explaining and testing learning as a mechanism has proved particularly challenging (Maske and Volden 2011). Simmons, Dobbin and Garrett (2006) take an expansive view similar to the concept of

28 Bayesian updating, suggesting that learning refers to a change in beliefs or change in 17 one s confidence in existing beliefs, which can result from exposure to new evidence, theories, or behavioral repertoires (see also Dobbin, Simmons and Garrett (2007)). More fundamentally, Simmons and Elkins (2004) assert that learning occurs any time new information is acquired. While an important foundation, however, the idea that the availability of new information about a policy leads to learning may overly simplify a more complex causal process. As Simmons and Elkins (2004) acknowledge, information can be acquired in several different ways. Specifically, they identify policy success, communication (assuming information is more easily shared among governments that are in especially close communication), and cultural reference groups as possible diffusion mechanisms. Although they classify all mechanisms as forms of learning, the majority of diffusion scholars have focused more on the apparent success of a policy as the relevant information considered by policymakers in the learning process (Gilardi 2012). This perspective is summarized by Shipan and Volden (2008), who argue, In learning, policymakers focus on the policy itself how was it adopted, was it effective, what were its political consequences? Yet while policymakers may intend to learn from the experiences of others, they are inherently limited by how the human brain processes new information (Gilardi 2012). As such, effectively attempting to measure learning based on perceptions of a policy s success in effect, attempting to disassociate policy outcomes from the actor establishing policy may be difficult and unrealistic in quantitative studies. Psychologists have argued that people rely on simplifying heuristics to evaluate information and that doing so may bias the ways they interpret and act upon that information (Kahneman 2011, 7). For example, cognitive shortcuts such as availability and representativeness suggest that policymakers

29 do not consider all new information equally: particularly vivid examples are more 18 influential than less striking events, and policymakers may draw unwarranted inferences from a limited empirical basis (Tversky and Kahneman 1974; Weyland 2005, 2007, 2009; Gilardi 2012). Ultimately, policymakers may or may not distinguish between policy success and political consequences and discount information that is not in line with their preferences (Gilardi 2010, 2012; Volden, Ting and Carpenter 2008). Based on this work, I suggest that two other diffusion mechanisms discussed throughout the literature imitation and social construction, sometimes both classified as forms of emulation and categorized separately from learning may actually just be modified manifestations of learning. As such, I believe it may be difficult to make a hard distinction between learning and these two possible forms of emulation in consistently quantifiable ways. However, I do attempt to develop hypotheses that may assess the ability of these slightly different, albeit similar, mechanisms to predict policy adoption. With this in mind, my foundational learning hypothesis simply seeks to capture the idea that the availability of new information over time about a policy may influence policy adoption. I assume that because countries can observe and interact with one another, policy adoptions elsewhere increase the amount of information available to a country about the potential implications of a policy. H2: The likelihood of adopting religious restrictions increases as the number of restrictive policies adopted throughout the world rises. Imitation Although some studies do not distinguish between different types of emulation, others suggest that imitation is actor-centric whereas social construction is community-focused. For example, Shipan and Volden (2008) distinguish between learning and imitation by

30 arguing that imitation involves a focus on the other government what did that 19 government do and how can we appear to be the same? Simmons and Elkins (2004) also highlight how the perceived overall success of other countries (as opposed to just observable policy-specific success) may incentivize emulation: The apparent success of others may in fact be a cognitive shortcut to assessing policy consequences; the relevant question in this process is, What policies are the high achievers pursuing? Dobbin, Simmons, and Garrett (2007) similarly argue that diffusion motivated by imitation may be reflective of power asymmetry, but is founded more in a willing desire by both countries to reduce asymmetry. Exploring one way countries may work to reduce this type of asymmetry, Susan Hyde (2009) proposes that individual states can contribute to the formation and diffusion of international norms through signaling. Hyde s signaling theory suggests that democracy promoters reward states they recognize as democratizing, in turn incentivizing governments with uncertain regime types to signal their commitment to democracy in an effort to obtain benefits. As all true democrats begin to adopt a common signal in her study, inviting international election monitors she expects even pseudo-democrats to mimic the behavior in an effort to mask their true type and probabilistically compete for the benefits associated with democracy. How might these theories extend to the adoption of restrictive religion policy? At a basic level, I suggest that imitation is a form of learning from countries that are perceived as politically or economically successful. Whereas Hyde focuses on the potential for regimes to signal their intent to potential benefactors by mimicking the behavior of true democracies, I would argue that the concept of signaling may equally apply in a different way: if true democracies or other powerful countries that shape opportunities for alliances and aid continue to adopt restrictive policies, they may implicitly signal to other regimes that such

31 policies are considered acceptable in practice. This approach may help to further explain 20 why countries sometimes appear to adopt conflicting positions (such as inviting election monitors but ignoring their conclusions, or adopting a constitutional provision that protects religious freedom yet continuing to legislate or otherwise restrict religious practice). As such, I propose that the probability of policy adoption will increase as countries imitate relatively successful or powerful countries that have adopted similar policies. Potential ways to assess and measure this concept are discussed in more detail below. H3: The likelihood of adopting religious restrictions increases as successful or powerful countries adopt restrictive policies. Social Construction In his seminal piece, Anarchy is What States Make of It, Alexander Wendt explains that while both realism and liberalism disagree fundamentally about various aspects of international relations, both primarily draw upon the idea of a self-interested state as the starting point for theory (1992, 392). With this introduction, he argues that an examination of constructed identities and associated interests should precede the assumption of selfinterested state. Since this assertion, the school of constructivist thought in international relations has assumed that culture and meaning are socially constructed, suggesting that the merits or limitations of political action may be culturally encoded. For example, one of the causes proposed by Huntington (1991) in his theory regarding a third wave of democratization assumed that changes in church doctrine and practice spread throughout countries that shared a culturally similar, Catholic heritage; religious teachings were, in his view, communicated across countries through the cultural network and thus interpreted with special consideration, leading in part to the adoption of government policies to emphasize the importance of individual rights following Vatican II. Some diffusion authors have

