Introduction
Why do governments discriminate against religious minorities? There are many answers to this question, which are demonstrably multiple, complex, and often crosscutting. Significant contributing factors include secular and religious ideologies (Philpott, Reference Philpott2009, Reference Philpott2019; Stark, Reference Stark2003), alliances between governmental and religious institutions (Gill, Reference Gill2008; Sarkissian, Reference Sarkissian2015), the desire to protect a country’s culture (Helbling and Traunmüller, Reference Helbling and Traunmüller2016), fears of cults (Thomas, Reference Thomas2001), and a belief that religious minorities pose a political (Aísa and Larramona, Reference Aísa and Larramona2021; Bailey, Reference Bailey2021), and security threat to the state (Cesari, Reference Cesari2013).
This study focuses on an additional under-theorized and rarely tested motivation for governmental religious discrimination (GRD), its popularity. Specifically, we argue that self-interested politicians more often enact discriminatory policies when they believe publics favor or even demand it. In return, these policies significantly contribute to popular perceptions of a government’s legitimacy. This argument is often mentioned in passing but is rarely theorized as an important link between societal attitudes and real-world discrimination. Most scholarship tends to focus on pressure placed on politicians via lobbying and social institutions (e.g. Grim and Finke, Reference Grim and Finke2011) rather than on the societal attitudes of constituent publics. Similarly, those who examine how government religion policy influences governmental legitimacy tend to focus on governmental religious support (GRS) rather than GRD (Gill, Reference Gill2005, Reference Gill2008; Koesel, Reference Koesel2014; Sarkissian, Reference Sarkissian2015).
This argument comes with some important caveats. We do not argue that discrimination is always popular. Rather, we argue that when it is, discrimination is likely to increase. Also, we do not seek to replace existing theories on the causes of GRD, but rather examine an additional contributing factor whose importance has generally been neglected.
We focus empirically on the link between GRD as measured by the Religion and State project, round 3 (RAS3) dataset and two sets of variables from the World Values Survey (WVS) in Christian-majority countries: dislike of religious minorities, measured as not wanting members of other religions as one’s neighbor, and perceptions of confidence in government, political parties, and parliament as expressed by members of the majority religion.
Our focus on religious discrimination follows from the availability of rich religious minority-specific discrimination data, which is lacking for other identity groups on a global scale. We limit analyses to Christian-majority countries based on previous studies demonstrating that patterns of discrimination differ across religious traditions (Fox, Reference Fox2020; Fox and Breslawski, Reference Fox and Breslawski2023) coupled with insufficient survey data from non-Christian-majority countries. Our results show that GRD is more common and intense in countries where members of the majority religion oppose having members of other religions as their neighbors and that governments, legislatures, and political parties are more trusted in countries with higher levels of GRD across multiple metrics. While these results are specific to religious minorities, there are strong theoretical reasons to believe they should generalize to many kinds of politically-relevant societal out-groups.
This study proceeds as follows. First, we examine the existing literature on the causes of GRD and argue that common government motives for GRD may also explain why it can be popular. Second, we develop a theoretical framework for understanding the political utility of GRD. Finally, we present our research design and analysis and discuss the implications of our findings.
The causes of religious discrimination
Although there has been considerable research examining the causes of GRD (e.g. Akbaba, Reference Akbaba2023; Fox, Reference Fox2020), the argument that these policies may increase the popularity of sitting governments is present but under-theorized and largely untested. The empirical literature has uncovered multifaceted, cross-cutting, and complex causes and explanations for GRD. Key examples include religious and secular governmental ideologies, structural incentives for state control over religion, and the desire to protect the state and society from cultural, political, and security threats (Fox, Reference Fox2020). Most studies we discuss here emphasize the ideological and coalitional interests of leaders to engage in GRD. However, they rarely explicitly explore how such policies are received by their constituents nor how such responses may encourage or discourage GRD. In our discussion of GRD’s causes and correlates, we highlight how each approach offers theoretical inferences that are at least suggestive of our primary one. We then argue that a key mechanism linking these causes to GRD includes politicians’ expectations that they will be rewarded by their constituents when GRD is popular. This argument does not deny other theorized causes of discrimination. Rather, we posit that accounting for this additional cause can significantly improve our understanding of this troubling phenomenon.
