Religiosity and Bellicosity: The Impact of Religious
Transcription
Religiosity and Bellicosity: The Impact of Religious
Religiosity and Bellicosity: The Impact of Religious Commitment on Patterns of Interstate Conflict Kathryn J. Alexander1 September 18, 2015 1 Department of Political Science, Duke University. Abstract Are states’ international conflict behaviors at all connected to the levels of religious commitment within their populations? Existing quantitative work within the area of religion and international relations tends to employ shallow conceptualizations of “religious” variables, overemphasizing superficial demographic indicators rather than proposing more robust measurements. This paper takes a first step toward resolving the issues presented in the literature by using a measurement of the level of religious commitment within a state–that is, the proportion of a state’s population to which religious beliefs and practices are important–to explore whether commitment may be a scope condition for religion’s role in international relations, more generally. Using both monadic and dyadic model specifications, the analysis uncovers a link between increasing levels of religious commitment within states and their propensity to initiate conflictual behavior with other states. In addition, this finding is also linked to the conditions under which states with different majority religions are more likely to initiate conflict with one another, indicating a need to understand the dynamics of religious commitment itself in order to more clearly unpack the processes that lead to interstate conflict. 1 Introduction In the popular mind, particularly in the post-9/11 international political climate, religion is often credited with playing a powerful role in shaping how countries relate to one another. From Iran’s theocratic regime and the rise of the Islamic State to the religious rhetoric of parades of American presidents and tensions between religious and secular identities in states like Israel and Turkey, issues tied to religion seem to consistently dominate news cycles and discussions of international affairs. The very prevalence of such issues seems enough to prompt the casual observer, at least, to conclude that religion is a factor of great significance on the world stage. This instinct is corroborated by more than just headlines, however. Perhaps now more than ever, religion remains a prominent dimension of modern life for billions of people; indeed, the global population of religious people is growing (Inglehart and Norris, 2004). When considering major factors that could condition political behavior, it is only prudent to give religion serious consideration. Yet, this conclusion has not traditionally been shared or supported by mainstream international relations work, particularly by its most influential theories. International relations scholarship prior to the end of the Cold War tended to assume that religion had no role in the international system, a view shaped by the continuing influence of the secularism of the Enlightenment and nineteenth century modernization (Fox and Sandler, 2004; Hurd, 2004; Philpott, 2009, 2013). Philpott (2009) explains: The dominant theories in this field assume that the states, nations, international organizations, parties, classes, businesses, interest groups, nongovernmental organizations, and lobbies that carry on politics pursue ends that include power, conquest, freedom, wealth, a redistribution of wealth, welfare provision, human rights, justice, envinternational relationsonmental cleanliness, and other goals, but they do not pursue religious ends and are not influenced by religious actors. Such theories reason as if religion has disappeared from politics. (187) Consequently, write Fox and Sandler (2004), “These powerful intellectual perspectives 2 transformed into the twentieth-century social science disciplines of sociology, political science, and psychology, resulting in a growing body of literature that hypothesized the irrelevance of religion” (23). Despite these challenges, a niche research literature has still managed to emerge which specifically emphasizes religion’s relationship(s) to international relations, and in recent years interest has even sharply increased (Hassner, 2011, 37), no doubt spurred on in great part by the aforementioned popular association of religion with global affairs. This silver lining is somewhat undermined, however, by the prevalence of some systematic methodological issues within the body of existing work, chief among them the matter of how religion is generally conceptualized and measured by recent studies: either too superficially or too contextually. Hassner (2011) provides a framework for conceptualizing these twin problems by drawing a distinction between what he terms broad and deep approaches to the study of religion and international relations, emphasizing the current existence of a tradeoff between generalizability and rigor in the literature. In Hassner’s classification, broad approaches emphasize religious behavior’s effects on the international sphere but treat the concept of ”religion” itself superficially: “These authors are interested in the effects of religion on international conflict and cooperation, diplomacy, or globalization, but they have been hesitant to trace these effects to their origins in religion itself” (46). Conversely, deep approaches employ rigorous contextualization in order to understand particular movements or religious traditions and offer in-depth insights as to their intersection with political processes, yet these works tend to be so focused on particularities as to have no systematic implications. Hassner’s solution to the broad/deep divide is what he calls thick religion, an approach that incorporates traits of both deep and broad scholarship: Deep, in that it traces the pathways by which religion affects international affairs to their origins in the content and meaning of religion, and broad, in that it offers generalizable implications across states and regions. A thick religion approach to the study of religion and international politics requinternational relationses an understanding of religious detail but also a willingness to gen3 eralize from particular religious movements, regions, or instances to arrive at broader conclusions. (49) The development of research agendas and methodologies that meet the qualifications of thick scholarship may appear a tall order, but it is by no means an unattainable goal. The spirit of Hassner’s call asks scholars of religion and international relations to take context seriously while retaining a willingness to generalize where appropriate. Such is the motivation of the present study, the primary objective of which is to propose a potential step forward in applying a more conceptually robust understanding of religious variables to large-N, quantitative studies of international relations. Specifically, this paper explores the connection between levels of religious commitment within states and their propensity for initiating international conflicts, moving away from the more common, broad practice of focusing exclusively on the religious traditions to which majorities of state populations belong. 2 Religious Commitment: A Potential Scope Condition? One of the most striking components of Hassner’s proposal for scholarship is his admonishment of the current quality of both quantitative, large-N comparative studies and in-depth qualitative ones. A primary way in which quantitative methods, in particular, can be helpful in advancing quality work in religion and international relations is by empirically identifying certain conditions under which religious factors can be tied to international phenomena. If scholars of religion and international relations hope to demonstrate religion’s importance to their more skeptical colleagues and see religious factors’ future inclusion in more mainstream international relations work, the identification of such scope conditions should be of high priority in their research, for to simply assert that “religion matters” to international relations is an unhelpful exercise. Religion, while an important factor in the lives of many, does not matter to everyone. Critics of religion’s inclusion in most international relations studies are not without ammunition: for 4 every international event with even a passing religious connection, there are numerous instances of interactions between states in the international system in which no evidence appears to justify even a minimal role for the consideration of religion, even in countries with ostensibly “religious” domestic regimes (Shaffer, 2006). Rather than asking whether religion matters