32 acknowledged the rise of these types of theories in constructivist IR literature and have 21 since suggested that social construction can be a mechanism for policy diffusion. Dobbin, Simmons, and Garrett (2007), for instance, suggest that the merits of a policy are socially constructed through epistemic communities and reference groups, increasing the probability of policy adoption following an adoption by a nation s self-identified peers. Simmons and Elkins (2004) further explain this assumption, highlighting the potentially significant role of cultural reference groups in learning: Learning takes place at least partially through analogy, and lessons are viewed as more relevant to the extent to which a foreign case is viewed as analogous. The policies of culturally similar countries are perceived to (and in fact may) contain highly relevant information on the appropriateness of a particular policy in a specific context of shared values. This perceived similarity may provide a cognitive shortcut for an individual in a group decision-making context. Simmons and Elkins original hypothesis about cultural similarity and diffusion (which evaluated the influence of similarity on the spread of economic policy) has not been extensively tested in systematic ways by other scholars, even though it seems reasonable to hypothesize that cultural similarity across countries may increase the likelihood of policy emulation. I believe this may be due in part to the difficulty of measuring the concept of cultural similarity in quantitative studies. Although definitively understanding which nations constitute a country s self-identified peers may require in-depth qualitative or cultural studies, potential ways to measure the concept of cultural similarity are discussed in more detail below. Here, I attempt to test and extend Simmons and Elkins hypothesis by applying the general concept to the potential diffusion of restrictive religion policy. H4: The likelihood of adopting restrictive religion policies increases when culturally similar countries adopt such a policy.

33 EVALUATING DIFFUSION OF GOVERNMENT RELIGION POLICIES 22 What Counts? Measuring Government Regulation of Religion To test these hypotheses, I adapt data from the Religion and State (RAS) Dataset compiled by Jonathan Fox and others as a part of the Religion and State Project at Bar-Ilan University (Akbaba and Fox 2011; Fox 2015, 2013, 2011b, 2009, 2006). This dataset includes several variables that record the presence or absence of over one hundred different kinds of religious regulations (and a basic measure of the level at which such legislation is enforced), several indices measuring the overall level of government involvement in religion, and the presence of a constitutional provision that codifies the separation of religion and state or the specific establishment of a state religion for each country in the dataset. This study specifically includes relevant and available observations from in the RAS dataset, which spans 175 countries with data for each year between 1990 and One potential limitation to the dataset is the tendency to emphasize government practice over the adoption of specific laws. That is, in some cases a country may be coded as adopting a restriction even if a law has not been passed so long as it is consistent government practice to enforce such a restriction. This practice is explained further in this excerpt from the RAS codebook: It is important to emphasize that these codings focus on the relationship between religion and the state apparatus. For a variable to be coded, there must either be a law or a consistent government practice. In cases where the two contradict, consistent government practice was coded. These codings also represent the practice of the federal or national governments of states, and not practices by local governments. However, if a large number of local or regional governments engage in a practice it is also coded (Fox 2009). Despite this potential uncertainty in the coded data, however, I believe the data can still be used to evaluate the general trend of concern that is, whether or not restrictions on

34 religious practice are indeed diffusing across countries and if so, in what ways? As such, I 23 treat the data as representative of a policy adoption without attempting to distinguish whether or not the new governmental norm is codified in law or simply manifest in consistent government practice. I largely accept this assumption recognizing that government religion policy can be difficult to quantify. 1 What, for instance, should be considered restrictive government religion policy? Fox (2013) notes that while terms such as religious discrimination and religious freedom seem at first glance straightforward, they are not. Each has multiple possible meanings and interpretations (13). To operationalize these concepts, Sarkissian suggests that religious regulation can be defined broadly to include all laws and rules that are enacted to govern religious affairs in a state, those that both support and oppose religion (2015, 27). Here, I agree with Sarkissian and others that regulations that are intended to repress religion and therefore policies that restrict it can include both overtly discriminatory policies as well as policies that support or favor some forms of religious expression (Sarkissian 2015, 27; Fox 2015). Acknowledging the complex normative and scholarly debate surrounding what types of policies constitute religious restrictions, I focused only on policy types that could fairly clearly be categorized as restrictive based on the following assumptions, rather than including all of the policies tracked in the RAS dataset. 2 First, I consider discriminatory policies to be those that clearly violate the principles articulated in the UN Declaration of Human Rights as 1 For a more detailed discussion of the challenges associated with data collection and measurement when studying religion and state, see Fox 2011b. 2 In future work, expanding the sample to include more policies from the RAS dataset may improve statistical analysis by increasing the number of observations. For this project, however, time constraints and 2 In future work, expanding the sample to include more policies from the RAS dataset may improve statistical analysis by increasing the number of observations. For this project, however, time constraints and theoretical concerns namely an effort to focus on restrictive policies rather than all forms of government religion policy suggested that the selection of a sample might be appropriate.

35 24 cited in the introduction (that is, the right of an individual to change his religion or belief and to manifest his religion or belief in teaching, practice, worship, and observance ). Second, I recognize that government support for certain religious groups or actions in the form of funding or other resources can also constitute religious repression, as it often results in discrimination or restrictions against those who do not benefit from government support and may also be intended to restrict the [favored] group itself (Sarkissian 2015, 27). With this in mind, I selected a sample of twenty specific types of policy from the RAS dataset that may be considered representative of a government s overall level of restrictive religion policy. The original RAS dataset separately categorizes policies based on whether or not they target minority, majority, or all religions, as well as whether or not they are overtly restrictive or more implicitly regulatory. Although it may be interesting in future research to determine if restrictive policies categorized by targeted group may diffuse in different ways, this paper is designed to assess the spread of restrictions generally. As such, an effort was made to reduce selection bias by including roughly the same number of policies for comparison from each of the original RAS categories. In short, because the literature has not yet examined the potential international diffusion of restrictive religion policy in a general sense, my sample includes policies that may be categorized by their substantive goals regardless of whether or not they target minorities exclusively or all religions generally. The twenty policies I selected may be grouped into the following substantive categories, with specific examples provided in Table 1: 1. Restrictions on formal religious organizations (including religiouslyaffiliated or inspired political parties or unions);

