Abstract
Who commits religious discrimination, and why? Recent findings at the country level show an inverse relationship between societal religiosity and social religious discrimination, but findings at the individual level show an increase in conservatism among religious practitioners. In this study, I seek to bridge the gap these levels with a novel framework of religious identity. I argue that the threat of secularism can cause religious communities to reinforce boundaries between themselves and other groups. This should manifest itself in hostile attitudes toward members of other religions. I test this theory using multilevel data. I find that as aggregate religiosity declines, individual religiosity increasingly predicts inter-religious hostility. At the same time, this effect is only the case concerning community- and practice-based aspects of religiosity, and not among beliefs in-and-of-themselves. This study bridges a gap between macro-level findings regarding religion and state and micro-level findings regarding the modernization hypothesis.
Why is religious discrimination increasing, and who is responsible for it? Recent studies by Fox (2013, 2020) and Fox et al. (2018) have examined these questions as one of the central components of resurgent religion in the twenty-first century. Such questions have far-reaching ramifications for the broader study of Comparative Politics, as religious cleavages have become an increasingly important part of the polarized landscape across the democratic world. Despite the increase in religious discrimination throughout both advanced and developing economies, however, religious practice as a whole is declining as economies develop (Norris and Inglehart, 2004). Adding to the complexity of this question are findings suggesting that religious self-identification has not declined as much as practice (Dhima and Golder, 2021; Lüchau, 2014), but also that religious self-identification does not predict religiously based illiberalism to the same degree as practice (Cingranelli and Kalmick, 2019). In other words, there appears to be a disconnect between the increasing religious particularism at the country level and the decreasing religious practice at the individual level.
Recent studies have increasingly looked to secular actors as potential sources of religious discrimination. Evidence by Fox et al. (2021) and Fox and Topor (2021) finds an inverse relationship between societal religiosity and discrimination; at the same time, because this relationship is highly conditional and varies from one religious group to another, the authors conclude that their results likely reflect the role of secular actors, rather than religious ones, in perpetuating discrimination. Other work on right-wing populism has demonstrated a clear link between secular actors and religious prejudices (Brubaker, 2017; Cremer, 2024 [2023]). Nevertheless, some research indicates that religious practitioners could become increasingly conservative or exclusive in secularizing environments (cf. Gaskins et al., 2013); other findings indicate that religious practice remains a source of chauvinism even while religious adherence does not (Cingranelli and Kalmick 2019; Dhima and Golder, 2021). At its heart, the disconnect between these studies centers around competing units of analysis and alternate definitions (Szendro, 2025). To date, the question of religious discrimination remains fixed at either the country (Akbaba and Fox, 2019) or minority (Fox, 2019; Fox et al., 2019) level, rather than the individual. In this study, I focus on the individual as the unit of analysis; I argue that religious actors may have powerful incentives to partake in discriminatory behavior in secularizing environments.
Drawing on findings from social psychology, I argue that religious practice indicates either positive identity—what a group is—or negative identity—what a group is not. When groups feel threatened, its members seek to reinforce boundaries between in-groups and out-groups. This would imply that although both religious and secular actors have an incentive to discriminate against religious minorities in a secular environment, this incentive is nevertheless greater for religious actors than their secular counterparts. Throughout this conceptualization, I move away from actions—occurring at the country level—and toward attitudes—at the individual level. I examine this theoretical framework within the context of inter-religious hostility, which I discuss as the attitudinal basis for discriminatory actions.
I test this theory using a multilevel, microfoundational approach. I find that as aggregate levels of religiosity decrease, individuals identifying with the majority religion show greater hostility toward other religions. This effect is substantially larger the more the individual in question practices their religion. In other words, while both secular and religious individuals show a greater tendency toward inter-religious hostility in a secular environment, this effect is larger in religious individuals. This implies that religious actors, rather than secular actors, are more likely to be sources of religious discrimination under such conditions.
This study fills an important gap between the literature surrounding religion and state and the literature surrounding modernization theory. In doing so, it also provides key insights for scholars of development and democratization.