Religious, ideologies, secular ideologies, intolerable practices, and anti-cult policies
Rodney Stark (Reference Stark2003: 32) summarizes the theorized link between religious ideology and GRD as follows: “Those who believe there is only One True God are offended by worship directed toward other Gods.” He argues that Christianity, like other religions, can contribute to faith in ultimate values that are not open to dissent. Although most discussion in the relevant literature generally focuses on religious intolerance and conflict, their arguments are fully compatible with the intuition that these beliefs should contribute to GRD. For example, Jelen and Wilcox (Reference Jelen and Wilcox1990: 69) argue that religion inculcates “ultimate values in its adherents – values which do not lend themselves to compromise or accommodation.” Nilsson and Svensson (Reference Nilsson and Svensson2020) and Köbrich and Hoffmann (Reference Köbrich and Hoffmann2023) similarly highlight religion’s capacity to limit even pragmatic accommodation. Sociologists Grim and Finke (Reference Grim and Finke2011: 46) further argue, “To the extent that religious beliefs are taken seriously and the dominant religion is held as true, all new religions are heretical at best. Thus, established religions will view the new religions as both dangerous and wrong.”
That the government’s constituents, especially the more religious ones, may share these values and therefore also desire discrimination is implied but rarely discussed. Saiya (Reference Saiya2019) and Henne, Saiya, and Hand (Reference Henne, Saiya and Hand2020) argue that one mechanism for this phenomenon is that religious organizations often seek to gain prominence by “outbidding” the state to demonstrate their religiosity. This often includes discrimination and even violent attacks on religious minorities. Many also argue that religious leaders outbid both each other and the state to gain popularity (Basedau and Koos, Reference Basedau and Koos2015; Birnir and Overos, Reference Birnir, Overos and Ratuva2019; Buckley and Wilcox, Reference Buckley and Wilcox2017; Isaacs, Reference Isaacs2017; Sandal, Reference Sandal2021), and violence itself can be a form of religious outbidding (Breslawski and Ives, Reference Breslawski and Ives2019; Zellman and Malji, Reference Zellman and Malji2023). While these arguments focus on societal harassment rather than government policy, the logic of outbidding assumes these behaviors increase the political popularity of involved parties.Footnote 1
Secular actors are not immune from these behaviors. While secularism is far from monolithic, including a diverse set of ideologies (Calhoun et al., Reference Calhoun, Juergensmeyer and VanAntwerpen2011; Philpott, Reference Philpott2009), some are overtly anti-religious. Philpott (Reference Philpott2019: 77) calls this “repressive secularism,” which “emanates from a western strand of thinking, vivified in the French Revolution, which holds that religion must be managed, controlled, and contained in order to make way for the modern state.”
These Western anti-religious manifestations of secularism evolved after the Thirty Years War and commonly see religion as dangerous, violent, irrational, and anachronistic, and advocate that it should be banished from the democratic public sphere (Casanova, Reference Casanova2009: 1052, Reference Casanova2011: 79–80). It correlates religion with “tradition, superstition, and supernaturalism and kindred categories, whereas secularity is aligned with modernity, rationality, and science” (Gorski and Altınordu, Reference Gorski and Altınordu2008: 61). Accordingly, “If they insist on clinging to religiosity, then legally and culturally religion should be a strictly private matter cordoned off from public life” (Hoover and Johnston, Reference Hoover and Johnston2012: 2). So understood, we would expect secular states to repress religion in general rather than have any particular interest in restricting religious minorities.
So why would such assertive or anti-religious secularism particularly target religious minorities? Fox (Reference Fox2020) argues this occurs for several reasons. Secular actors within a given culture are better acquainted with the majority religion so its practices are seen as normal and are much less likely to be construed as transgressive, anti-social, or inappropriate. Also, both secular and religious ideologies are intolerant of practices that contradict their core tenets. For religious Christians, this might include polytheistic practices they see as devil worship. Even more to the point, Jews and Muslims stand out as engaging in at least three shared practices that are commonly condemned by secular ideologies. 1) Male infant circumcision is criticized as a barbaric ceremony that violates a child’s right to bodily integrity. 2) Many secular actors consider female modest dress, particularly head-coverings, a form of women’s repression. 3) Both Kosher and Halal ritual slaughter have also been criticized as violations of animal rights (Fox, Reference Fox2020).
Religious ideologies can also motivate the banning of minority religious practices that the majority considers intolerable. For example, until 2006, cremation was banned in Greece because it is against the tenets of the Greek Orthodox Church. This effectively banned the funereal rites of religions such as Buddhism and Hinduism, which include cremation (Reuters, 2006; U.S. Department of State, 2006). In practice, cremation was not possible for almost a decade after the ban was removed due to a lack of cremation facilities in Greece (Boukas, Reference Boukas2013).
Anti-cult discrimination can also be motivated both by religious intolerance and secular fears of dangerous religions. Moreover, both mainstream religious and secular actors often associate cults with practices they consider intolerable. While definitions are contested, cults are commonly described as small often temporary religions that are new to a region (though not necessarily new outside that region). Their leaders are seen as charismatic and their ideological indoctrination as brainwashing, presenting a threat to both their members and society as a whole. As such, restrictions on cults considerably overlap with other sources of GRD (Peretz and Fox, Reference Peretz and Fox2021).