36 2. Restrictions on religious speech and/or the dissemination of religious 25 materials; 3. Restrictions on public observances of religion, including holidays and any observation of the Sabbath or its equivalent; 4. Restrictions on religious proselytizing; and 5. Legislation that supports some forms of religious expression but may, as described above, be used by the government in restrictive ways (i.e. the presence of religious education in publicly-funded schools or the provision of government funding for building, repairing, or maintaining religious sites). For each policy, the original RAS dataset simply provides information from year to year about whether or not a policy is in place and the extent to which it is enforced. To assess policy diffusion, however, the dependent variable of interest should capture whether or not a nation adopts new policies over time. Available data on these policies are thus recoded dichotomously to reflect years of policy adoption and combined across categories to ensure that there are sufficient observations (at a country-year level of analysis) with which to test the hypotheses. This approach of evaluating policy diffusion with categories of different types of legislation instead of focusing on one specific policy has also been applied in other prominent studies of policy diffusion. This article employs the same logic as Shipan and Volden (2008), who examine policy diffusion across cities throughout the United States by evaluating anti-smoking policy adoption with three different categories of anti-smoking laws. As they highlight, examining multiple policy choices in the same years for the same countries allows a policy adoption to be considered a type of repeated event, as a policy in any of the

37 categories could be adopted at any time in any order (Box-Steffensmeier and Zorn 2002; 26 Wei, Lin, and Weissfeld 1989; Shipan and Volden 2006, 2008). In short, I first summarize the number of policies (of the twenty policies included in the study) adopted by each country in each year. While many countries had years without any policy adoptions, the maximum number of policies adopted in a given year by one country was seven. Figure 1 highlights the number of policies adopted in each year of the study, and Figure 2 notes that by the end of 2008, worldwide policy adoption (in the categories sampled) totaled an additional 203 policies. I then use this information to create a consolidated dependent variable that is initially set equal to 0 and is then set equal to 1 in the years in which a country adopts additional restrictions, resulting in 140 observations of adoption at a country-year level. Due to the dichotomous nature of the dependent variable, I employ logistic regression for the purpose of conducting standard binary event history analysis (as a commonly utilized tool throughout diffusion literature) (Berry and Berry 1990; Wotipka and Ramirez 2008; Graham, Shipan and Volden 2013).

38 27

39 28

40 Assessing Diffusion Mechanisms 29 My general spatial diffusion hypothesis suggests that a country will be more likely to adopt a restriction if its geographic neighbors have already done so. To test this hypothesis, I construct the variable Geography, which captures the total number of policies adopted by countries in the same geographic region between 1990 and the prior year (i.e. for 1996, it represents the total number of policies adopted in the region between 1990 and 1995). Geographic region is coded based on a country s regional designation by the UN Statistics Division (United Nations n.d.). Due to its broad nature, this variable may in turn capture some of the effect of more specific mechanisms, but may be helpful in that sense as I can then compare the goodness of fit of a model based on this variable as opposed to models based on the more specific mechanisms. I expect this variable to have a positive coefficient, assuming that the probability of adoption will increase as the number of restrictive policies adopted by neighboring states increases. Considering learning as a relatively broad category in and of itself, I attempt to first evaluate the learning hypothesis with Learning, a variable that counts the total number of restrictions adopted across the world between 1990 and the beginning of the year in question. Although this measure could again potentially capture some aspects of imitation or social construction, it fundamentally assumes that the higher the number of policies adopted across the world, the more information there is available to a country interested in adopting such a restriction, improving their understanding of potential consequences of the policy. Thus, I expect Learning to have a positive coefficient: as the number of restrictive policies adopted throughout the world increases, so to should the likelihood of future adoption. Distinguishing between and measuring imitation and social construction is more difficult. To do so, however, I return to an understanding of imitation as actor-centric and

41 social construction as community-centric. As discussed above, I assume that imitation 30 would primarily be driven by attempts to emulate larger or more powerful countries that have adopted similar restrictions. Different ways of operationalizing the concept of powerful or successful countries, however, may have different theoretical implications. Joseph Nye (2004, 1-2) argues, Power is like the weather. Everyone depends on it and talks about it, but few understand it.power is also like love, easier to experience than to define and measure, but no less real for that. The dictionary tells us that power is the capacity to do things. The dictionary also tells us that power means having the capabilities to affect the behavior of others to make those things happen. So more specifically, power is the ability to influence the behavior of others to get the outcomes one wants. But things are not as straightforward as they first appear. Power always depends on the context in which the relationship exists. What is the relevant context of the relationships in this study? I find it unlikely that powerful countries would deliberately exert some form of influence to convince a country to adopt restrictive religion policies that limit personal freedoms. As such, I return to a modified signaling theory as articulated above: whether intentionally or not, I argue that when powerful countries that could influence others adopt restrictions, they signal to other countries that those policies will not be challenged on the world stage. To measure this potential phenomenon, I construct variables that assess the total number of policies previously adopted by powerful countries. I suggest two possible proxies that may capture which nations would be considered powerful or influential in this sense: first, the extent to which a country controls a significant share of military and economic resources available throughout the world, and second, whether or not a country is a member of the OECD. While there is some overlap in the countries that will be considered powerful between these two categories, they represent theoretically different perspectives. The first measure approximated here using the Composite Index of National Capability developed for version five of the National Material Capabilities dataset in the Correlates of

42 War project more closely aligns with conventional definitions of hard power (Singer ). This variable totals the number of restrictive policies previously adopted by the thirtyfive countries with the highest share of power in the CINC database prior to the year in question. In contrast, OECD member countries are more likely associated with soft power leadership based on culture, values, etc. as implied in the organization s mission statement: to promote policies that will improve the economic and social well-being of people around the world by a shared commitment to market economies backed by democratic institutions and focused on the well-being of citizens ( About the OECD ). This variable totals the number of restrictive policies previously adopted by OECD member countries prior to the year in question. Given this emphasis on democratic institutions and freedoms, it would be particularly interesting from a normative perspective if countries imitate OECD members in adopting restrictive religion policies. In short, I propose two different variables that could separately assess imitation as a diffusion mechanism. For each, I expect a positive coefficient, assuming that a desire or willingness to emulate powerful countries will increase the probability of adoption. With respect to social construction, I primarily seek to test and extend the Simmons and Elkins (2004) cultural similarity hypothesis as it applies to the adoption of restrictive government religion policy. Simmons and Elkins measured cultural similarity in three ways: 1) whether or not a country shared the same dominant language, 2) whether or not the country had a common colonial heritage, or 3) whether or not a country shared the same dominant religion. Interestingly, they found the strongest support for diffusion among countries with similar religious cultures, with less conclusive results for the influence of former colonial background or dominant language.