Literature Review
In the study of religion-state relations, religious discrimination has become a focal point for analysis; as a particularly tangible cleavage between religious groups, the rise in discrimination directly challenges previous, archaic models such as secularization theory (cf. Fox, 2020). 1 Existing studies have explored discrimination by both government and societal actors (Fox, 2020; Fox et al., 2018, 2019). Other recent studies have focused on examining individual-level components in relation to state-level ones (Arikan and Ben-Nun Bloom, 2019; Fox et al., 2017). The exploration of religious discrimination serves as the backbone of numerous studies suggesting that the modern relationship between religion and state is marked by a pronounced religious–secular cleavage (Fox, 2015, 2016). These studies largely use structural data to expand upon previous theories of religious markets to demonstrate the effect that state policies can have on societal practices. Religious markets theory views religious groups as firms competing in a marketplace; religious policies can expand or contract the market in the same vein that fiscal policy can affect fiscal markets (cf. Iannaccone, 1992; Iyer, 2016). Exploring a secular–religious cleavage—particularly in secularizing environments—requires incorporating concepts of market forces (cf. Grim and Finke, 2007, 2010).
One repeated finding that begs further examination is that there exists an inverse relationship between societal religiosity and social religious discrimination (SRD). 2 Although both Fox et al. (2021) and Fox and Topor (2021) observe this relationship in aggregate, both studies find this relationship to be highly conditional. Among Christian-majority countries, SRD against non-Christian minorities is inversely related with societal religiosity, but SRD against Christian minorities is directly related with societal religiosity. 3 The authors suggest that while religious actors in secularizing environments see other religious practitioners as allies against the onslaught of secularism, religious minorities within the same family present potential competition for adherents. The conditionality of their findings thus suggests that in secularizing environments, it is secular actors, rather than religious ones, who seize on religious identity as a divider between in-groups and out-groups. Fox and Topor add that in a secularizing world, a “religious realignment” in which religious groups find common ground is more likely than a scenario in which they compete (89).
These conclusions are drawn not only from aggregate data but also from experimental findings by Helbling (2014). Several recent studies of right-wing populism in Europe present a similar concept—that religious discrimination largely stems from secular actors that use religious identity as “floating signifiers” (Lamour, 2022). Work by Brubaker (2017) and Cremer (2024 [2023]), for example, demonstrates the use of religious aesthetics among secular right-wing populists. According to Cremer, a secularized religious identity is an important part of a complex cleavage structure that incorporates elements of civilization identity relative to globalized identities. Cremer (2024 [2023]: 33–35) goes as far as to distinguish between the category of “belief without belonging” and “belonging without belief” to describe the use of religious cleavages as rhetorical devices by secular actors. By the same token, Cremer (2024 [2023]: 246) also analyzes “supply-side” sources of the post-religious right, highlighting the degree to which religious practitioners have reluctantly associated with secular populists due to a “perceived lack of political alternatives.”
A growing body of literature, then, focuses on the role of secular actors in exacerbating religious cleavages, particularly in relation to globalization and secular–religious competition. These studies likewise highlight the importance of renewed religious cleavages to an increasingly unstable and hotly contested democratic landscape. Nonetheless, these studies have focused less on religious practitioners in and of themselves than on broader trends that may or may not incorporate practitioners. It is important, however, to examine this theory further. Although government policy can be examined in terms of institutional structures, societal actions are made up of individuals. It is thus problematic to draw conclusions on individual incentives based on group-level findings (cf. Seligson, 2002). Because SRD is viewed in aggregate, it does not by itself indicate who is committing the actions in question. Examining conditionalities in the relationship between religiosity and SRD provides important evidence in this regard, but without an accompanying analysis of individual-level incentives, it is difficult to paint a full picture.
In this study, I offer an individual-level argument regarding the inverse relationship between religiosity and discrimination. In contrast to existing findings, I suggest that there is an incentive for religious actors to engage in discriminatory behavior in secularizing environments. This is not to suggest the secular actors do not engage in discriminatory behavior, but rather to suggest that religious actors can play a role in this phenomenon as well. In exploring this concept, I refocus the analysis from SRD—a phenomenon occurring in aggregate—to inter-religious hostility, an attitude occurring among individuals. Drawing on functionalist views of religious practice as well as social identity theory, I argue that secularism can create a so-called “bottleneck pressure” on a religious market. “Bottleneck pressures” refer to evolutionary changes that occur when populations go through contractions; in a functionalist context, evolutionary adaptation can explain shifts in religious practices or attitudes (Wilson, 2002).
Ultimately, I suggest that these pressures can incentivize religious practitioners to act defensively by drawing stronger boundaries between in-groups and out-groups. In such an environment, I argue that members of minority religions can be viewed as agents of secularism, because they undermine religious homogeneity and thus introduce elements of relativism and liberalism to previously religious societies. In the section that follows, I will detail the theoretical schools of thought that lend themselves to this argument. I will then detail the construction of a research design based on these theories.