Religious monopolies and political threats
The religious monopoly approach, as applied to GRD, argues that governments support religious freedom or, alternatively, a religious monopoly when it is in their interests to do so. The motivation here is self-interested rather than ideological. Politicians desire to stay in power, rule more efficiently, and maximize resources. GRS can do all of this by legitimizing a government, which reduces opposition and increases moral behavior among the ruled (Gill, Reference Gill2005, Reference Gill2008; Sarkissian, Reference Sarkissian2015). However, there is general agreement that religious monopolies cannot be sustained without repressing minority religions (Casanova, Reference Casanova2009; Gill, Reference Gill2008: 45; Grim and Finke, Reference Grim and Finke2011: 70; Stark and Finke, Reference Stark and Finke2000: 199).
This argument is echoed in other literatures. For example, Aisa and Larramona (2021: 4) argue that “minority religions face discrimination because they represent undesirable competition to the state-supported religion.” Bailey (Reference Bailey2021) similarly reasons that religious minorities pose challenges to a government’s cultural hegemony. This encourages governments to recognize only a narrow range of religions which fit within their conception of national culture, while heavily restricting all others. Finke and Goff (Reference Finke, Goff, Flake and Oman2023: 46) concur that “governments seek restrictions on religion in an attempt to curb political, cultural, economic, or ethnic competitors. This often results in an alliance with a dominant religion and increased discrimination against minority religions.”
This literature also indicates that religious minorities are particularly likely to experience GRD if they are seen as threatening to the regime. This type of threat is often one of potential political opposition rather than a security threat. This can occur after demographic shifts in favor of the minority (Grim and Finke, Reference Grim and Finke2011: 46) or when the group otherwise engages in opposition activity (Gill, Reference Gill2008; Sarkissian, Reference Sarkissian2015). That is, “when the state views religions as offering ideological or political alternatives, the state’s restrictions of religious freedoms will increase” (Finke, Reference Finke2013: 300). As such, there is a clear recognition that both governmental support of a majority religion and repression of minorities may be in leaders’ political interest.
Nationalism and protection of culture
Although religion is not identical to culture, it is often integral to or even constitutive of it, particularly at the “national” level (Juergensmeyer, Reference Juergensmeyer1993; Kunovich, Reference Kunovich2006; Smith, Reference Smith2000). Accordingly, the imposition of GRD against minority religions may be a means to actively preserve and protect national culture from perceived outside influences. Helbling and Traunmüller (Reference Helbling and Traunmüller2016: 415) argue that “when the political, social, and cultural life of a public is defined by strong references to religious tradition, religious minorities pose a direct threat to this collective identity. Their practices are not easily accommodated because this would entail a loss in what is essentially perceived a zero-sum game.” Christian nationalism and populism, in particular, are often associated with intolerance toward religious out-groups and xenophobic attitudes (Broeren and Djupe, Reference Broeren and Djupe2024; Buckley et al., Reference Buckley, Brooke and Kleinsteuber2022; Davis et al., Reference Davis, Perry and Grubbs2024).
Ethnoreligious conflict and security risks
There is also a wide body of theory that understands discrimination against minorities in general as an outgrowth of conflict and societal insecurity. While most studies of minorities and conflict focus on discrimination as itself being a cause of conflict (e.g. Akbaba and Taydas, Reference Akbaba and Taydas2011; Basedau et al., Reference Basedau, Fox and Zellman2023; Gurr, Reference Gurr2000), those who address the topic also tend to argue that minorities perceived to be an enemy or a security threat are more likely to “attract” higher levels of GRD (Cesari, Reference Cesari2013; Deitch, Reference Deitch2022).
This direction of inquiry is closely complemented by securitization theory, which theorizes how democratic (and non-democratic) societies often justify repression of minorities, religious or otherwise (Sarkissian, Reference Sarkissian2015). This approach posits that if a group is “securitized”—depicted as a security threat by leaders using “speech acts”—this removes treatment of the group from the realm of ordinary politics. Once perceived as a security threat, policies such as discrimination, which were previously unjustifiable, become necessary to control or contain the securitized out-group (Buzan and Segal, Reference Buzan and Segal1998; Mabee, Reference Mabee2007; Wæver, Reference Wæver1995). While the only religious minority discussed extensively in this literature are Muslims in the West, the theory certainly has broader applicability (Laustsen and Wæver, Reference Laustsen and Wæver2000).