43 32 Returning to Maske and Volden s (2011) assertion that diffusion mechanisms may vary across policies with different attributes, however, I suggest that re-testing Simmons and Elkins (2004) findings with different policy types is important for developing a better understanding of the potential impact of social construction in policy diffusion. As such, I measure cultural similarity in two ways similar to Simmons and Elkins. First, I use politicogeographic region codings from the V-Dem dataset as a proxy to group countries that have common colonial history, institutions or language (Coppedge et al. 2016). 3 Specifically, politico-geographic region was defined in the codebook as regions based on geographic proximity as well as characteristics that contribute to regional understanding as identified by scholars in studies of democratization (i.e. grouping post-communist countries in Eastern Europe and Central Asia together or grouping Australia and New Zealand in a single category with Western Europe and North America). This variable, in particular, also builds upon and refines more basic ideas of geographic diffusion. I anticipate that this variable will have a positive effect. Figure 3 highlights the percentage of total policies adopted during the time period by each politico-geographic region. Second, I develop a measure to assess the relationship between countries with the same dominant religion. This variable is constructed to count the number of previous adopters with the same religious majority as the country (i.e. the number of predominantly Sunni Muslim countries that have adopted a restriction prior to another predominantly Sunni Muslim country). 4 My codings are based on religious demographic statistics found in version 3 While taken from the V-Dem database for this project, this variable was originally developed in the Quality of Government Standard Dataset (Teorell et al. 2017). 4 Because the demographic data on religious affiliations only exists in five year increments, annual codings were inferred based on whatever the dominant religious group was at the previous measurement (i.e. if a country was predominantly Catholic in 1990 but had a higher demographic share of Other Christian in the 1995 data, it was coded as Catholic for years and Other Christian from ).

44 1.1 of the World Religions dataset through the Correlates of War project; I identify a 33 country as one of 12 categories (Catholic, Orthodox Christian, Other Christian, Jewish, Sunni Islam, Shia Islam, Other Islam, Buddhist, Hindu, Other (i.e. Confucian, Animist), Non- Religious, and No Clear Majority) based on the religious group with the highest share of the population (Maoz and Henderson 2013). I believe this measure in particular should have a relatively strong, positive influence in this case: it may be even more reasonable to assume that nations with similar religious cultures feel similarly about the relationship of religion and state (relative to Simmons and Elkins (2004) assumptions and findings that similar religious cultures may influence the adoption of economic policy). 5 5 Historians have long noted that dominant religious culture can strongly influence the relationship between religion and state. For an example of this literature, see Kalkandjieva 2011.

45 Controlling for Domestic Influences 34 To better assess the potential impact of diffusion mechanisms, I include several controls for domestic factors that may influence government religion policy. First, I construct a variable to control for the extent to which a country is democratic. In an effort to reduce the likelihood of endogeneity, I create an index variable based on several indicators in the V- Dem dataset that exclude measures otherwise often included in proxies for democracy (such as protections of civil liberties that may include measures of religious freedom or more general protections for freedom of expression and association) (Coppedge et al. 2016). Specifically, I equally weigh measures gauging the share of the adult population with suffrage, the extent to which elections are held and can be considered "free and fair," whether or not the chief executive is elected, and the extent to which the judiciary or legislature are able to act independently and exercise checks and balances over executive power. 6 I anticipate that this variable will have a negative effect (as the level of democracy increases, the likelihood of adopting a restriction falls). As previous literature has not conclusively identified the impact of a constitutional separation of religion and state (SRAS) clause on government religion policy, I separately include a control variable from the Religion and State dataset which is set equal to 1 if such a clause is present. However, I assume that this measure is more likely to influence the general type of government religion policy in a country as opposed to the extent to which a country is susceptible to the diffusion of policy, and thus anticipate that any (potentially negative) effect will likely be mild and not significant. Next, I include a control Regime Duration to account for the number of 6 For more detailed information on these individual indicators, consult the V-Dem v. 6 Codebook for the variables v2x_suffr, v2xel_frefair, v2x_accex, v2x_jucon, and v2xlg_legcon (Coppedge et al. 2016).

46 consecutive years of the current regime type in a country (Coppedge et al. 2016) Again, I expect this variable to have a negative coefficient (that is, the longer a regime type remains stable, the less likely a country will be to adopt additional restrictions). To control for economic factors, I include the growth rate of GDP per capita using data from the World Bank (2017). Finally, to account for possible time trends, I include a cubic polynomial approximation of the number of years prior to failure or the end of the observation period (Carter and Signorino 2010). 7 While taken from the V-Dem database for this project, this variable was originally developed in Boix, Miller, and Rosato (2013) and captures changes in government type (i.e. non-democracy to democracy) rather than shifts between individual leaders in a government.

47 RESULTS 36 Table 2 presents basic descriptive statistics for the variables included in analysis. Although many diffusion scholars turn to basic logistic regression for event history analysis, however, it is possible that standard logit results may be biased in this case. As reflected in the mean value of the dependent variable in Table 2 (0.042, suggesting that the outcome of interest was only observed approximately 4.2 percent of the time), this dataset may be classified as capturing rare events: out of 3325 country-year observations, only 140 failures were observed. King and Zeng (2001) note that when working with rare event data, a logit model often compresses the estimated probabilities, potentially obscuring substantive effects. They then propose a number of remedies for addressing such bias as well as reconceptualizing the substantive effect observed in original logit coefficients. As such, the regression results included in the body of this paper were generated using King and Zeng s (1999, 2001) corrections for modeling rare events (Tomz, King and Zeng 1999). For comparison, the output of standard logit models as well as graphs visualizing the changes in predicted probabilities based on the uncorrected logit are included in the appendix.

48 How are these variables related? Table 3 shows the correlation matrix between 37 different variables. Notably, the correlation matrix provides some support for my theoretical assumption that imitation and social construction may just be modified forms of learning: in particular, both ways I measure imitation are highly correlated with learning (at and respectively); the social construction variables are also moderately correlated (at and respectively). Similarly, the spatial diffusion variable (geography) and first proxy for cultural similarity (politico-geographic region) are also highly correlated (0.716). Based on both the theory and these correlation results, I do not include more than one of these related mechanisms at a time in a logit model. None of the mechanism variables appear closely correlated with the dependent variable or controls. Table 4 presents the corrected logit coefficients and standard errors of rare events models (Models 1-6) that separately assess each diffusion mechanism. Taken individually, each mechanism is statistically significant to some degree, and as expected, all had a positive effect on the odds of policy adoption.