Theory
I argue that religious actors have an incentive to discriminate against members of other religions in secularizing environments. This inter-religious hostility stems from the relationship between religious markets on the one hand and social identity on the other. A long body of literature on the functionalist school of religion in social sciences views religious practice in terms of its social role, as a vehicle for collective action and its ability to delineate in-groups and out-groups. This view places a heavier emphasis on the practice of religion, as something that visibly occurs in the public sphere, than identification with religion, a separate phenomenon that can, in secular environments, occur primarily in private. Based on this line of argumentation, I present a research design in which religious practitioners can be expected to demonstrate stronger levels of inter-religious hostility than their secular counterparts. Below, I summarize the functionalist school of thought, the role of social identity theory, and the concept of “bottleneck pressures” both in theory and in practice.
The functionalist school of thought views religious practice in terms of its social purpose, rather than in terms of individual experience (cf. Berger, 1974). This conceptualization postulates that collective action is made possible through a common understanding of norms; religious institutions use symbolism and signaling to communicate these norms and thus advocate for collective action (Durkheim, 1912; Graham and Haidt, 2010; Haidt, 2013 [2012]). Religions also foster interpersonal relationships through repeat interactions within communal networks (Fukuyama, 2001; Putnam, 1995). As such, religious institutions both rely on and create social capital (Szendro, 2021). Taken together, religious institutions aggregate individuals into a cohesive unit by using symbols to delineate in-groups and out-groups (Wilson, 2002). Modern views of functionalism, such as Wilson’s, draw explicit comparisons with concepts of evolutionary biology as well as religious markets, suggesting that religious competition within markets resembles a form of “multilevel natural selection.”
The key takeaways from the functionalist school are: 1) religious institutions commonly translate individual attitudes into collective action and 2) religious institutions feature mechanisms that neatly divide between in-groups and out-groups. Both aspects are critical for examining the potential individual roots of religious discrimination. Fox (2020) pointedly separates attitudes from actions; a functionalist framework suggests that a key element of religious practice, as distinct from identity, is the ability to translate one to the other. In-group-out-group divisions, meanwhile, establish the motivation for such attitudes and actions. Scholars suggest that religious adherence—encompassing both practice and identity—has exclusive properties that other forms of belonging often do not (cf. Reynal-Querol, 2002). Religious practitioners, then—being distinct from those who merely identify with religion—should adapt attitudes and actions in relation to threat more efficiently than their secular counterparts.
The role of threat in shaping a response, and the types of threats that engender response, are further explored in the literature surrounding social identity theory. According to the long-standing theories of Tajfel (1978) and Tajfel and Turner (1979), individuals within groups strive for a positive group identity; specifically, that members of a group desire a higher standing for their group than others. However, when experiencing a negative identity, members of a group will seek to compete with other groups in order to raise their status. In this context, “positive” and “negative” relate to a group’s conception of itself in relation to other groups. More recent findings suggest that individual members of a group experiencing negative identity may experience identity-uncertainty. Uncertainty can result not only from a group’s negative standing but also from an individual’s sense of displacement within a group (Hogg and Adelman, 2013; Hogg and Reid, 2006). Individuals experiencing uncertainty will attempt to “maximize the ratio of intergroup differences to intragroup differences” (Hogg and Reid, 2006: 10). In doing so, members of the group reinforce an “entitative” boundary between the group and members of other groups. These theories have specifically been applied to the study of religious practitioners, as a functionalist framework suggests that religious rituals serve as signals to delineate these entitative boundaries.
In effect, the positive group identity is defined by in-group favoritism—what a group is—while the negative group identity is defined by out-group hostility—what a group is not. When members of a group experience uncertainty, negative group identity becomes increasingly salient independent of positive group identity. In a religious context, this means that “inter-religious hostility”—an individual’s negative attitude toward other religions—is not directly dependent on an individual’s religiosity; rather it is in reaction to a specific environment of threat. Existing research focuses on the role of secular threat in shaping religious collaboration. Given the exclusive properties of religious adherence, however, I suggest that in some secularizing environments, secular threat may promote inter-religious hostility. This is because religiosity can predispose an individual to certain types of threats. Returning to the concept of religious markets, existing evidence such as Fox et al. (2021) points to a declining religious share of the of the secular–religious market cleavage as encouraging cooperation. In an environment of low religiosity, however, I suggest that a so-called “bottleneck pressure,” a phenomenon in which contracting populations lead to group-level adaptations, should in fact lead to greater competition within an increasingly constrained market.