to international relations, we should instead interrogate the conditions surrounding when and how it matters. To identify some potentially interesting scope conditions that will be worth exploring, it is necessary to ask a rather basic, but fundamental question: what is special about “religion” as a factor that can be relevant to international affairs? According to Fox (2001), there are three major categories of mechanisms via which religion’s impact on international politics can be identified. first, decision making and/or policymaking can be influenced by the beliefs of leaders and/or constituents. Here, religion acts as part of the process that yields actual policies, whether through leaders or the society they represent. Second, religion can be a source of legitimacy. In this mechanism, religion’s influence is not necessarily in informing policies but is instead used as a means by which to legitimize and defend them. Finally, domestic religious issues, including conflict, can spread or otherwise become international issues. This paper focuses on a potential component of Fox’s first category, the beliefs of leaders and constituents, positing that there are two primary roles for religion at the individual level, in particular, that could make it pertinent to how states relate to one another. first, religion as a source of ideas can introduce preference structures and incentives that differ from and compete with default material interests in foreign policy issues. For example, individuals holding particular religious views may find their time horizons and/or payoff structures altered when it comes to particular events and policies: “The logic is simple: the physical self is mortal, and hence temporary; the religious self, however, is potentially immortal and eternal. Thus, sacrificing the temporary and mortal to obtain the eternal and immortal is not only rational but also desirable” (Toft, 2007, 100). As Lumsdaine (1993) explains, some actions in international relations–such as the distribution of foreign aid–can run completely contrary to state material interests: 5 A view of human action that sees only self-interest is far too simple. It errs in leaving out the dark side of human character, often astonishingly powerful, as well as in ignoring the strength of compassionate feelings, or hatred of injustice. Principled refusals to do wrong, and acts of love and compassion, are common, as are folly, unnecessary hatred and domineering, and self-defeating behavior. Human beings are a mixture of self-interest, idealism, and pointless destructiveness. All three elements operate, in varying proportions, in civil society and politics and in international affairs as well as in the life of the individual. (9) Though Lumsdaine does not emphasize the role of religion in conditioning the above motivations and behaviors, it is key to note that religious beliefs can provide concrete motivations for behaviors unmotivated by self-interest. Not all unselfish behaviors are necessarily motivated by religious belief, but morality is unquestionably an enormous part of almost every religious tradition, and with a world population that is only getting more religious, the relevance of morality in international relations is guaranteed to show overlap with the relevance of religion in international relations. Religious individuals may also possess certain perceptions of whether other actors should be classified as friends or enemies because of religiously-prescribed concepts like empathy, altruism, or divine justice. This is to say nothing of the powerful possibilities for in-group/out-group identity construction created by perceptions of shared religious beliefs, which may condition how states perceive threats from “outsiders” (Wendt, 1999; Rousseau and Garcia-Retamero, 2007). Shared religious ties and participation in religious communities may also contribute to the building of social capital within societies, which Putnam (1993) and Helliwell and Putnam (1995) tie to the quality of government performance. Second, religious beliefs may contribute to cognitive biases when situations of international interaction arise. This can take place via skewed perceptions of the likelihood of victory, actors’ resolve, or how intervening information should be interpreted. As Akbaba and Taydas (2011) point out, “Religious views are non-relativistic, meaning that believers insist that what they believe is true and hence they do not accept alterna6 tive answers” (162). Furthermore, religion can be conceptually distinguished from other ideational forces that might provide competing accounts. Toft, Philpott and Shah (2011) argue that “other forms of belief and belonging,” such as nationalism, are conceptually distinct from religion: Nationalism and religion are not the same. Even where some religions are nontheistic (as in the cases of Theravada Buddhism, Confucianism, and Jainism), all religions by definition seek understanding of, and harmony with, the widest reaches of transcendent reality–the quality that distinguishes them from political ideologies such as Marxism or secular nationalism that are sometimes thought to be functionally equivalent to religion. Religions offer answers to universal questions about the origins of existence, the afterlife, and realities that transcend humanity; nations generally do not. (21) Religion offers answers to questions unsatisfied by other elements of human society, particularly regarding why life exists and the nature of its ultimate meaning. While even strong ideologies such as nationalism and religiously-divorced systems of morality may provide answers to questions of how humans should live with regard to themselves and one another, religion is the term we can apportion to these questions of why. Religions tend to provide answers to the “how” questions as well, but these answers are conditioned upon how the “whys” were first formulated. Other types of ideological systems lack this initial foundation. It is because religion deals with these most central questions of human existence that it is able to condition the behaviors and outlooks outlined above. Religion can be reasonably regarded as a unique force with the potential to shape people’s views on politics, including perceptions of how the states in which they live (or that they govern) should conduct themselves on the international stage. The key word here, however, is potential. The cognitive perspectives provided by religious belief, behavior, and belonging will vary across individuals as a function of how prominently the perspectives feature vis--vis other interests and values. We cannot expect religious factors to impact anyone’s views on foreign policy if religion itself is not actually salient to the actors in whom we are interested. Put differently, for religion to provide an alternative to 7 baseline–let us say realpolitik–preferences for policy, individuals must actually consider it relevant. The notion that religion’s importance to individuals would be directly tied to the formation of their policy preferences is rooted in conclusions from psychology: “According to theories of cognitive consistency, those for whom religious beliefs are highly salient should feel greater pressure to bring their political attitudes into congruence with their religious convictions than those for whom such beliefs are less salient” (Smidt, Kellstedt and Guth, 2009, 20). Capturing this dimension of “religiosity,” as some have called it, or “religious commitment,” necessitates an understanding and measurement of more than just straightforward religious affiliation. This crucial connection is particularly overlooked in most of the existing quantitative scholarship dealing with the topic of “religion” in international relations. Existing work tends to operationalize religious variables with rudimentary indicators of whether a particular faith is dominant within a state (Hassner, 2011). These are measured for statistical models usually by the percentage of the population that identifies as belonging to a broad faith tradition, such as “Christianity” or “Islam,” or a dummy variable to indicate the religion to which ascribes the majority of a state’s population. The issue with exclusively relying on either proxy is that simply noting the demographic fact that people identify with a particular faith does not say anything about the degree to which that faith actually matters to them and, therefore, could be a force in shaping political preferences. This is not to say that such studies are not valuable, only that they inevitably only explore one dimension of religion’s potential political impact and paint an incomplete picture, one that overwhelmingly dominates cross-national studies