This theory was originally applied to security-related issues such as immigration restrictions and arrests of terror suspects. However, Cesari (Reference Cesari2013) and d’Appollonia (Reference D’Appollonia2016) argue that once a minority is securitized, leaders can create a hysteria that can also justify broader non-security related restrictions. This hysteria can easily extend to restrictions on minority religious activities and institutions, even when there is no obvious or perhaps only a tenuous connection between the actual activity and the presumed security threat. Baradaran-Robison, et al (Reference Baradaran-Robison, Scharffs and Elizabeth2004: 908) argue that in some cases protection of the national religion and national security are explicitly linked, noting, “National security is compromised without state protection of the favored religion. States respond to perceived national security threats by limiting the rights of minority religions through criminal punishment, anti-cult legislation, and government investigations” (also see Helbling and Traunmüller, Reference Helbling and Traunmüller2020).
Government versus Society
To this point, we have emphasized explanations for GRD that center on political and institutional causes. We have not examined how the literature engages with societal interests and pressures. To be sure, how leaders’ ideologies influence their policy preferences and how individual ideologies influence attitudes toward otherness are closely related. Religious ideologies can clearly contribute to individual intolerance directed at those who believe incorrectly, particularly religious minorities. Secular ideologies can make individuals intolerant of religious minorities who behave objectionably as well as intolerant of religion in general, especially as expressed by unfamiliar minority groups. Nationalism too is a classic motivation for prejudice. Fear and distrust of cults is common. Finally, individuals can also fear minorities seen as threats. In fact, securitization theory specifically argues that fear-mongering by politicians is intended to induce these feelings among the populace in order to advance their political agenda (Fox, Reference Fox2020).
Perhaps the only exception is the religious monopoly argument, which truly is more focused on the institutional incentives of religious control and discrimination than on societal parallels. When this literature does discuss repressing religious minorities, it generally focuses on these minorities as a political threat or that governments have an interest in gaining legitimacy (Koesel, Reference Koesel2014; Sarkissian, Reference Sarkissian2015). While this implies that governments have an interest in repressing religious minorities when this policy is popular, the argument is not well developed.
There is some precedent in the literature for our argument that societal factors can influence levels of GRD. However, the predominant focus is the link between societal religious discrimination and GRD, that is, how societal attitudes and behaviors encourage parallel government ones, rather than whether GRD contributes to a government’s popularity or legitimacy. For example, Grim and Finke (Reference Grim and Finke2011) argue that GRD has societal origins and is often the result of pressure on governments by religious institutions seeking to increase their religious monopoly. Grim and Finke (Reference Grim and Finke2011: 9) also argue that “the enforcement of any type of legal restrictions relies on social cooperation.” One example of this would be “the preexistence of widespread anti-Semitism throughout Europe prior to the Holocaust. This anti-Semitism eased the enactment of regulations against Jews and enhanced the enforcement of such regulations” (Grim and Finke, Reference Grim and Finke2011: 9). Their empirical analysis focuses on the link between societal restrictions on religion and GRD. Their only references to legitimacy focus on how governments legitimate discrimination, not whether discrimination makes the government more legitimate.
Some argue against this relationship altogether. For example, in his study of Indonesia, Buehler (Reference Buehler2023) finds no relationship between societal discrimination and human rights violations by the state. He argues that instead, “governments may emulate other governments and adopt laws and regulations discriminating against religious minorities without any immediate societal pressure preceding it” (Buehler, Reference Buehler2023: 468). Others such as Klocek, et al (Reference Klocek, Ha and Sumaktoyo2023) argue that both societal and religious discrimination are the result of the same underlying causes. This is consistent with our argument that the motivations for GRD can also exist at the societal level.
In sum, while the argument that government-based causes of GRD can also manifest from societal pressures is consistent with the current literature, it is an argument that is rarely made and, to the best of our knowledge, has never been empirically tested. In the next section we develop a more explicit theory to explain how this dynamic functions.
The political utility of GRD
To reiterate, we acknowledge that there are many causes of GRD ranging from the ideological to the structural. However, we also posit that a key additional motivation for states to engage in discrimination is the interests of politicians themselves. Namely, leaders pursue discriminatory policies targeting religious minorities because they understand that GRD is frequently popular and likely to pay considerable political dividends.
In part, this argument draws from the aforementioned literature which recognizes that discrimination targets out-groups who challenge behavioral, cultural, ideological, or moral norms of the dominant society. In turn, the consistent rhetorical representation of minorities as threatening via their securitization enables politicians to pursue policies that would otherwise be deemed illegitimate. Increased GRD should therefore also improve the political and popular standing of leaders who engage in such securitization.