The potential for competition in a constrained religious market stems from the role of religious homogeneity in preventing the advent of secularism. According to Taylor (2007), one of the challenges religions face from secularism is that of axiomatic secularization; this is the notion that religions, once taken as axiomatically true, can no longer be accepted as such due to the existence of alternatives. This logic implies that religious majorities see religious minorities as agents of secularism, because both secularism and religious pluralism challenge the axiomatic nature of the religious majority (Szendro, 2025). Such formulations can indeed be found in noteworthy cases of religious actors themselves, including the writings of Ayatollah Khomeini (1970: 79) regarding Jews, Christians, and Ba’hai, or Pat Robertson (1991: 218) regarding Hindus and Muslims. A noteworthy case outside the Abrahamic canopy is Myanmar, where one line of argument suggests that anti-Muslim sentiment among Buddhist monks follows from secular challenges to the monastic order (Zin, 2015). 4
In support of this argument, other areas of literature seem to suggest a more assertive religious profile in reaction to modernization. The Gaskins et al. (2013) model, or GGS, for example, demonstrates that religious practitioners grow more conservative in secularizing environments. One reason that these findings have not been explored in relation to each other is that they utilize different units of analysis. Most of the analysis of religious discrimination has occurred either at the state level or at the minority level. Recent work on religion–state relations has increasingly incorporated individual-level data, but primarily in relation to state religious policy (cf. Fox and Breslawski, 2023; Fox et al., 2024). Research on religious cleavages, then, has yet to explore the incentives behind discrimination or its relationship to secularism. Findings at the individual level have long suggested that economic development goes hand-in-hand with secularism due to an overall decrease in existential insecurity (Norris and Inglehart, 2004), but recent findings suggest that this process exerts more of an effect at the country level rather than individual level (Höllinger and Muckenhuber, 2019).
In addition, while economic development predicts a decline in individual religious practice, individuals’ religious self-identification has not declined in relation to the same mechanisms (Dhima and Golder, 2021; Lüchau, 2014). Findings comparing societal identity and government policy further suggest that religious self-identification is not associated with chauvinistic views or policy support, but religious practice is (Cingranelli and Kalmick, 2019). Each of these findings suggest that pinpointing shifts in attitudes or behaviors among religious actors requires a fine-grained analysis and that religious actors represent a diverse group, reacting to secularizing environments in diverse ways. One of these ways may be discriminatory behavior, as an extension of the chauvinistic views that grow in relation to the threat of secularism. By asserting the boundaries between religious communities, religious actors can protect their status when it is threatened by a decline in practice or membership in religious organizations.
This theory, then, views one potential outcome of secularism as a so-called “bottleneck pressure” on religious markets. In evolutionary biology, a population bottleneck is an event in which a sudden decline in population size leads to evolutionary adaptations. Scholars such as Wilson (2002) have used concepts from evolutionary biology to explain the development of religious practices and institutions; this view utilizes frameworks on religious markets as well as functionalism to explore the religion as a process of group-based adaptations. One recurring finding in the study of religious markets is that strict sects grow more easily than lax ones, because these entities can eliminate free-rider costs. One mechanism by which this occurs is to promote exclusivity (Iannaccone, 1994; Tamney and Johnson, 1998; Wilson, 2002). In applying a bottleneck pressure approach to religious markets, I argue that if there is a decline in the overall level of religious practice in a given society, the remaining religious practitioners may adapt by growing more protective of their institutions. This type of protectionism can manifest itself in reinforced boundaries between in-groups and out-groups. Against this backdrop, religious practitioners can come to view religious minorities as undermining the religious homogeneity and, by extension, conviction of a society. In this conception, the pressure of religious bottlenecks could cause inter-religious competition. This suggests that religious actors have an incentive for discrimination—particularly if they perceive a threat from secularization.
In short, key elements of these literatures have effectively bypassed each other in communication, leading to a number of contradictory results. This is in part because differing units of analysis have been deployed in order to divine incentives based on structural observations. In the section that follows, I will develop a theoretical framework to bridge the aggregate and disaggregate components of this analysis. This framework will discuss the notion of macro-level events as products of micro-level attitudes and identities. Rather than religious discrimination at the macro level, I will focus on “inter-religious hostility” as an individual attitude signifying exclusive social identities constructed on a religious basis.