in international relations. This project therefore follows the example of a growing number of scholars in the field of American politics in attempting to conceptualize and measure religious commitment for cross-national comparison. Within the American case, individuals with higher levels of religious commitment have been linked to the highest levels of support for removing Saddam Hussein from power and invading Iraq, as well as higher likelihoods to view Islam as encouraging more violence than other religions (Smidt, 2005). Religious commitment 8 is also associated with higher levels of voter turnout and support for H.W. Bush in the 1988 election. And, perhaps most interestingly, the least committed groups within each religious tradition are shown to have similar preferences (Kellstedt et al., 1996). In other words, at low levels of commitment, differences between nominal religious traditions do not appear to impact political views. Could such case-specific findings have any international implications? Can religious commitment be linked at all to not only differences in individual policy preferences, but also to patterns of interstate interactions? For example, could we expect states with more invasion-supporting, committed populations to be more prone to undertaking invasion, overall? If policy preferences appear to converge between religious traditions at the lower ends of the religious commitment spectrum, does that mean that states with lower levels of religious commitment may be more cooperative with one another than states with higher levels? By operationalizing a measurement for states’ overall levels of religious commitment, it may be possible to sketch the outline of a potential scope condition for religion’s role in international relations beyond the American case. This paper’s analysis explores the possibility that these findings could resonate across different state contexts and have implications for how states interact with one another. Because the study functions largely as a plausibility probe on this front, emphasis is placed on determining whether any empirical link exists between religious commitment and interstate interactions, not necessarily on testing expectations about what such a link–if it does exist–looks like. Indeed, while popular expectations tend to conflate religion and conflict, it is important to note alternative narratives that instead tie religious factors to peaceful and positive international outcomes.1 For this reason, the analysis is broken into four parts, each addressing a particular question about the potential relationship between levels of religious commitment and states’ propensity for bellicose (conflictual) behavior. These questions are as follows: 1 These points of view tend to emphasize the sacredness of human life revealed in so many faith traditions, as well as values that promote not only peaceful but positive relationships with other human beings (Thompson, 1990; Templeton, 1999). Faith can also provide an emphasis on restoration of right relationship to facilitate reconciliation, countering what might be otherwise bellicose impulses (Philpott, 2007b,a, 2009). These accounts might therefore lead to expectations for higher levels of religious commitment to be associated with preferences for less bellicose, more peaceful foreign interactions. 9 I. Is a state’s level of religious commitment at all connected to its overall likelihood to initiate confict with other states? II. Is the likelihood that a state will initiate conflicts against a state with a different majority religion affected by the initiating state’s domestic level of religious commitment? III. Does the level of religious commitment within State B in a dyad appear to condition the impact of State A’s level of religious commitment on its likelihood to initiate conflict with State B? IV. Does the joint level of religious commitment between State A and State B in a dyad (level of commitment of State A * level of commitment of State B) appear to condition whether State A is more likely to initiate conflict when State B has the same/different majority religion? The next sections present the research design for the analysis and a discussion of the results, addressing each of the above questions. These are followed by an exploration of some theoretical narratives that could potentially explain the findings. 3 3.1 Research Design Data Short of peering inside of someone’s brain, how might one attempt to determine an individual’s level of religious commitment? This study uses two strategies: looking at religious behaviors and self-evaluations of religiosity. Religious behavior refers to actions undertaken by an individual associated with observance of a particular faith–attending religious services outside of weddings, funerals, and such special occasions, for example. Religious behavior can be reasonably assumed to track with religious commitment because of the costs of time and energy associated with participation in behaviors versus simple identification. On the other hand, a self-evaluation of religiosity is an assessment undertaken by an individual regarding the importance of religion to his or her life. 10 Ideally, at least one cross-national survey asking about both religious behavior and religiosity would be administered in every country, every year, but in reality, such surveys are far more scarce, and most tend to ask about either behavior or religiosity, if they ask about them at all. To take advantage of as much survey coverage as possible for the purposes of this study, I standardize responses to both types of questions, using them to generate a generalized measure for respondents’ ”religious commitment.” This procedure is consistent with standard practices in the sociology of religion (Kellstedt et al., 1996, 177), and strategies to mitigate some of the potentially problematic elements of this approach are elaborated upon below and in the Methods subsection. Coding rules for this standardization are presented in Appendix A. It is also necessary to acknowledge that these measurements of religious behavior and self-evaluations of religiosity have been selfreported in surveys. As with any data that relies upon individuals relating information about themselves, we should be appropriately aware of results that may carry a reporting bias. This is no reason to eschew survey data, but it is a well-debated caveat. To generate the primary variable of interest for my analyses, individual-level data was drawn from multiple cross-national surveys, including the World Values Survey, the International Social Survey Programme, Latinobaromtro, Arab Barometer, Asian Barometer, Caucuses Barometer, Afrobarometer, and a number of smaller, one-wave survey projects. This study, in fact, brings together more survey sources than any previous scholarship on these themes.2 As a result, one of the major contributions of this project is the provision of better data coverage and greater confidence in the generalizability of results than preceding work. From the individual-level response data, responses were aggregated by country-year to deliver the total of respondents who fall within each of four categories of religious commitment: “very religious,” “somewhat religious,” “not very religious,” and “not at all religious.” Each total was then divided by the total number of respondents for the countryyear (including individuals who elected not to answer or had their response omitted for some other reason) to deliver a proportion of total respondents. Then, taking into account 2 Previous work has tended to rely on one of these surveys at a time, so they have been either regionally-focused or limited in their coverage. 