We also expand upon a rational choice approach to religion and politics pioneered by Gill (Reference Gill2005, Reference Gill2008). He argues that government religion policy is determined by a politician’s interests. These interests can be distilled into three axioms: “[1] Politicians are primarily interested in their personal survival…[2] Politicians will also seek to maximize government revenue, promote economic growth, and minimize civil unrest…[3] Politicians seek to minimize the costs of ruling” (Gill, Reference Gill2008: 47).
Gill focuses on state support for religious monopolies—a condition where a single religion is the dominant religion in a country—arguing that when governments support a religious monopoly, it is usually because the government believes this policy will increase its legitimacy. This, in turn, should improve its chances at political survival and lower the costs of governing by reducing opposition. To be clear, Gill (Reference Gill2005, Reference Gill2008) does not argue that GRS will increase a government’s legitimacy under all circumstances and even argues that in some cases when it does, the costs may not be worth this benefit. However, when governments do choose to support a religious monopoly, he argues it is likely because they expect it to increase their legitimacy.
Many dispute the specific costs and benefits of GRS (e.g. Grim and Finke, Reference Grim and Finke2011; Henne et al., Reference Henne, Saiya and Hand2020; Koesel, Reference Koesel2014; Saiya, Reference Saiya2019; Sarkissian, Reference Sarkissian2015) or argue that modes of mobilization are contextual (Birnir and Satana, Reference Birnir and Satana2023). Others claim Gill downplays the role of religious ideology (e.g. Murphy, Reference Murphy2008; Philpott, Reference Philpott2009). However, none of these critics dispute that GRS can, under some circumstances, increase a government’s legitimacy. The majority of this conversation focuses on GRS, but, as we note earlier in this study, when the topic of religious minorities is addressed in discussions of religious monopolies, there is general agreement that these monopolies are not sustainable without repressing religious minorities. Thus, this repression is seen as an element of state support for religion and is generally not addressed separately.
Given this, we expand on these arguments in two respects. First, we argue that engaging in GRD can increase a government’s legitimacy in and of itself, not just as part of a more general policy of GRS. Second, the popularity of discrimination can motivate GRD even in cases where politicians do not find it in their interests to otherwise support a religious monopoly. Put differently, we disentangle GRS and GRD theoretically and argue they can have distinct and sometimes different impacts on a government’s legitimacy.
One empirical study does examine this relationship focusing on the link between GRS and trust in government, namely Fox and Breslawski (Reference Fox and Breslawski2023) who find that GRS, in general, decreases trust in government. They argue that this is because of the increasing popularity of political secularism and beliefs among religious people that religions are more authentic and likely to thrive when they are outside of government control. They include a control variable for GRD in their tests as a control and find it correlates with increased trust in governments and legislatures. However, unlike the present study, they neither theorize on this issue nor examine what types of GRD are more likely to influence trust in government.
Finally, there are multiple studies that demonstrate using prejudice against minorities can be a useful tool for political mobilization by political and especially populist parties (Engelhardt, Reference Engelhardt2021; Molle, Reference Molle2019; Newman et al., Reference Newman, Merolla and Shah2021). Recent scholarship agrees that trafficking in religious legitimacy enables populists to build, and in some cases consolidate, political power via identifying culturally-salient scapegoats for their (or their countries’) perceived cultural, economic, or political failures (DeHanas and Shterin, Reference DeHanas and Shterin2018; Yabanci and Taleski, Reference Yabanci and Taleski2018; yet also see Driessen, Reference Driessen2019; Eisenstein, Reference Eisenstein2008; Kühle, Reference Kühle2017; Philpott and Anderson, Reference Philpott and Anderson2017). Religion provides a wide array of rhetorical tools and political opportunities for populists, even when mobilizing an aggrieved religious majority is not their primary goal (Buckley et al., Reference Buckley, Brooke and Kleinsteuber2022; Gurses, Reference Gurses2014; Guth, Reference Guth2012). Given all of this, there is certainly a theoretical basis to argue that GRD may be sufficiently popular to increase levels of trust in government.
When combined with the insight that all theorized government-based causes of GRD can also manifest at the societal level, this suggests that when members of society support and desire GRD for these reasons, governments engaged in GRD will be more popular and, therefore, more legitimate and trusted by members of the religious majority. To put these assertions into more formal terms, we hypothesize as follows:
Hypothesis 1: States whose majority-religion citizens are more intolerant of religious others will have governments that engage in higher levels of GRD.
Hypothesis 2: Governments that engage in higher levels of GRD will be perceived as more legitimate by members of the majority religion, all else being equal.