49 38

50 39

51 40 King and Zeng (2001) suggest that one of the best ways to interpret the substantive significance of regression results in rare events models is to compare the percentage change in relative risk between different values of a variable. 8 They argue that at a baseline, even a ten to twenty percent increase in relative risk may be considered substantively important in rare event studies. Based on this assumption, these results suggest a meaningful substantive impact in addition to the statistical significance noted in Table 4. The strongest effect was associated with imitation based on representations of power in the CINC database; holding other factors constant, the change in relative risk between the 25 th and 75 th percentile was 74.1 percent. Measures of general learning and the imitation of OECD countries had the next highest impact, with a 57.9 percent change in the likelihood of adoption as more countries around the world adopted restrictive policies and a 47 percent change as the number of policies adopted by OECD members increased. As countries within the same geographic region adopted restrictive policies (assuming some form of spatial diffusion), the increase in relative risk was 30.1 percent. Finally, both measures of social construction presented the smallest (albeit still notable) differences in relative risk: The adoption of policies within the same politico-geographic region was associated with an increase the predicted probability of adoption of 21.1 percent; as countries with the same religious majority adopted restricted policies, the change in relative risk was 15.8 percent. Coefficients for the control variables largely supported the theoretical assumptions outlined above. Notably, the control measure for democracy was the strongest and only statistically significant control with a negative coefficient in all models as predicted; holding other factors constant, democracy was consistently associated with a decrease of 8 For each mechanism, I calculated the percent change in relative risk (that is, in the probability of adoption) between values at the 25th percentile and 75th percentile using the formula recommended by King and Zeng (2001): 100 * [rr 1], where rr is the relative risk calculated in STATA 12 following a rare events regression.

52 approximately 80 percent in the odds of adoption. Regime duration and economic growth 41 similarly held negative coefficients although statistically insignificant. As anticipated, a separation of religion and state clause was not a significant predictor, but maintained a positive coefficient across all six models. All in all, the calculated p values associated with Wald χ 2 statistics for all six models suggest that for each model, a null hypothesis assuming no effect of the variables included should be rejected in favor of these models that account to some extent for policy diffusion across countries. In short, these results provide moderate support for the assumption that restrictive government religion policy is diffusing across countries in some way. How should these potential diffusion mechanisms be compared? While previous diffusion literature often combines all proposed mechanisms in a final model to isolate odds ratios or statistical significance when taken together, I argue that this may not be the best approach. For instance, as noted above, mechanisms that may differ theoretically in important ways but are highly or moderately correlated (such as general measures of learning compared to modified forms of learning such as imitation of specific actors) could fail to achieve significance when assessed together in the same model. The results presented in Table 5 where Models 7-10 attempt to compare learning, imitation, or social construction mechanisms to geography provide further support for this assumption: even when less closely-related theoretical concepts (yet still moderately correlated variables) are tested together, almost all significance for diffusion mechanisms is eliminated while a control for democracy remains strongly predictive. These apparently negligible results, however, may still have substantive significance if considered outside the boundaries of conventional statistical significance (Rainey 2014). For instance, similar comparative models developed as a purely theoretical check (which include only

53 international diffusion variables without domestic controls see Table 6) continue to 42 provide statistical support for various mechanisms (Achen 2002, 2005).

54 43

55 44

56 CONCLUSION 45 Taken together, these results provide some support for the idea that global restrictions on religious freedom are not only increasing, but that the increase in policy adoption may be partially explained by international diffusion of ideas rather than exclusively occurring as a result of domestic factors. This extends the current body of literature on the relationship between religion and state and on the origins of government religion policy, which has so far focused almost entirely on internal determinants of policy. These findings also extend the diffusion literature by suggesting that mechanisms commonly proposed as independent such as learning, imitation, and social construction may be more closely interrelated than they are perhaps always treated. Yet the finding that many diffusion mechanisms identified by other scholars as largely separate can be highly interrelated complicates previous research. In short, the theoretical discussion advanced in the paper merits additional examination in future work. Are differences in findings across the literature primarily driven by differences in the operationalization and measurement of concepts? Is it really possible to accurately disentangle social construction, learning, and imitation? Or should these mechanisms be reframed to examine an interaction of factors? Can qualitative studies or case studies shed further light on these findings and help to better distinguish the roles of these mechanisms in international policy diffusion? Could scholars actually learn more about diffusion by interacting domestic and international factors (i.e. the level of democracy interacted with measures of imitation based on hard power)? Finally, what can be learned about diffusion by looking beyond simple policy adoption? In particular, I believe that examining the extent to which a policy is enforced or its duration in practice may further illuminate the strength and longevity of policy diffusion,

57 similar to Shipan and Volden s (2008) conditional and temporal effects hypotheses. All in 46 all, understanding the diffusion of restrictions of religious freedom may complicate the growing literature that has traced the diffusion of political and economic liberalism over the past several decades.

58 BIBLIOGRAPHY 47 Achen, Christopher Toward a New Political Methodology: Microfoundations and ART. Annual Review of Political Science 5: Let s Put Garbage-Can Regressions and Garbage-Can Probits Where they Belong. Conflict Management and Peace Science 22, no. 4: Akbaba, Yasemin and Jonathan Fox The Religion and State Minorities Dataset Journal of Peace Research 48, no. 6: Antoun, R. and M. Hegland (eds.) Religious Resurgence: Contemporary Cases in Islam, Christianity and Judaism. Syracuse: Syracuse University Press. Almond, G Introduction: A Functional Approach to Comparative Politics. In The Politics of the Developing Areas, edited by G. Almond and J. C. Coleman. Princeton: Princeton University Press. Apodaca, Clair Global Economic Patterns and Personal Integrity Rights after the Cold War. International Studies Quarterly 45: Apter, D The Politics of Modernization. Chicago: University of Chicago Press. Barro, Robert J., and Rachel M. McCleary Religion and Economic Growth Across Countries. American Sociological Review 68, no. 5: Which Countries Have State Religions? Quarterly Journal of Economics 120, no. 4: Beckford, J. A The Insulation and Isolation of the Sociology of Religion. Sociological Analysis 46, no. 4: Berry, Frances Stokes, and William D. Berry State Lottery Adoptions as Policy Innovations: An Event History Analysis. American Political Science Review 84, no. 2: Box-Steffensmeier, Janet M., and Christopher Zorn Duration Models for Repeated Events. Journal of Politics 64, no. 4: Boyle, Elizabeth Heger, and Sharon E. Preves National Politics as International Process: The Case of Anti-Female-Genital-Cutting Laws. Law & Society Review 34, no. 3: Brinks, Daniel, and Michael Coppedge "Diffusion is No Illusion: Neighbor Emulation in the Third Wave of Democracy." Comparative Political Studies 39, no.4:

59 Buckley, David and Luis Felipe Mantilla God and Governance: Development, State Capacity, and the Regulation of Religion. Journal for the Scientific Study of Religion 52, no. 2: Bueno de Mesquita, Bruce, Alastair Smith, James Morrow, and Randolph Siverson The Logic of Political Survival. Cambridge: MIT Press. Bueno de Mesquita, Bruce, George Downs, Alastair Smith, and Feryal Marie Cherif Thinking Inside the Box: A Closer Look at Democracy and Human Rights. International Studies Quarterly 49: Bunce, Valerie J., and Sharon L. Wolchik Defeating Authoritarian Leaders in Postcommunist Countries. New York: Cambridge University Press, Casanova, J Public Religions in the Modern World. London: University of Chicago Press. Carpenter, R. Charli "Vetting the Advocacy Agenda: Network Centrality and the Paradox of Weapons Norms." International Organization 65, no. 1: Carter, David, and Signorino, Curtis Back to the Future: Modeling Time Dependence in Binary Data. Political Analysis 18, no. 3: Chilton, Adam and Mila Versteeg Do Constitutional Rights Make a Difference? Coase-Sandor Institute for Law and Economics Working Paper No Collier, Ruth Berins Paths Toward Democracy: The Working Class and Elites in Western Europe and South America. New York: Cambridge University Press. Collier, David, and Richard E. Messick Prerequisites versus Diffusion: Testing Alternative Explanations of Social Security Adoption. American Political Science Review 69: Coppedge, Michael et al V-Dem Codebook v6. Varieties of Democracy (VDem) Project. Davenport, Christian Multi-Dimensional Threat Perception and State Repression: An Inquiry into Why States Apply Negative Sanctions. American Journal of Political Science 39, no. 3: Human Rights and the Democratic Proposition. Journal of Conflict Resolution 43, no. 1: Deutsch, K Nationalism and Social Communication. Cambridge, MA: MIT Press. Dobbin, Frank, Beth Simmons, and Geoffery Garrett The Global Diffusion of Public Policies: Social Construction, Coercion, Competition, or Learning? Annual Review of Sociology 33:

60 Donno, Daniela Who is Punished? Regional Intergovernmental Organizations and the Enforcement of Democratic Norms. International Organization 64: Driessen, Michael Religion, State, and Democracy: Analyzing Two Dimensions of Church-State Arrangements. Politics and Religion 3, no. 1: Fearon, James and David Laitin Ethnicity, Insurgency, and Civil War. The American Political Science Review 97, no. 1: Feldmann, Horst Protestantism, Labor Force Participation, and Employment across Countries. American Journal of Economics and Sociology 66, no. 4: Finke, Roger "Religious deregulation: Origins and consequences." Journal of Church and State 32, no. 3: "Presidential Address Origins and Consequences of Religious Freedoms: A Global Overview." Sociology of Religion 74, no. 3: Finke, Roger, and Robert R. Martin "Ensuring Liberties: Understanding State Restrictions on Religious Freedoms." Journal for the Scientific Study of Religion 53, no. 4: Forsythe, David P The Internationalization of Human Rights. Lexington, Mass.: Lexington Books. Fox, Jonathan Paradigm Lost: Huntington's Unfulfilled Clash of Civilizations Prediction into the 21st Century. International Politics 42: World Separation of Religion and State into the 21st Century. Comparative Political Studies 39, no. 5: Do Democracies Have Separation of Religion and State? Canadian Journal of Political Science 40, no. 1: Quantifying Religion and State: Round Two of the Religion and State Project. Politics and Religion 2, no. 3: a. Out of Sync: The Disconnect Between Constitutional Clauses and State Legislation on Religion. Canadian Journal of Political Science 44, no. 1: b. Building Composite Measures of Religion and State. Interdisciplinary Journal of Research on Religion 7, no. 8: An Introduction to Religion and Politics: Theory and Practice. New York: Routledge.

61 Political Secularism, Religion, and the State: A Time Series Analysis of Worldwide Data. New York: Cambridge University Press. 50 Fox, Jonathan and Deborah Flores Religions, Constitutions, and the State: a Cross- National Study. Journal of Politics 71, no. 4: Franzese, Robert J., Jr., and Jude C. Hays Interdependence in Comparative Politics: Substance, Theory, Empirics, Substance. Comparative Political Studies 41, no. 4/5: Garrett, Geoffrey Global Markets and National Politics: Collision Course or Virtuous Circle? International Organization 52, no. 4: Globalization and Government Spending Around the World. Studies in Comparative International Development 35, no. 4: Garrett, Geoffrey, and Deborah Mitchell Globalization, Government Spending and Taxation in the OECD. European Journal of Political Research 39, no. 2: Garrett, Kristin, and Joshua Jansa Interest Group Influence in Policy Diffusion Networks. State Politics and Policy Quarterly 15, no. 3: Gilardi, Fabrizio Who Learns from What in Policy Diffusion Processes? American Journal of Political Science 54, no. 3: Transnational Diffusion: Norms, Ideas and Policies. In Handbook of International Relations, edited by Walter Carlsnaes, Thomas Risse and Beth Simmons, Thousand Oaks: SAGE Publications. Gill, Anthony The Political Origins of Religious Liberty. New York: Cambridge University Press "Religious Liberty and Economic Development: Exploring the Causal Connections." The Review of Faith & International Affairs 11, no. 4: Graham, Erin, Charles Shipan and Craig Volden The Diffusion of Policy Diffusion Research in Political Science. British Journal of Political Science 43, no. 3: Gray, Virginia Innovation in the States: A Diffusion Study. American Political Science Review 67: Grim, Brian J., Greg Clark, and Robert Edward Snyder "Is Religious Freedom Good for Business? A Conceptual and Empirical Analysis." Interdisciplinary Journal of Research on Religion 10, no. 4: np. Grim, Brian, and Roger Finke The Price of Freedom Denied: Religious Persecution and Conflict in the 21 st Century. New York: Cambridge University Press.