11 the potentially vague conceptual boundaries between ”very” and ”somewhat” religious, the proportion totals of these categories were combined, resulting in a measurement of the proportion of a country’s population in a given year reporting to be at least reasonably religiously committed. If multiple surveys were available for a country for a particular year, their yielded proportions were averaged. Finally, because the study utilizes a cross-sectional methodological approach with a dependent variable that is a count over a particular time period, each country’s proportions of committed respondents were averaged for a single commitment measurement. Therefore, a state’s level of religious commitment consists of the average proportion of its population that self-identifies as “very religious” or “somewhat religious” over the time period in question. Different religious traditions can constitute vastly different constellations of beliefs and, by association, different practices that might be considered applicable as ”religious behaviors.” This could accordingly bias any results toward explaining the religious commitment of adherents to religions like Christianity and Islam, which have more clear-cut prescriptions of observable religious behaviors than Buddhism. This issue is at least somewhat addressed by the question wording of surveys given in different areas. The Asian Barometer, for example, asks, ”About how often do you practice religious services or rituals these days?” which allows for a variety of behaviors in accordance with the majority religions of East Asia. However, to more deliberately control for what might constitute different expressions of religious commitment across faiths, I include a variable in the model that indicates the majority religion of a state, taken from the International Religious Freedom Dataset provided by the Association of Religion Data Archives. The inclusion of this control also enables the analysis to build upon the existing literature in religion and international relations that uses such dummy variables as their only religious indicators, evaluating whether previously discovered effects (or lacks thereof) linking religion to interstate conflict may in fact be overlooking the impact of the degree to which religion is actually salient to state populations. In the monadic analysis, this variable manifests as a dummy variable indicating the broader religious tradition to which the majority of a country’s population ascribes, while the dyadic analysis includes an indicator 12 for whether or not the two states in a dyad share the same majority religion. Determining what sorts of interstate interactions to examine through the lens of religious commitment is not necessarily a straightforward task–religious traditions can be both theoretically and empirically linked to all kinds of human behaviors–but in light of popular assumptions that often associate religion with violence, as well as the fact that existing scholarship on religion in international relations tends to focus on conflict, I opt to test the viability of my chosen scope condition in association with antagonistic, bellicose state actions. A natural step for future work in this area would be to examine cooperative and altruistic state behaviors, such as the distribution of foreign aid. To operationalize conflictual interstate behaviors, a count of initiations of militarized interstate disputes (MIDs) was employed as the outcome of interest. As will be discussed in further detail below, the analyses are conducted on a cross-sectional, non-temporal basis, and thus the dependent variable for the models is an aggregate count of MID initiations by individual country over a particular time period. For the monadic model, this total count is over the time period 1990-2008, while for the dyadic models, the count is for 1990-2001. Both periods are determined by the availability of data from the Militarized Interstate Dispute (MID) dataset compiled by the Correlates of War (COW) Project. MIDs are more common than interstate wars, offering more variation across cases and more occurrences to examine statistically, and their employment in scholarship on international conflict is relatively standard practice. A militarized interstate dispute is a case of conflict in which ”the threat, display or use of military force short of war by one member state is explicitly directed toward the government, official representatives, official forces, property, or territory of another state” (Jones, 1996, p. 163). For this study, I include cases where the conflict action constitutes the display or use of military force, omitting cases where only threats have been issued in order to distinguish costly conflict actions from “cheap talk.” It also bears repeating that country counts are based on MID initiations, where a state is the first mover in a conflict, not on mere MID involvement, owing to potentially distinct sets of mechanisms that might link religious commitment to the latter. 13 3.2 Methodology The modeling strategy employed for this project is a cross-sectional analysis that compares across countries without any temporal element. Relying upon a cross-sectional approach for this study has the advantage of providing a more conservative test of the viability of religious commitment as a scope condition than would a time-series crosssectional approach because it reduces the problems of serial-correlation inherent to panel data. In addition, as has been previously mentioned, there are not applicable surveys available for every country in every year; running cross-sectional analyses helps to deal with this reality, as well, as no imputation is required for dealing with missingness. If anything, we can actually be more confident in the precision of the level of religious commitment variable for each state included in the models, since the proportions for each country are averaged across multiple surveys, which might each otherwise come with their own idiosyncratic effects. This provides a method by which to account for potential issues with standardizing across different types of questions that go into the commitment measurement, as well. The analysis evaluates states in two different ways, the first of which is monadic– looking at the general propensity for countries to initiate MIDs, regardless of who those MIDs are with–and the second of which is dyadic–taking into account the natures of target states. Because the dependent variable is a count of militarized interstate disputes, a Poisson generalized linear model with a Log link function is used to estimate the effects of the explanatory and control variables identified above in both the monadic and dyadic components of the analysis. I have already described the procedure for computing the level of religious commitment variable in preparation for the cross-sectional; each of the other independent variables is held at its mean or median over the relevant period of time (1990-2008 for the monadic component, 1990-2001 for the dyadic), generating non-temporal country snapshots. These individual country profiles each constitute an observation in the monadic model. Each directed dyad (combining states in pairs twice, with each state having the opportunity to be the initiator [State A] as well as the target [State B]) constitutes an observation for the dyadic models. 14 Control variables included in the models are drawn from the greater literature on conditions for interstate conflict. These are made up of a state’s CINC score (Correlates of War National Material Capabilities Dataset) to account for countries’ varying levels of capability; Polity score (Polity IV Project), to control for factors associated with regime type, including whether democratic and/or autocratic institutions may enhance or restrict the impact of religious factors; an indicator of major power status (EUGene), because major powers may be overall more likely to be involved in conflict; and a variable capturing neighboring country dynamics (Correlates of War Direct Contiguity dataset). For the monadic analysis, this neighbors control takes the form of number of neighboring countries, because states with more neighbors can have more opportunities and/or reasons for interstate disputes. The dyadic analysis includes an indicator for whether the two states in a dyad are contiguous. Results of the analyses are presented in the form of substantive effects plots, which chart a predicted value of the dependent variable–here, that value is always the predicted number of MID initiations by a state during the time period covered by the analysis–over changes in the value of the variable of interest and under any particular conditions set by the researcher, for example, two states sharing the same majority religion. Presenting model results in this way has a number of advantages. first of all, substantive effect plots are much more straightforward in their interpretation than raw coefficients, particularly in the case of non-linear models and when interaction terms are employed–both of which are true of this study. Related to this point, unlike the presentation of raw coefficients, substantive effect plots allow the observer to determine the “significance” of a particular factor by simultaneously viewing the uncertainty associated with an estimate and the substantive magnitude of the relationship between the variable of interest and the dependent variable. Exclusive reliance on p-values would miss the complexity of these relationships. Tables containing the point estimates of the models, however, are included in Appendix B. 15 4 Results and Discussion Question 1: Is a state’s level of religious commitment at all connected to its overall likelihood to