Research design
To empirically evaluate these hypotheses, we conduct a series of statistical tests employing data drawn from the WVS, a global project that has, to date, conducted seven waves of standardized public opinion surveys, from 1981 to 2022, across 120 countries which include, in principle, nearly 90% of the world population (Inglehart et al., Reference Inglehart, Moreno and Welzel2022). Our state-level measures for treatment of religious minorities and religion in general are drawn from the Religion and State project, round 3 (RAS3). This country-year level dataset covers all countries with populations of at least 250,000 as well as a sampling of smaller states, 183 unique states in all, measured annually from 1990 to 2014 (Fox, Reference Fox2020).
In this article, we limit our analyses to Christian-majority states. We do so for two reasons. The first is theoretical: the manner in which states relate to religion as well as the policy consequences of these relations are often distinct across cultures and major religious denominations (Fox, Reference Fox2020). As such, to include multiple categories of religious-majority states in the same empirical analysis would run the risk of theoretically obscuring more than clarifying the relationship between state-religion policy and governmental legitimacy. The second is practical: despite the ostensibly broad coverage of the WVS, when considering religious categorizations, only Christian-majority countries enjoy both substantial spatial and temporal coverage. This enables us to draw empirically sound conclusions on the basis of this sub-sample for which we would have significantly less confidence if applied to other majority religion states, say those with Muslim or Buddhist majorities.
Given the temporal and geographic overlap between WVS and RAS3, we are able to examine 144 country-surveys conducted over 5 survey waves in 58 unique countries, with 201,761 respondents in all, 122,790 of whom we identified as members of each state’s majority Christian denomination. These countries include twenty post-socialist states, sixteen Western democracies, thirteen Latin American states, eight sub-Saharan African states, and the Philippines. Examined by Christian denomination, 24 are majority Catholic, 13 are majority Orthodox, 6 are majority Protestant, and a further 15 are majority Christian but have no obvious predominant denomination. We more specifically limit analysis to respondents who were identified as belonging to the state’s majority Christian denomination. We provide complete details regarding this country sample as well as the WVS participants captured by our study in the online Appendix.
Our tests employ this sample to examine (1) the correlation between negative popular attitudes by members of the religious majority against minorities on GRD and (2) the influence of GRD on popular confidence in three key political institutions: government, political parties, and parliament. We measure popular religious intolerance via WVS’s dichotomous variable A124_12, inquiring whether respondents would object to having members of another religion as a neighbor. Insofar as we theorize that politicians should enact discriminatory policies against religious minorities because they expect them to be popular, these effects should be most apparent via popular attitudes toward these institutions. To measure these effects, we draw from WVS survey questions regarding respondent “confidence” in each, scaled from (1) none to (4) a great deal. The relevant WVS variables are E069_11 for the government, E069_12 for political parties, and E069_07 for parliament.
To measure GRD, we rely on three variables from RAS3. The first is an index measure for all forms of GRD measured by RAS, defined as restrictions imposed upon minority religious groups and not imposed upon the religious majority by the state. This index includes 36 distinct forms of discrimination, each coded on an intensity score from 0, absence, to 3, sharp prohibitions on all or nearly all religious minorities in the state, for a maximum possible score of 108. In our first set of models, to address potential endogeneity concerns, use a year-led version of this variable, that is, testing the relationship between our covariates and levels of religious discrimination in the year after each given WVS country-survey was conducted. In our second set of models, we address endogeneity concerns in the opposite direction by lagging this measure by one year from the year in which each given WVS country-survey was conducted. The maximum observed score is 57 from Belarus in 2010, one year prior to the survey fielded there in 2011. The minimum observed score is 0, inclusive of Andorra (2004), Canada (1999, 2005), Estonia (1995), Hungary (1997), New Zealand (1997, 2003, 2010), Philippines (1995, 2000, 2011), South Africa (1995, 2000, 2005), and Uruguay (1995, 2005, 2010). It is important to emphasize that these scores may not reflect broader trends across the whole temporal range of RAS3. For instance, while Hungary had a GRD score of 0 in 1997 and 2 in 2008, religious discrimination in Hungary jumped to a GRD score of 14 in 2011.
We also employ two sub-measures of GRD, also calculated from individual discriminatory policies, that examine the two most pervasive forms of religious discrimination in the Christian-majority world. The first are restrictions on the religious practices of minority groups, such as curtailing their observance of religious holidays, rituals, dietary guidelines, public dress, and life cycle events (marriage, burial, etc.). Again, the most significant offender is Belarus in 2010 with the maximum observed score of 16, followed by a score of 11 for Belarus again in 1995 and Russia with a score of 10 in 2005 and 2010. The second are discriminatory measures against religious minority institutions and clergy, including restrictions on the construction, maintenance, and use of places of worship, access by minority clergy (chaplaincy) to jails, military bases, and hospitals otherwise granted to the majority religion, ordination of minority clergy, and legal registration. Here, both Belarus (2010) and Russia (2005, 2010) received the maximum observed score of 20, followed again by Belarus (1995) and Armenia (1998, 2010) with scores of 16 and 15, respectively.