62 51 Guiso, Luigi, Paola Sapienza, and Luigi Zingales, People s Opium? Religion and Economic Attitudes. Journal of Monetary Economics 50, no. 1: Helbling, Marc and Richard Traunmuller How State Support of Religion Shapes Attitudes Toward Muslim Immigrants: New Evidence From a Sub-National Comparison. Comparative Political Studies 49, no. 3: Henderson, E. A., & Tucker, R Clear and Present Strangers: The Clash of Civilizations and International Conflict. International Studies Quarterly 45: Hofferbert, Richard, and David Cingranelli Democratic Institutions and Respect for Human Rights. In Human Rights and Developing Countries, edited by David Cingranelli. Greenwich: JAI Press. Horowitz, Michael C "Nonstate Actors and the Diffusion of Innovations: The Case of Suicide Terrorism." International Organization 64, no. 1: Hulsether, Mark Religion and Culture. In The Routledge Companion to the Study of Religion, edited by John Hinnells, New York: Routledge. Huntington, Samuel P The Third Wave: Democratization in the Late Twentieth Century. Norman: University of Oklahoma Press The Clash of Civilizations and the Remaking of the World Order. New York: Simon and Schuster. Hurd, Elizabeth Shackman Beyond Religious Freedom: The New Global Politics of Religion. Princeton: Princeton University Press. Hyde, Susan D "Catch Us If You Can: Election Monitoring and International Norm Diffusion." American Journal of Political Science 55, no. 2: Iannccone, Laurence Introduction to the Economics of Religion. Journal of Economic Literature 36, no. 3: Kahneman, Daniel Thinking, Fast and Slow. New York: Farrar, Straus and Giroux. Kalkandjieva, Daniela A Comparative Analysis on Church-State Relations in Eastern Orthodoxy: Concepts, Models, and Principles. Journal of Church & State 53, no. 4: Karch, Andrew Emerging Issues and Future Directions in State Policy Diffusion Research. State Politics and Policy Quarterly 7, no. 1: Kaufman, Robert R., and Alex Segura-Ubiergo Globalization, Domestic Politics, and

63 Social Spending in Latin America: A Time-Series Cross-Section Analysis, World Politics 53, no. 4: Keith, Linda Camp Constitutional Provisions for Individual Human Rights : Are They More than Mere Window Dressing? Political Research Quarterly 55: Kepel, G The Revenge of God: The Resurgence of Islam, Christianity and Judaism in the Modern World. Cambridge: Polity Press. King, Gary and Langche Zeng "Estimating Absolute, Relative, and Attributable Risks in Case-Control Studies," Department of Government, Harvard University, available from Logistic Regression in Rare Events Data. Political Analysis 9: Landes, David The Wealth and Poverty of Nations: Why Some Are So Rich and Some So Poor. New York: W. W. Norton. Latest Trends in Religious Restrictions and Hostilities Pew Research Center. Accessed November 19, Linz, Juan J., and Alfred C. Stepan Problems of Democratic Transition and Consolidation: Southern Europe, South America, and Post-Communist Europe. Baltimore, Md.: Johns Hopkins University Press. Lutz, James M Regional Leadership Patterns in the Diffusion of Public Policies. American Politics Quarterly 15: Maoz, Zeev and Errol A. Henderson The World Religion Dataset, : Logic, Estimates, and Trends. International Interactions 39: Maske, Todd and Craig Volden The Role of Policy Attributes in the Diffusion of Innovations. The Journal of Politics 73, no. 1: McLaren, Lauren and Mark Johnson Resources, Group Conflict and Symbols: Explaining Anti-Immigration Hostility in Britain. Political Studies 55, no. 4: Mooney, Christopher Z Modeling Regional Effects on State Policy Diffusion. Political Research Quarterly 54: Mooney, Christopher Z., and Mei-Hsien Lee Legislating Morality in the American States: The Case of Pre-Roe Abortion Regulation Reform. American Journal of Political Science 39: Most, Benjamin A. and Harvey Starr Theoretical and Logical Issues in the Study of

64 International Diffusion. Journal of Theoretical Politics 2, no. 4: Mungiu-Pippidi, A., & Mindruta, D Was Huntington Right? Testing Cultural Legacies and the Civilization Border. International Politics 39, no. 2: Nongbri, Brent "Dislodging "Embedded" Religion: A Brief Note on a Scholarly Trope," Numen: International Review for the History of Religions 55, no. 4: Norris, P., & Inglehart, R Sacred and Secular: Religion and Politics Worldwide. New York: Cambridge University Press. Nye, Joseph S Soft Power: The Means to Success in World Politics. New York: PublicAffairs. O Donnell, Guillermo A., Philippe C. Schmitter, and Laurence Whitehead, eds Transitions from Authoritarian Rule. Baltimore, Md.: Johns Hopkins University Press. OECD "About the OECD." Accessed April 18, O Loughlin, John et al The Diffusion of Democracy, Annals of the Association of American Geographers 88, no. 4: Omer, Atalia, Scott Appleby and David Little The Oxford Handbook of Religion, Conflict, and Peacebuilding. Oxford: Oxford University Press. Philpott, Daniel "Has the Study of Global Politics Found Religion?." Annual Review of Political Science 12: Poe, Steven C., and C. Neal Tate Repression of Human Rights to Personal Integrity in the 1980s: A Global Analysis. American Political Science Review 88, no. 4: Poe, Steven C., C. Neal Tate, and Linda Camp Keith Repression of the Human Right to Personal Integrity Revisited: A Global Cross-National Study Covering the Years International Studies Quarterly 43, no. 2: Przeworski, Adam, Michael E. Alvarez, Jose Antonio Cheibub, and Fernando Limongi Democracy and Development: Political Institutions and Well-Being in the World, Cambridge: Cambridge University Press. Rainey, Carlisle Arguing for a Negligible Effect. American Journal of Political Science 58, no. 4: Ramirez, Francisco O., and Elizabeth H. McEneaney From Women s Suffrage to Reproduction Rights? Cross-National Comparisons. International Journal of Comparative Sociology 38, no. 1 2: 6 24.