initiate confict with other states? [Figure 1 about here] Figure 1 presents the substantive results of the first, monadic analysis. The plot represents what might be termed a “median state” in the international system, according to the data–that is, a state with median capabilities, number of neighbors, and polity score. It is a not a major power and has Christianity as a majority religion, as it is the median faith tradition within the group of states examined. The “median state” is not intended to represent any actual, existing country–though there may be countries within the data that fit its profile–but rather to demonstrate how, when other factors are held constant at their medians, the predicted count of MID initiations for a state changes over various levels of religious commitment. As Figure 1 shows, as the proportion of the median state’s population that is religiously committed (level of religious commitment) increases, there is a corresponding steady increase in the predicted number of militarized interstate disputes that it initiates– a substantive increase from an estimated 0.5 MIDs when close to 0 percent of a state’s population identifies as religiously committed to an estimated 1 MID as that percentage approaches 100. Considering that the median number of initiations for states in the monadic dataset is zero, and that the mean is 1.08, a predicted substantive increase by 0.5 MIDs is not inconsequential. Having controlled for typical explanatory factors for conflict, this is particularly striking, and an initial indication that religious commitment could not only be a scope condition for understanding religion’s role in international affairs, but also a potentially salient characteristic to examine in more mainstream scholarship on interstate conflict. 16 Substantive Impact of Religious Commitment (Monadic) Predicted Count: MID Initiation 1.5 1.0 0.5 0.0 0.2 0.4 0.6 0.8 Proportion of Population Religiously Committed Figure 1 17 Question 2: Is the likelihood that a state will initiate conflicts against a state with a different majority religion affected by the initiating state’s domestic level of religious commitment? [Figures 2 and 3 about here] Figures 2 and 3 display the first results of the transition from examining states exclusively on their individual characteristics (monadically) to interrogating how the natures of their potential adversaries may impact the likelihood for conflict (dyadically). In this part of the analysis, the model accounts for the capabilities and polity scores of each state within a dyad, as well as for whether either is a major power and whether they are contiguous.3 The variable of interest, level of religious commitment of State A (the potential/actual initiating state), is interacted with whether the pair of states has a majority religion in common–an indicator which takes a value of 0 or 1. The reasoning behind this interaction is that levels of religious commitment may have a different relationship to MID initiations when states share a majority religious tradition than when they do not, an expectation that follows narratives expecting differences between religions to lead to, or at least contribute to, conflict behavior.4 This view can be logically extended to an expectation that such differences could be magnified and hostilities exacerbated when states’ populations are more committed to their own religious traditions. However, as Figures 2 and 3 show, there is an almost nonexistent difference between the predicted count of militarized interstate disputes initiated by a median state A as the proportion of its population that is religiously committed increases when it shares a majority religion with a median State B (Figure 2) and when it does not (Figure 3). 3 Raw estimates of this model are presented in Table 2 in Appendix B. Perhaps the most well-known, and controversial, expectation of such a relationship is put forth in Samuel Huntington’s The Clash of Civilizations (1993), in which Huntington predicts that conflicts in the post-Cold War era will fall along “civilization” lines rather than other cleavages or grievances. Civilizations, as Huntington conceives of them, are the broadest shared cultural grouping short of humanity itself, and he ties their natures to shared ideas–fueled predominantly by religion. Some of the implications from the prediction are that such shared ideas exist, that they are significant at the societal level, and that they motivate how states interact with one another, in terms of conflict or otherwise. The thesis indicates that differences between civilizations are ultimately irreconcilable at some level, and Huntington assumes that these irreconcilable differences will inevitably lead to conflict. For empirical responses to Huntington, see Bolks and Stoll (2003), Gartzke and Gleditsch (2006), and Chiozza (2002), among others. 4 18 Substantive Impact of Initiating State's Religious Commitment: Target Has Same Majority Religion Predicted Count: MID Initiations 0.020 0.015 0.010 0.005 0.000 0.0 0.2 0.4 0.6 0.8 Proportion of Population Religiously Committed, Initiating State Figure 2 19 Substantive Impact of Initiating State's Religious Commitment: Target Has Different Majority Religion Predicted Count: MID Initiations 0.020 0.015 0.010 0.005 0.000 0.0 0.2 0.4 0.6 0.8 Proportion of Population Religiously Committed, Initiating State Figure 3 20 In other words, whether State A and State B have the same majority religion or not does not appear to matter to State A’s likelihood to initiate conflict. However, there is another clear connection between level of religious commitment and predicted counts of MID initiations. Though the values on the Y axes of Figures 2 and 3 are significantly smaller than in Figure 1–seeming to suggest much less of a substantive effect–one must note inherent differences between the monadic and dyadic data, a most significant one of which being that counts of initiations are not totals for individual states any longer, but rather totals for directed pairs of states. The median number of initiations for a dyad in the data is again 0, while the mean is down to 0.019. The substantive relationship between level of religious commitment and predicted MID initiations presented in Figures 2 and 3 is actually, therefore, a relevant one. One may also note that the increase from approximately 0.001 predicted MID initiations when State A’s levels of religious commitment are close to 0 percent to a count of close to 0.0075 initiated MIDs when those levels are closer to 100 percent represents a 650 percent increase. So while shared majority religion may not be much of a substantively impactful factor, levels of religious commitment appear to be meaningful in dyadic relationships just as they were in monadic propensity. Question 3: Does the level of religious commitment within State B in a dyad appear to have any effect on the likelihood that State A will initiate conflict with State B? [Figures 4 and 5 about here] While the answer to the previous question focused on the level of religious commitment within initiating states, considering the findings thus far, it makes sense to consider whether target states’ levels of commitment may also factor into the likelihood that a State A will initiate a MID against a State B. Is whether State A’s commitment matters to conflict propensity somehow contingent upon State B’s religious commitment profile? Exploring this possibility, a second dyadic analysis5 emphasizes an interaction of the levels 5 The raw estimates of this model are presented in Table 3 in Appendix B. 21 Substantive Impact of Initiating State's Religious Commitment: Target Has Minimum Religious Commitment Predicted Count: MID Initiations 0.020 0.015 0.010 0.005 0.000 0.0 0.2 0.4 0.6 0.8 Proportion of Population Religiously Committed, Initiating State Figure 4 22 Substantive Impact of Initiating State's Religious Commitment: Target Has High Religious Commitment Predicted Count: MID Initiations 0.020 0.015 0.010 0.005 0.000 0.0 0.2 0.4 0.6 0.8 Proportion of Population Religiously Committed, Initiating State Figure 5 23 of religious commitment within both States A and B in a dyad. The model here therefore includes an interaction term between the level of religious commitment within State A and that within State B, the implication here being that we might expect that whether State A’s level of religious commitment comes into play is contingent upon the nature of State B’s own level of religious commitment. To display a relatively fine-grained picture of what this relationship looks like over changes in State A’s level of religious commitment–as has been the case in the previous figures and thereby enabling comparison between