In each model, we also employ a host of control variables. From RAS3, we employ an additive index for Governmental Religious Support (GRS), in light of recent findings demonstrating that greater such support correlates with significantly reduced confidence in government (Fox and Breslawski, Reference Fox and Breslawski2023).Footnote 2 This variable measures 52 different types of said support, including laws on relationships, sex, and reproduction, institutions that enforce religion, government funding for religion, and various forms of state-religion institutional entanglement. As with GRD, in our second set of tests we lag this variable by one year to address endogeneity concerns. The maximum level of GRS is at score 16 observed in Sweden (1995, 1998), while the lowest level of GRS is at score 2 observed in Rwanda (2006, 2011), South Africa (2000), Uganda (1994, 1998), Uruguay (1995, 2005), and the United States (1994, 1998).
From the WVS, we derive a number of individual-level control variables commonly theorized to correlate with both popular attitudes toward religion and government. These include an ordinal measure for regularity of church attendance, scaled from 1 to 8 wherein 1 denotes zero attendance and 8 denotes attendance greater than once a week. We also include dichotomous measures for whether survey participants self-identify as either atheist or religious. At the state level, we include an overall measure of the percentage of respondents who report attending church at least once a month, the percentage who self-identify as atheist, and the percentage of those who self-identify as religious. Returning to the individual level, we include WVS’s ten-point ordinal scale for life satisfaction, an eight-point scale for highest level of education received, a five-point scale for self-identified social class in descending order, and a ten-point scale for political orientation from left to right.
From the World Bank Word Development Indicators, we include logged GDP per capita and logged state population (World Bank, 2022). From the Polity dataset, we include the polity2 democracy score, an ordinal measure of country regimes on a scale of -10 being the most autocratic to +10 being the most democratic, as well as Polity’s regime durability score, which counts the number of years since the last major change in a given country’s regime score (Marshall and Jaggers, Reference Marshall and Jaggers2009). We also include a dichotomous control for post-Soviet states, given their considerably uniform higher levels of irreligiosity. In the online Appendix, we alternatively control for all world regions, with western democracies as the index category, to no differential effect. From the Religious Characteristics of States-Demographics dataset, we draw a percentage measure of each country’s population belonging to the religious majority as well as a score for religious diversity by country-year (Brown and James, Reference Brown and James2018). This latter value ranges between 0 and 1 and represents the likelihood that any two random people in a given country will belong to the same religious group.
In our statistical analyses examining the influence of popular religious intolerance on GRD, we employ linear models with robust standard errors clustered to country-survey. Given that GRD is measured at the country rather than individual level, we conduct our analyses on the effect of religious intolerance, using the country-year percentage of respondents who indicated their objection to having neighbors of a different religion. These results are presented in Table 1, wherein each column corresponds to the use of a different dependent variable: the General GRD index, GRD-Practices, and GRD-Institutions, respectively.
Table 1. Influence of religious intolerance on GRD in Christian-majority states
Robust standard errors, clustered by country-survey, are in parentheses.
*** p < 0.001, ** p < 0.01, * p < 0.05.
In our analyses examining the influence of GRD on government legitimacy, we employ multilevel mixed models. Multilevel models are the most appropriate approach when examining WVS survey responses, as our dependent variables of interest are based upon individual-level responses nested within larger country contexts. This is also why these models are inappropriate for our first set of analysis. More specifically, we employ mixed effects ordered logits, as values for each of our dependent variables are both ordinal and constrained to a relatively short range between 1 and 4. Because the coefficients of ordered logits cannot be readily directly interpreted, we discuss our substantive results in terms of odds ratios, the full results of which we provide in the Appendix. We present our primary results in Tables 2, 3, and 4, wherein each table corresponds to the use of the General GRD index, GRD-Practices, and GRD-Institutions as the independent variables of interest, respectively. Each table contains three models examining popular confidence in government, political parties, and parliament, respectively.
Table 2. Influence of GRD-general on confidence in state institutions by majority Christians
Standard errors are in parentheses.
*** p < 0.001, ** p < 0.01, * p < 0.05.
Table 3. Influence of GRD-practices on confidence in state institutions by majority Christians
Standard errors are in parentheses.
*** p < 0.001, ** p < 0.01, * p < 0.05.
Table 4. Influence of GRD-institutions on confidence in state institutions by majority Christians
Standard errors are in parentheses.