65 Rudra, Nita Globalization and the Decline of the Welfare State in Less-Developed Countries. International Organization 56, no. 2: Russett, B. M., Oneal, J. R., Cox, M "Clash of Civilizations, or Realism and Liberalism Déjà Vu? Some Evidence." Journal of Peace Research 37: Sahliyeh, E. (ed.) Religious Resurgence and Politics in the Modern World. Albany: State University of New York Press. Sanghani, Radhika Burka Bans: The Countries Where Muslim Women Can t Wear Veils. The Telegraph. Accessed November 19, Sarkissian, Ani Religious Regulation and the Muslim Democracy Gap. Politics and Religion 5, no. 3: The Varieties of Religious Repression: Why Governments Restrict Religion. Oxford: Oxford University Press. Seiple, Chris, Dennis Hoover and Pauletta Otis The Routledge Handbook of Religion and Security. New York: Routledge. Shipan, Charles R., and Craig Volden Bottom-Up Federalism: The Diffusion of Antismoking Policies from U.S. Cities to States. American Journal of Political Science 50, no. 4: The Mechanisms of Policy Diffusion. American Journal of Political Science 52, no. 4: Policy Diffusion: Seven Lessons for Scholars and Practitioners. Public Administration Review 72, no. 6: Simmons, Beth A., Frank Dobbin, and Geoffrey Garrett Introduction: The International Diffusion of Liberalism. International Organization 60, no. 4: Simmons, Beth A., and Zachary Elkins The Globalization of Liberalization: Policy Diffusion in the International Political Economy. American Political Science Review 98, no. 1: Singer, J. David "Reconstructing the Correlates of War Dataset on Material Capabilities of States, " International Interactions 14: Smith, D. E Religion and Political Development. Boston: Little, Brown Religion and Political Modernization. New Haven, CT: Yale University Press. 54

66 Stark, Rodney, and Roger Finke Acts of Faith: Explaining the Human Side of Religion. Berkeley: University of California Press. 55 Starr, Harvey Democratic Dominoes: Diffusion Approaches to the Spread of Democracy in the International System. Journal of Conflict Resolution 35, no. 2: Teorell, Jan, et al The Quality of Government Standard Dataset, version Jan17. University of Gothenburg: The Quality of Government Institute, Tomz, Michael, Gary King, and Langche Zeng RELOGIT: Rare Events Logistic Regression, Version 1.1 Cambridge, MA: Harvard University, October 1, Trends in Global Restrictions on Religion Pew Research Center. Accessed November 19, Tversky, Amos and Daniel Kahneman Judgment Under Uncertainty: Heuristics and Bias. Science 185, no. 4157: Volden, Craig, Michael M. Ting and Daniel P. Carpenter A Formal Model of Learning and Policy Diffusion. American Political Science Review 102, no. 3: Walker, Jack L The Diffusion of Innovations among the American States. American Political Science Review 63: Weber, Hannes National and Regional Proportion of Immigrants and Perceived Threat of Immigration: A Three-Level Analysis in Western Europe. International Journal of Comparative Sociology 56, no. 2: Weber, Max The Protestant Ethic and the Spirit of Capitalism. 2nd edition. New York: Routledge, Wei, L. J., D. Y. Lin, and L. Weissfeld Regression Analysis of Multivariate Incomplete Failure Time Data by Modeling Marginal Distributions. Journal of the American Statistical Association 84, no. 408: Wendt, Alexander Anarchy is What States Make of It: The Social Construction of Power Politics. International Organization 46, no. 2: Westerlund, D. (ed.) Questioning the Secular State: The Worldwide Resurgence of Religion in Politics. London: Hurst & Co. Westhus, K The Church in Opposition. Sociological Analysis 37, no. 4:

67 Weyland, Kurt Theories of Policy Diffusion: Lessons from Latin American Pension Reform. World Politics 57: Bounded Rationality and Policy Diffusion: Social Sector Reform in Latin America. Princeton: Princeton University Press The Diffusion of Revolution: 1848 in Europe and Latin America. International Organization 63: The Diffusion of Regime Contention in European Democratization, Comparative Political Studies 43, no. 8-9: The Arab Spring: Why the Surprising Similarities with the Revolutionary Wave of 1848? Perspectives on Politics 10, no. 4: Wilson, B. R Religion in Secular Society. Baltimore: Penguin Religion in Sociological Perspective. Oxford, UK: Oxford University Press. The World Bank World Development Indicators. Washington, D.C.: The World Bank (producer and distributor). Wotipka, Christine Min and Francisco Ramirez World Society and Human Rights: an Event History Analysis of the Convention on the Elimination of All Forms of Discrimination Against Women. The Global Diffusion of Markets and Democracy, ed. Beth Simmons, Frank Dobbin, and Geoffrey Garrett. New York: Cambridge University Press, USCIRF (United States Commission on International Religious Freedom) Annual Report. Accessed November 19, f b. China: USCIRF Condemns the Government s Ramadan Restrictions. Accessed November 19, c. Russia: USCIRF Condemns Enactment of Anti-Terrorism Laws. Accessed November 19, United Nations Universal Declaration of Human Rights United Nations. Article 18. Accessed November 19, United Nations. n.d. "UNSD Methodology." Accessed April 18,

68 APPENDIX LOGIT TABLES AND FIGURES 57

69 The following tables and figures present the uncorrected (standard) logistic 58 regression results corresponding to those reported in the paper above. Table 7 compares with Table 4; Figures 4 through 9 plot the change in predicted probabilities associated with the logistic models estimated in Table 7. Tables 8 and 9 correspond with Tables 5 and 6 in the main text, respectively. The uncorrected results may be summarized as follows: Some of the strongest effects were associated with the imitation of OECD countries and spatial diffusion: holding other factors constant, the odds of adoption increased by 10.8 percent and the predicted probability of adoption increased from.036 to.078 when OECD countries had previously adopted restrictive policies. Similarly, the predicted probability of adoption increased from.037 to.093 as countries within the same geographic region adopted restrictive policies and the odds of adoption increased, on average, by 3.5 percent. Next, the impact of imitation based on representations of power in the CINC database was an increase in the predicted probability of adoption from.028 to.076; here, however, the percentage change in the odds of adoption was 2.8 percent. Both proxies of social construction produced similar results: the adoption of policies within the same politico-geographic region was associated with an increase the predicted probability of adoption from.037 to.071 and a 1.2 percent change in the odds. As countries with the same religious majority adopted restricted policies, the odds of adoption increased by 1.1 percent and the predicted probability of adoption rose from.04 to.086. Finally, as countries learn from the adoption of policies worldwide, the predicted probability of adoption increased from.030 to.074 and odds of adoption only increase by roughly.5 percent.

70 59

71 60

72 61

73 62