them–five different scenarios were created, differing on State B’s level of religious commitment. In the first scenario, presented in Figure 4, States A and B are “median” in every way except in terms of their religious commitment variable; State A’s level of religious commitment is allowed to vary from the observed minimum to the observed maximum (shown along the X axis) while State B’s level of religious commitment is held at its observed minimum. In other words, Figure 4 shows the change in the predicted count of MID initiations by State A as its level of religious commitment changes when State B’s level of religious commitment is minimal. The other scenarios follow this example, with State B’s level of religious commitment set to the observed 1st quantile value (0.36), the observed median value (0.51), the observed 3rd quantile value (0.79), and the observed maximum value (0.98). Figures 5 presents the 3rd quantile value (“high” religious commitment) for comparison here, while the plots for the remaining scenarios are available in Appendix C. These plots tell a fascinating story about the interactive impact of joint levels of religious commitment within dyads. When State B has lower levels of religious commitment (Figures 4 and 5), we observe essentially no change in the predicted MID initiation count for State A over increases of its own level of religious commitment. This changes ever so slightly when State B’s proportion of the population that identifies as religious increases to around 50 percent, at which point State A’s predicted number of MID initiations gradually increases as its own commitment level increases. The most drastic changes by far, however, appear when high levels of religious commitment in both states are associated with relatively much higher MID initiation counts (see Figure 5), the curve becoming 24 ever steeper in both graphs as State A’s religious commitment level increases. These results drive home the conclusion that higher levels of religious commitment are somehow connected to states’ propensities for initiating conflict. But what is more, the results also emphasize a key dyadic dynamic to this connection, in which the likelihood that we will observe a state initiating conflict with another is as much about the level of religious commitment within the targeted state as it is about the level within the initiating state. Though the monadic analysis revealed a higher general bellicosity associated with higher religious commitment levels, further scope conditions appear to have been placed on that empirical finding. Question 4: Does the level of religious commitment within State B in a dyad appear to condition the impact of State A’s level of religious commitment on its likelihood to initiate conflict with State B when State B has a different majority religion? [Figures 6 and 7 about here] The final component of the dyadic analysis brings together the specifications of its preceding parts. Perhaps the results of Part 3, which emphasized the impact of the interacted levels of religious commitment within a dyad on the likelihood that State A would initiate a MID, could help explain the somewhat counter-intuitive takeaway of Part 2, which found no difference between the substantive impact of levels of religious commitment on predicted counts of MID initiations when states had the same or different majority religions. The results of this final model, which includes another interaction–this time between the levels of religious commitment of States A and B and the indicator for whether the states share the same majority religion–are presented in two more substantive effects plots (Figures 6 and 7).6 The first plot (Figure 6) shows the effects in a scenario when otherwise median states share the same majority religion, and the second plot (Figure 7) shows the effects in a scenario when those majority religions are different. Both scenarios set State B’s level of religious commitment to “high,” following the findings of Part 3. 6 The raw estimates of this model are presented in Table 4 in Appendix B. 25 Substantive Impact of Both States' Religious Commitment: Target Has High Commitment, Same Majority Religion Predicted Count: MID Initiations 0.020 0.015 0.010 0.005 0.000 0.0 0.2 0.4 0.6 0.8 Proportion of Population Religiously Committed, Initiating State Figure 6 26 Substantive Impact of Both States' Religious Commitment: Target Has High Commitment, Different Majority Religion Predicted Count: MID Initiations 0.020 0.015 0.010 0.005 0.000 0.0 0.2 0.4 0.6 0.8 Proportion of Population Religiously Committed, Initiating State Figure 7 27 The results of Part 3 revealed that not only State A’s level of religious commitment matters to its propensity for bellicose behaviors, and indeed we can see from the current results that the conditioning impact of State B’s level of religious commitment (set to “high” in this scenario) continues. While in Part 2 no discernable difference existed between the relationship of religious commitment to MID initiation under conditions of shared or different majority religions, the puzzle of that counter-intuitive finding appears to have been unlocked by introducing the dynamics of State B’s level of religious commitment to the analysis. When we take the dyadic dynamics of religious commitment into account, we can suddenly see what common narratives of religion and conflict might have suspected: on average, states with higher levels of religious commitment are more likely to initiate conflict with states that also have high levels of religious commitment and differ in their majority faith tradition. In other words, it appears that when two states have populations that are each highly committed to different traditions, they are more likely to experience conflict. Introducing religious commitment as a factor to the study of international conflict uncovers an important dynamic that would be missed by continuing to focus only on differences in majority religions within dyads. 5 Exploring Mechanisms: Why and How Does Religious Commitment Matter? As the above results have demonstrated, religious commitment appears to be an empirically viable scope condition for understanding the roles that religion can play in international relations. What is it about religious commitment itself that could be driving the observed substantive effects? This particular section of the paper is intentionally brief, for while identifying the potential mechanisms that could be behind these empirical findings will be important to developing a robust understanding of how religion factors into interstate interactions, such identification will require much more attention and additional inquiry than can be provided here. The goal of this project, first and foremost, has been to introduce an as-yet unexplored factor to the quantitative study of religion and inter28 national relations, and in the process take a step forward in identifying the conditions under which religion may be a factor of importance to international affairs, generally. Even with both of these objectives achieved, however, this piece would be remiss in not offering some preliminary ideas regarding the processes that yield the presented results. Hopefully these potential explanations can provide a jumping-off point for future work on this topic that focuses on the dynamics of causal mechanisms. Dyadic differences in majority religion only appear to matter when both states’ levels of religious commitment are high (not only State A’s). This is certainly an interesting finding and a highly relevant scope condition to identify. It is also important to recall, however, that religious commitment itself was clearly linked to State A’s MID initiation behavior in each step of the analysis, under every tested set of conditions. It could be that religious commitment plays some sort of role in hardening existing inter-religious animosities that could make armed conflict more likely, but the overall results of this study also appear to indicate that there is something about religious commitment in its own right that could be generally linked to state bellicosity. What might be the nature of this link? Perhaps the prevalence of religious commitment itself makes it a factor in not only conflict initiation, but indeed in the antagonistic relationships and bargaining behaviors that come before the step where use of force takes place. So while State A may be the first mover (initiator) in a MID, the actions beforehand by State B could also be connected to its own levels of religious commitment, increasing the probability that escalation to a MID takes place. If this is the case, the expansive literature on international bargaining may be able to provide some