To summarize, I argue that the religious–secular cleavage can produce a “bottleneck pressure” on religious practitioners. Given that religious identities can be divided into “positive” and “negative” components, corresponding to in-group favoritism and out-group hostility, “negative” components should become increasingly salient under an atmosphere of threat. The need to reinforce homogeneity and discipline among coreligionists may thus lead to inter-religious hostility rather than collaboration. If this is indeed the case, then these findings would have far-reaching implications for the study of religious discrimination. Indeed, such findings would provide a deeper understanding of religious cleavages in secular environments, highlighting the role of religion in the contentious political arena of modern democracies.
Regarding the apparent inverse relationship between religiosity and SRD, this suggests that religious actors are in fact a potential source of discrimination in a secular environment. Because discrimination as a societal practice cannot be observed at the individual level, I will examine inter-religious hostility as an individual attitude, held by the majority religion, toward minority religions. I expect that under conditions of increasing secularism, individual religiosity will increasingly predict inter-religious hostility. In brief:
H1: As society becomes less religious, individual religiosity among the majority religion will increasingly predict negative attitudes toward other religions.
This framework allows for “positive” and “negative” aspects of religiosity to be viewed as independent from one another; I argue that secular actors identifying with the majority religion will still perceive a threat of waning religious practice, due to their personal identification, but that this perception of threat will be smaller than that of religious actors. 5 In other words, pursuant to the alternative theories offered above, I expect a similar, albeit lesser, trend among diminishing levels of practice of the majority religion. This would imply that non-practicing individuals of the dominant faith may still react to declining practice as a potential threat. In this regard, I expect both secular and religious actors to demonstrate inter-religious hostility, so long as they identify with the majority religion. Thus, this line of argumentation does not directly contradict alternative theories but instead subsumes them within a broader context. Essentially:
H2: As society becomes less religious, non-practicing members of the majority religion will show greater negative attitudes toward other religions, albeit to a lesser degree.
Research Design
I use a series of hierarchical models to gauge the relationship between individual religiosity, aggregate religiosity, and inter-religious hostility. Table 1 lists descriptive statistics and summary information regarding each of the variables, as described below. The unit of analysis is individuals within a given country year; in the main text, I have limited this sample to members of the religious majority within Christian-majority countries. Respondents were included in the sample if they were members of the majority religious denomination, or members of the majority religious family in a country where that religious family represents a supermajority of the population. This decision reflects several countries in which Christianity represents a decisive supermajority of the population, but no specific denomination is documented as having a majority. In Appendix 2, I compare results across a sample of Muslim-majority countries and a pooled sample of all countries. I also compare results using only respondents from a majority denomination, specifically.
Descriptive Statistics.
Both individual-level religiosity and aggregate religiosity are either drawn from or calculated using data from the World Values Survey (Haerpfer et al., 2022). I have used data from Waves 6 and 7, corresponding to limitations on availability for certain variables of interest. Religious majorities were determined using data from the Religious Characteristics of States (RCS) project (Brown and James, 2017). Wherever country-level data was unavailable for a given year of the WVS, I used the most recently available data.
Dependent Variables
The dependent variables are displays of an individual’s inter-religious hostility. I use two dependent variables encompassing inter-religious hostility. The first is a question asking respondents “to what extent do you agree: the only acceptable religion is my religion?” It is a four-point Likert-type scale ranging from “Strongly Disagree” to “Strongly Agree.” In these models, higher values correspond to agreement. The second is a binary variable as part of a set of questions asking respondents who they do not want as a neighbor. It is coded as “1” if respondents stated that they “do not want a member of another religion” as a neighbor. Just 12.05% of respondents in the sample answered as such. Although the second dependent variable is binary, I use a hierarchical linear model due to difficulties in convergence with hierarchical logit models (cf. Fox et al., 2024). Although hierarchical estimates for binary outcomes do not vary much from linear ones, I have performed robustness tests with hierarchical logit models using QR decomposition to aid convergence. These tests, which can be found in Appendix 3, are in fact less conservative than the tests included in the main text.
Independent Variables
I use four independent variables gauging both religious practice and religious identity at both the individual level and country level. These include service attendance, membership in religious organizations, respondent prayer and self-identification as a religious person. I include an interaction term between individual respondents and the country-level percent of respondents answering a certain cut-off. Two of these terms, service attendance and active membership, are explicitly public forms of practice rather than private forms of identification: one, prayer, is a form of practice that is both public and private in nature, and the other, self-identification, does not encompass public aspects of religiosity.