explanatory assistance. And because the operationalization of “level of religious commitment” in this study is a proportion of a state’s population that purports to be religious, work on bargaining that affords some sort of role to the general public may be particularly illuminating. The literature on so-called “audience costs” could prove a fruitful place to start. The basic premise of the “audience costs” concept initially comes from the seminal work of James Fearon (1994), which builds upon Schelling’s (1956) outline of tactics 29 such as “tying one’s hands” that lock in one side’s proposals in international bargaining. Fearon argues that state leaders who back down in international crises will be punished by their domestic constituents for being unsuccessful at foreign policy.7 This domestic political cost, suffered by leaders–whose ultimate aim is to remain in power–is a helpful tool for avoiding war, accoring to Fearon, as leaders who can generate “audience costs” can credibly signal to their international opponents their unwillingness to back down once they have issued a threat, revealing previously private information about intentions/resolve and locking into a proposed course of action that eliminates questions of bluffing. States interacting with states that are able to generate audience costs are therefore theoretically able to adjust their offers in bargaining based upon more complete information. If an outcome still exists within the opponent’s bargaining range with this knowledge, then they can propose an offer and attempt to avoid costly conflict. For Fearon, audience costs should be associated with less bellicose interstate behavior because they encourage deterrence and restraint as actors look ahead to the costs and consequences of costly war. However, a number of scholars have identified a double-edged nature for audience costs that could actually link them to conflictual behavior. As Leventoglu and Tarar (2005) point out, making public demands that thereby generate audience costs is a dominant strategy for each negotiator in a bargaining situation, making it unlikely that either side will back down. In an international crisis situation, we might then expect the generation of audience costs to be associated with conflict. Similarly, Debs and Goemans (2010) argue that when leaders’ tenures are less sensitive to the outcomes of international interactions (when they have fewer audience costs when acting internationally), they are more likely to offer concessions for peaceful outcomes, pointing to more conflictual expectations for leaders facing audience costs.8 7 While much of the audience cost literature has been written in response to the empirical phenomenon of the “democratic peace,” the audience costs mechanism has also been expanded to many different kinds of regimes (e.g.De Mesquita et al. (1999); Weeks (2008)). 8 Further, we may note that Fearon’s optimism about audience costs is specified primarily for the phenomenon of war, which constitutes complete bargaining breakdown. Militarized interstate disputes, on the other hand, might themselves be part of a greater bargaining process on the road to war–or to a settled negotiation. Peaceful expectations for audience costs rely upon dynamics of deterrence and restraint associated with the high cost of war, but MIDs may be comparatively less costly and actually used as demonstrations of resolve and to communicate private information. 30 If this pessimistic conclusion is in fact applicable, then it is encouraging to note that according to Slantchev (2006), generating audience costs is actually quite difficult. Leaders condition their behavior on the chance that citizens will find out about policy quality, making audience costs quite dependent on institutional features and media freedoms. But perhaps high levels of popular religious commitment supply a condition under which audience costs can be more effectively generated by leaders, thereby leading to decreased concessions in international bargaining behavior and a higher probability for conflict. Religious commitment could, for example, provide a means by which to circumvent some of the intrinsic difficulties of audience cost-generating process by offering leaders an opportunity to utilize shared language, symbolism, and values systems to bypass the ambiguity of institutional structures and media issues, regardless of regime type. By this logic, the greater the population of a state that is religiously committed, the more that leaders’ rhetoric steeped in religious language or references can be used as an effective tool to cue costs. Religious commitment could also impact the credibility and salience of audience costs. Smith (1998) attributes the micro-foundations of audience costs to audiences being uncertain of leaders’ types–i.e. are they competent or incompetent? Whether citizens actually care about foreign policy or not is not necessarily important; citizens can use leader performance in international crises to form opinions about that leader in other areas–even domestic policy. Smith argues that reneging on an international threat, despite it being “cheap talk,” signals a leader’s incompetence to his domestic audience, allowing them to differentiate his competence type. Utilizing religious rhetoric to paint policies in terms of particular salience, such as good versus evil, and appealing not only to citizens but, indeed, to some kind of higher power to whom they are ultimately accountable, an executive can provide additional means which a committed public can learn more about his or her “true” type– not only as an authority figure, but as a religious compatriot. A leader who backs down under such conditions might then expect to not only be punished for having less resolve and losing, but also for capitulating to an existential and evil adversary. In the process, a 31 leader could also sacrifice the trust relationship with his or her committed public, cutting into any ability to cue public opinion. The incentives to maintain resolve and avoid backing down would then only ratchet up when leaders preside over publics with high levels of religious commitment. And if both sides have a dominant strategy to generate such costs if they are able, then it would follow that we should expect more frequent conflict in dyads where both states with higher levels of religious commitment–and this is exactly what the results of this project convey. Future work will provide opportunities to test the applicability of audience cost logics to this topic, or to apply other causal pathways and theoretical proposals. The primary takeaway of the present work, however, is that there at least exists an empirical connection between increasing levels of religious commitment within states and their propensity to initiate conflict with other states. This is an interesting finding both for scholars interested in interrogating religion’s connections to international processes as well as those who study the foundations of and conditions for international conflict behaviors, more generally. This project has also taken a step toward refining the conceptualization and measurement of religious variables for “thick” scholarship utilizing quantitative methodologies. “Religion” as an amorphous concept may be difficult to successfully link to international processes, but more concrete operationalizations of factors such as religious commitment can offer useful insights into how the beliefs and practices of people around the world are connected to how states interact. 32 Appendix A: Coding Procedures for Generating “level of religious commitment” Variable For questions of how important religion was to life of respondent: • Very important = Very religious • Somewhat important = Somewhat religious • Not too (or ”not very”) important = Not very religious • Not at all important = Not at all religious For questions of how religious the respondent would consider him/herself: • Very religious = Very religious • Moderately (or ”somewhat”) religious = Somewhat religious • Lightly (or ”not very”) religious = Not very religious • Not religious at all = Not at all religious For questions of how often the respondent practiced religious behaviors: • More than once a week = Very religious • A few time a month/Once a month/Once a week = Somewhat religious • At least once a year/A few times a year = Not very religious • Not at all = Not at all religious 33 Appendix B: Point Estimates of Cross-Sectional Models Table 1: Monadic Cross-Sectional (Intercept) Level of Religious Commitment Muslim Majority State No Majority Religion/Indigenous Majority Buddhist Majority State Hindu Majority State Jewish Majority State Major Power Polity CINC Score Number of Neighbors Estimate Std. Error z value -1.7584 0.6597 -2.67 0.7100 0.6501 1.09 0.2304 0.3667 0.63 0.1124 0.4115 0.27 0.5549 0.5868 0.95 1.2693 0.5115 2.48 2.1477 0.4883 4.40 1.2553 0.3327 3.77 0.0813 0.0306 2.66 -0.1924 0.5145 -0.37 0.0813 0.0239 3.40 Pr(>|z|) 0.0077 ** 0.2748 0.5298 0.7847 0.3444 0.0131 0.0000 *** 0.0002 *** 0.0078 ** 0.7085 0.0007 *** Table 2: Dyadic Cross-Sectional 1 (State A Religious Commitment and Same Majority Religion) (Intercept) Level of Religious Commitment, State A Same Majority Religion Major Power, State A Major Power, State B Contiguous Polity, State A Polity, State B CINC, State A CINC, State B Level of Religious Commitment, State A* Same Majority Religion 34 Estimate Std. Error z value Pr(>|z|) -4.0493 0.5863 -6.91 0.0000 *** 1.6866 0.5405 3.12 0.0018 ** -0.4863 0.3950 -1.23 0.2183 1.5932 0.3338 4.77 0.0000 *** -0.5809 0.3497 -1.66 0.0967 . 