Service attendance is a 7-point scale asking respondents to report the frequency with which they attend religious services. Here, higher numbers indicate greater service attendance, and lower numbers indicate lower levels of attendance. The highest category is “more than once a week,” while the lowest category is “never or practically never.” As is custom in the literature using WVS data, I have deployed this measure as a continuous rather than categorical measure; tests utilizing a categorical measure can be found in Appendix 3. Following Cingranelli and Kalmick (2019), Fox et al. (2021) and Fox and Topor (2021), the country-level variable indicates the percent of respondents who answer “5” or higher on the scale in a given country-year. “5” corresponds to “at least once month.” Ultimately, aggregate service attendance rates in the included sample range from 8.71% to 93.01%, with an average of 47.21 and a median of 44.78.
Membership in religious organizations encompasses three categories: not a member, an inactive member, and an active member. This variable has been deployed categorically, due to its relatively few categories that are not ordinally comparable to each other. In the main text, the aggregate measure is the percent of respondents in a given country-year who report active membership in a religious organization. The aggregate percent of active membership in the included sample ranges from a low of 1.55% to a high of 74.01%, with an average of 25.08 and a median of 19.58.
Prayer is an 8-point scale asking respondents to report the frequency with which they pray. The lowest of these is “never or practically never,” and the highest of these is “several times a day.” Here, “5” corresponds to “only at religious services.” Like service attendance, this variable is typically used as a continuous variable, but I have included a categorical test in Appendix 3. For the aggregate variable, I use the percentage of respondents reporting “6” or higher in a given country-year. Aggregate prayer rates range from 18.41% to 95.43%, with an average of 64.1 and a median of 68.18.
Finally, the “religious person” question asks respondents whether they are a religious person, not a religious person, or an atheist. As with membership, this question is used categorically rather than continuously. For aggregate percentage, I have used the percent of respondents identifying as religious people within a given country-year. For the included sample, these values range from a low of 30.93% to a high of 97.29%, with an average of 70.67 and a median of 74.85.
Control Variables
At the individual level, I control for respondent income, education level, sex and age. At the country level, I control for a country’s official support for religion, GDP per capita, population size and level of democracy. Official support for religion is taken from the Religion and State project; although its values can potentially range from 0 to 52, this sample only contains values ranging from 2 to 15. GDP per capita and population size are taken from the World Bank’s World Development Indicators (World Bank 2024); in both cases, I use log-values to account for skew. Level of democracy is taken from the Polity V dataset, using the most recent values available for a given country-year (Marshall and Gurr, 2018). I also deploy fixed effects for the year the survey was conducted, and the geographic region in which a country lies. Although models in this area of study typically control for religious diversity or majority percent, I exclude them here as they are potentially colinear with aggregate indicators of religiosity (Voas et al., 2002).
Results
Table 2 contains the output of Model 1 across four sets of interactions, as illustrated in Figure 1. Controls are repressed for space but can be found in Appendix 1. The first set of results concerns the statement “the only acceptable religion is my religion.” Across service attendance, membership in religious organizations, and frequency of prayer, results follow the hypotheses. On their own, each of these expressions of individual religiosity are positively associated with inter-religious hostility, significant at the 0.01 level (p < 0.01); across each of these variables, the interaction term between individual- and aggregate-level religiosity is negatively associated with inter-religious hostility, significant at the 0.01 level (p < 0.01).
Results for “the Only Acceptable Religion is My Religion.”
Standard errors in parentheses.
p < 0.01, ** p < 0.05, *p < 0.1.
Results of Model 1.
What these interactions mean in practical terms varies according to the prediction. At low levels of aggregate service attendance, individual service attendance is associated with a modest but highly significant (p < 0.01) increase in inter-religious hostility. Over the range of aggregate service attendance, however, the effect of individual service attendance falls some 75%. Although the effect is still positive, the size of the effect is just one-fourth of the effect at low levels. Concerning membership in religious organizations, both active and inactive membership are significantly associated with inter-religious hostility at low levels of aggregate membership (p < 0.01), but neither is significant at high levels. Concerning prayer, meanwhile, the effect size of individual prayer is reduced by roughly half over the range of aggregate prayer levels. The weaker effect of aggregate religiosity in this regard likely reflects the nature of prayer as both a private and public act.
Each of these tests show religious practitioners growing more hostile to other religions in secular environments. Of tests concerning “the only acceptable religion is my religion,” only one marker of religiosity—a respondent’s self-identification as a religious person—does not follow the same pattern. When fewer people in a given country-year identify as religious, an individual’s self-identification is not associated with inter-religious hostility; at high levels, it is significant at the 0.01 level (p < 0.01). The same also holds true for respondents who do not identify as religious, relative to those who identify as atheists, although at low levels of aggregate religiosity this association is negative. The fact that this finding does not follow the same pattern is likely because it is the only one of the four variables that does not contain a public- or practice-based component; rather, it is purely based on identification. There is no relative rise in effect sizes for non-religious respondents in secular environments.
Table 3 contains the output of Model 2 across four sets of interactions, as illustrated in Figure 2. Controls are repressed for space but can be found in Appendix 1. Each of these effect sizes are weaker concerning the second dependent variable, a respondent starting that they do not want a member of another religion as a neighbor. Although the interaction terms themselves are not significant, two of the results demonstrate a strong interactive relationship within the variables’ marginal effects. At low levels of aggregate service attendance, individual service attendance is significantly associated with the outcome (p < 0.01); at high levels of aggregate service attendance, it is not. Likewise, at low levels of aggregate membership in religious organizations, active individual membership is also positively associated with the outcome (p < 0.01); at high levels of aggregate membership, it is not. There are fairly wide confidence intervals at high levels of aggregate membership due to a lack of observations, but the slope of individual membership converges on zero at about 60% or the 94th percentile of observations. The effect of individual prayer does not change over the range of aggregate prayer, and the effect of individual self-identification is insignificant at all levels.
Results for “Do Not Want Neighbor of Another Religion.”
Standard errors in parentheses.
p < 0.01, ** p < 0.05, * p < 0.1.
Results of Model 2.
Taken together, these results demonstrate that inter-religious hostility is higher among religious practitioners at low levels of aggregate religiosity; at high levels of aggregate religiosity, the effect of individual religious practice is either insignificant or significantly lower than at high levels. Effect sizes are weaker, non-existent, or, in some cases, the opposite when using variables that are less focused on public practice. At the same time, results do not show non-religious respondents growing more hostile to religious out-groups at low levels of aggregate religiosity. It is worth noting that because these tests only included members of a country’s religious majority, secular respondents in the sample refer to those who still identify as being part of the majority religion. Likewise, because non-practicing or non-member respondents are the reference categories for variables concerning religious practice, the effect size of individual religiosity is in comparison to secularism.
These tests strongly suggest that religious practitioners can become more hostile to religious out-groups in secular or secularizing environments. As such, they offer another potential source of religious discrimination at the country level. Most studies thus far suggest that secular actors are a source of discrimination in secularizing environments. While the results of this study do not directly contradict these claims, they do suggest that religious practitioners can play a role as well.
Nonetheless, there are three caveats to these findings. First, the results themselves do not distinguish between target groups. This is to say that while the main text tests only include members of the religious majority, they do not indicate which groups respondents associate with “members of another religion” or which groups are included in the phrase “my religion.” It is, accordingly, not clear whether respondents feel differently about members of minority groups within their own religious family as opposed to those outside the religious family, or, likewise, if they differentiate between different religious categories, such as those religions that are associated with immigrant groups or subject to securitization measures. Future surveys should explore religious practitioners’ feelings toward specific religious groups, as well is in relation to a respondent’s perception of secularism and its presence in a given country or community.
The second caveat is that these results show a relationship between an individual’s religious practice and the aggregate levels of practice—especially public practice—in a given country-year. These results do not demonstrate a relationship between an individual’s religious identity and inter-religious hostility in secular environments. Future studies should further explore the distinction between practice and identity, and how political dynamics concerning religious adherents may be different across these distinct expressions of religiosity.
The third caveat is that this study utilizes a particular sample and should not be over-generalized. Although preliminary findings in Appendix 2 suggest similar dynamics in Muslim-majority countries, it is difficult to draw conclusions on alternative samples that may either lack sufficient observations or contain too much internal variation. Studies of religious discrimination have indeed shown wide variation across both regions and religious groups (Fox, 2020). While this study suggests that more attention should be paid to the role of religious actors in secularizing environments, it does not intend to suggest that this role is homogeneous across different types of environments. Likewise, it does not intend to suggest that the role of secular actors should not be explored or that the role of religious actors will be the same in relation to each religious group.