4.0362 0.1943 20.78 0.0000 *** -0.0111 0.0176 -0.63 0.5275 -0.0364 0.0125 -2.91 0.0036 ** -2.3772 0.9272 -2.56 0.0104 * -2.2744 0.7349 -3.09 0.0020 ** 0.2460 0.7427 0.33 0.7405 Table 3: Dyadic Cross-Sectional 2 (Interaction: Religious Commitment [State A], Religious Commitment [State B]) (Intercept) Level of Religious Commitment, State A Level of Religious Commitment, State B Same Majority Religion Major Power, State A Major Power, State B Contiguous Polity, State A Polity, State B CINC, State A CINC, State B Level of Religious Commitment, State A* Level of Religious Commitment, State B Estimate Std. Error z value Pr(>|z|) -3.1377 0.9280 -3.38 0.0007 *** -0.7895 1.5234 -0.52 0.6043 -0.9482 0.9746 -0.97 0.3306 -0.2902 0.2639 -1.10 0.2714 0.6500 0.4408 1.47 0.1403 -0.4881 0.3540 -1.38 0.1680 4.9205 0.4356 11.29 0.0000 *** -0.0636 0.0239 -2.66 0.0077 ** -0.0114 0.0243 -0.47 0.6384 -3.4323 1.2048 -2.85 0.0044 ** -2.8206 1.0022 -2.81 0.0049 ** 4.3629 1.9363 2.25 0.0242 * Table 4: Dyadic Cross-Sectional 3 (3-Way Interaction: Religious Commitment [State A], Religious Commitment [State B], Same Majority Religion) (Intercept) Level of Religious Commitment, State A Level of Religious Commitment, State B Same Majority Religion Major Power, State A Major Power, State B Contiguous Polity, State A Polity, State B CINC, State A CINC, State B Level of Religious Commitment, State A* Level of Religious Commitment, State B Level of Religious Commitment, State A* Same Majority Religion Level of Religious Commitment, State B* Same Majority Religion Level of Religious Commitment, State A* Level of Religious Commitment, State B* Same Majority Religion 35 Estimate Std. Error z value Pr(>|z|) -3.8327 1.0486 -3.66 0.0003 *** 1.4417 1.8909 0.76 0.4458 -0.7243 1.1135 -0.65 0.5154 -0.4368 1.3869 -0.31 0.7528 0.6776 0.4528 1.50 0.1345 -0.4789 0.3565 -1.34 0.1792 4.9169 0.4536 10.84 0.0000 *** -0.1215 0.0289 -4.20 0.0000 *** -0.0067 0.0244 -0.27 0.7840 -3.4458 1.2166 -2.83 0.0046 ** -3.2019 1.0555 -3.03 0.0024 ** 3.8991 2.2704 1.72 0.0859 . 1.5035 3.2207 0.47 0.6406 4.1658 2.5492 1.63 0.1022 -10.8111 5.8038 -1.86 0.0625 . Appendix C: Additional Scenario Plots for Question 3 Substantive Impact of Initiating State's Religious Commitment: Target Has Low Religious Commitment Predicted Count: MID Initiations 0.020 0.015 0.010 0.005 0.000 0.0 0.2 0.4 0.6 0.8 Proportion of Population Religiously Committed, Initiating State Figure 8 36 Substantive Impact of Initiating State's Religious Commitment: Target Has Average Religious Commitment Predicted Count: MID Initiations 0.020 0.015 0.010 0.005 0.000 0.0 0.2 0.4 0.6 0.8 Proportion of Population Religiously Committed, Initiating State Figure 9 37 Substantive Impact of Initiating State's Religious Commitment: Target Has Maximum Religious Commitment Predicted Count: MID Initiations 0.020 0.015 0.010 0.005 0.000 0.0 0.2 0.4 0.6 0.8 Proportion of Population Religiously Committed, Initiating State Figure 10 38 References Akbaba, Yasemin and Zeynep Taydas. 2011. Religion, Security Dilemma, and Conflict: The Case of Iraq. In Religion, Identity, and Global Governance, ed. Patrick James. University of Toronto Press pp. 159–186. Bolks, Sean and Richard Stoll. 2003. “Examining conflict escalation within the civilizations context.” Conflict Management and Peace Science 20(2):85–109. Chiozza, Giacomo. 2002. “Is there a clash of civilizations? Evidence from patterns of international conflict involvement, 1946-97.” Journal of Peace Research 39(6):711–734. De Mesquita, Bruce Bueno, James D Morrow, Randolph M Siverson and Alastair Smith. 1999. “An institutional explanation of the democratic peace.” American Political Science Review pp. 791–807. Debs, Alexandre and Hein E Goemans. 2010. “Regime type, the fate of leaders, and war.” American Political Science Review 104(03):430–445. Fearon, James D. 1994. “Domestic political audiences and the escalation of international disputes.” American Political Science Review pp. 577–592. Fox, Jonathan. 2001. “Religion as an overlooked element of international relations.” International Studies Review 3(3):53–73. Fox, Jonathan and Shmuel Sandler. 2004. Bringing religion into international relations. Cambridge Univ Press. Gartzke, Erik and Kristian Skrede Gleditsch. 2006. “Identity and conflict: Ties that bind and differences that divide.” European Journal of International Relations 12(1):53–87. Hassner, Ron E. 2011. Religion and International Affairs: The State of the Art. In Religion, Identity, and Global Governance. University of Toronto Press. Helliwell, John F and Robert D Putnam. 1995. “Economic growth and social capital in Italy.” Eastern Economic Journal pp. 295–307. Huntington, Samuel P. 1993. “The clash of civilizations?” Foreign affairs pp. 22–49. Hurd, Elizabeth Shakman. 2004. “The political authority of secularism in international relations.” European Journal of International Relations 10(2):235–262. Inglehart, Ronald and Pippa Norris. 2004. “Sacred and secular.” Religion and politics worldwide 1. Kellstedt, L. A., J. C. Green, J. L. Guth and C. E. Smidt. 1996. Grasping the essentials: The social embodiment of religion and political behavior. In Religion and the culture wars: Dispatches from the front. pp. 174–192. Leventoglu, Bahar and Ahmer Tarar. 2005. “Prenegotiation public commitment in domestic and international bargaining.” American Political Science Review 99(03):419–433. Lumsdaine, David Halloran. 1993. Moral vision in international politics: the foreign aid regime, 1949-1989. Princeton University Press. Philpott, D. 2007a. “Religion, reconciliation, and transitional justice: The state of the field.” Social Science Research Paper Working Paper (17 October). Philpott, D. 2007b. “What religion brings to the politics of transitional justice.” Journal of International Affairs 61(1):93–110. Philpott, Daniel. 2009. “Has the Study of Global Politics Found Religion?” Annual Review of Political Science 12(1):183–202. Philpott, Daniel. 2013. The Religious Roots of International Relations Theory. In Religion and International Relations: A Primer for Research. The Report of the Working Group on International Relations and Religion of the Mellon Initiative on Religion Across the Disciplines. Putnam, Robert D. 1993. “The prosperous community.” The american prospect 4(13):35– 42. Rousseau, David L and Rocio Garcia-Retamero. 2007. “Identity, Power, and Threat Perception A Cross-National Experimental Study.” Journal of Conflict Resolution 51(5):744–771. Schelling, Thomas C. 1956. “An essay on bargaining.” The American Economic Review pp. 281–306. Shaffer, Brenda. 2006. The limits of culture: Islam and foreign policy. MIT Press. Slantchev, Branislav L. 2006. “Politicians, the media, and domestic audience costs.” International Studies Quarterly 50(2):445–477. Smidt, C., L. Kellstedt and J. Guth, eds. 2009. The Oxford handbook of religion and American politics. Oxford Handbooks Online. Smidt, Corwin E. 2005. “Religion and American attitudes toward Islam and an invasion of Iraq.” Sociology of Religion 66(3):243–261. Smith, Alastair. 1998. “International crises and domestic politics.” American Political Science Review 92:623–638. Templeton, John. 1999. Agape Love: Tradition In Eight World Religions. Templeton Foundation Press. Thompson, Henry O. 1990. “Praxis: Peace-Preachment and Practice.” International Journal on World Peace pp. 87–100. Toft, Monica Duffy. 2007. “Getting religion? The puzzling case of Islam and civil war.” International Security 31(4):97–131. Toft, Monica Duffy, Daniel Philpott and Timothy Samuel Shah. 2011. God’s century: resurgent religion and global politics. WW Norton & Company. Weeks, Jessica L. 2008. “Autocratic audience costs: Regime type and signaling resolve.” International Organization pp. 35–64. Wendt, Alexander. 1999. Social theory of international politics. Cambridge University Press.