Discussion
In this study, I sought to provide a new framework for exploring the motivations of religious and secular actors alike as it concerns the emergence of religious discrimination. Religious discrimination remains among the most prominent counterpoints to secularization theory, as it is a tangible political cleavage that resists expectations regarding religion and modernization. Such avenues of study suggest far-reaching ramifications for the broader study of comparative politics, as religious cleavages have become an increasingly important part of the political landscape. It is perhaps for this reason that discrimination has received more attention than other facets of the religion–state relationship. At the same time, the growing trend toward secularism at the individual level has continued to demonstrate a relationship between modernity and secularism. Prior studies using aggregate data have suggested that this may be because secular actors, rather than religious actors, are more likely to discriminate.
The framework developed in this study suggests an alternative explanation that secularism on the one hand and religious cleavages on the other can not only coexist but perhaps even exacerbate each other. If negative religious identity—what a religion is not—rises in tandem with the loss of aggregate positive religiosity—what a religion is—then both secular and religious actors can perceive an explicit threat from both secularization and religious heterogeneity. The findings above indicate that this effect is stronger among higher levels of religiosity than lower levels; nevertheless, it does indeed support this conclusion.
These results provide a microfoundational approach to the macro-level phenomenon of religion-state relations. In doing so, they bridge the gap between two areas of literature that have often indicated contradictory results due to conflicting units of analysis. As such, they do not contradict the empirical work by Fox et al. (2021) but suggest that a different theoretical interpretation may be more apt when examining microfoundational dynamics.
Such results, however, are not an endpoint for analysis. First, I have not discussed intra- or trans-religious hostility encompassing separate denominations within religious families. This was a deliberate choice in an attempt to establish a straightforward framework that can be explored further. Second, I have not discussed cases in which there is no religious majority; the question of how religions may compete with each other or with secularism may yield different results in an environment without a specific, prevailing dogma. Third, the extant data does not provide material insights into the specific motives of religious actors; there are limits as to how many conclusions can be drawn from this analysis. Drawing on the writings of some prominent such actors, I have suggested that religious actors may equate religious minorities with secularism due to the “axiomatic” principle of secularization as described by Taylor (2007). In order to answer questions regarding personal motivations, however, further data collection must be conducted—particularly with regard to interview research and survey analysis. Finally, I have not tackled the question of when and how attitudes of inter-religious hostility aggregate to discrimination. While this is an important question—and indeed, a logical next step—it is tangential to establishing the framework required here.
At the same time, these findings carry widespread implications not only for literature surrounding religion and modernity but also for scholars of development and democratization. Religious institutions play a key role in democratic consolidation, state-building, civil society, and political mobilization. Understanding the dynamics of religion and modernity can also inform policymakers’ approaches to development. Resistance to liberalization and economic modernization may go hand-in-hand with the threat perceptions felt by religious practitioners. Democratic consolidation relies on “twin tolerations” between religion and state (Stepan, 2000); such arrangements may challenge the status of religious practitioners and foment backlash.
The case of Myanmar, for example, shows a direct relationship between religious cleavages and sudden shifts in a country’s social, economic, and politic structure; the past decade has seen violence against Myanmar’s Rohingya Muslim minority rise to a fever pitch amid a backdrop of liberalization, globalization, and indeed—secularization (Walton and Hayward, 2014; Zin, 2015).
In short, understanding the microfoundations of religious cleavages can better inform the social-scientific understanding of religion and its role in modernization. The framework provided in this study establishes a means of bridging the study of religion and state not only with secularization theory but also with broader questions of development as a whole.
Supplemental Material
sj-docx-1-psx-10.1177_00323217251323404 – Supplemental material for Defining the Contours of Religion, State, and Modernity
Supplemental material, sj-docx-1-psx-10.1177_00323217251323404 for Defining the Contours of Religion, State, and Modernity by Brendan Szendro in Political Studies
Footnotes
Acknowledgements
The author thanks the following for their input and support: Professors Gregory A. Robinson and Ekrem Karakoc (Binghamton University); Dr. Joseph B. Phillips (Cardiff University); Professor Richard Burke (Hampden-Sydney College); as well as to the reviewers and editors at Political Studies for their many helpful comments.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material
Additional Supplementary Information may be found with the online version of this article.
Appendix 1: Main Text Models with Controls.
Table A1: Main-text Model 1 with full controls.
Table A2: Main-text Model 2 with full controls.
Appendix 2: Alternate Samples.
Table A3: Model 1, Muslim majority sample.
Table A4: Model 2, Muslim majority sample.
Table A5: Model 1 using majority denomination only.
Table A5: Model 2 using majority denomination only.
Appendix 3: Alternate Models.
Table A6: Models 1 & 2 using categorical measures.
Table A7: Model 2 using Hierarchical Logit with QR decomposition.
Notes
Author Biography
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