*** p < 0.001, ** p < 0.01, * p < 0.05.
Empirical analyses
The results of our linear regression analysis in Table 1 clearly and directly confirm H1, which proposes that states whose majority-religion citizens are more intolerant of religious others demand from their governments more intolerant policies toward religious minorities. These outcomes are significant across all three tests, gauging year-lead measures of General GRD, across the entire spectrum of religiously discriminatory policies measured by RAS3, as well as for the more specific measures of GRD targeting religious minority practices and institutions. These findings hold even when controlling for GRS and domestic levels of religiosity, both factors commonly theorized to explain GRD. It is also encouraging that our findings are statistically significant across all three models, despite the rather constrained sample for our tests, based on available data from WVS. Although likely less appropriate given that all GRD measures are specific to country-years rather than individual respondents, these same results hold when running linear regressions on the full religious-majority WVS sample. These robustness checks, as well as alternative models employing ordered logits, with again essentially identical results, are available in the Appendix. Altogether then, the first stage of our causal model, wherein we argue that governments are more likely to enforce higher levels of GRD when such discrimination is popular, are strongly confirmed.
For the second stage of our explanatory model, we evaluate H2, which expects higher levels of GRD, themselves demonstrated to be at least partially the product of the popularity of anti-religious minority discrimination, should also translate into legitimacy gains for state political institutions and therefore political leaders themselves. In Table 2, we find a strong and highly significant correlation between a state’s year-lagged level of GRD and the degree to which members of its religious majority express confidence in its political institutions, whether the government, political parties, or the parliament, with each coefficient significant with at least 99.9% confidence. Interpreted in terms of odds ratios, for each additional level of GRD, our models predict a 3.4% increasing likelihood of expressing confidence in government, 1.5% increasing likelihood of expressing confidence in political parties, and 2.1% increasing likelihood of expressing confidence in parliament. Further encouraging for the validity of our models are our findings that increasing year-lagged levels of state support for religion (GRS) are consistently and significantly associated with reduced confidence in these political institutions, confirming Fox and Breslawski (Reference Fox and Breslawski2023). We also find increasing confidence in these political institutions to be consistently correlated with higher levels of life satisfaction, education, more right-wing political views, and higher levels of political democracy, generally speaking.
Faced with the potential critique that RAS3’s GRD index is too broad to reliably capture the “true” relationship between anti-religious minority discrimination and government popularity, we turn to our sub-measures. In Table 3, we present nearly identical models to Table 2 with the exception that we substitute RAS3’s measure of governmental restrictions on minority religious practices for the general GRD index. Here we again find that increasing levels of year-lagged discrimination are closely and significantly correlated with popular approval for all political institutions examined with at least 99.9% confidence. Per the models’ odd ratios, we find a 14.5% increased likelihood of confidence in government, 11.0% increased likelihood of confidence in political parties, and 11.3% increased likelihood of confidence in parliament with each increasing level of these forms of discrimination. Given that the observed range of this variable is considerably shorter than that of the general GRD index (0–16 versus 0–57), we should be cautious in assuming that restrictions on minority practices are necessarily more influential on confidence in political institutions, being roughly a 1:3.6 ratio. To be clear, whereas a one unit change in discrimination versus religious practices is associated with the above mentioned 14.5% increase in the likelihood of confidence in government, a 3.6 unit equivalent change in the general GRD measure is associated with only a slightly smaller increase at 12.8% (being the GRD odds ratio in model 1a to the power of 3.6).
Finally, in Table 4, we present nearly identical models to the previous two tables but substitute RAS3’s measure of governmental restrictions on religious minority institutions. Here we observe highly significant correlations with 99.9% confidence between increases in these restrictions and confidence in government and parliament, however we detect no such significant correlations for confidence in political parties. In terms of odds ratios, we observe a 7.3% increase in the likelihood of confidence in government and a 5.1% increase in the likelihood of confidence in parliament with each additional level of restrictions on religious minority institutions. Here the comparable ranges between these more specific discriminatory measures and the general GRD index (0–20 versus 0–57) are similar to those for discrimination against religious minority practices, a roughly 1:2.9 ratio. An equivalent 2.9 unit change in the general GRD measure is therefore associated with a 10.2% increase in confidence in government, suggesting general GRD better accounts for increased popular support in political institutions than discrimination targeting religious minority institutions.
Directly comparing odds ratios for discrimination against religious minority practices versus religious minority institutions, again vis-a-vis government, their respective value ranges (0–16 versus 0–20) have a 1:0.8 ratio. Here an equivalent 0.8 unit change in the level of discrimination against religious minority institutions is associated with a 5.8% increase in the likelihood of confidence in government. This compared to a 14.5% increase in the likelihood of confidence in government based upon increased discrimination against religious minority practices strongly suggests that it is these forms of discrimination, rather than those directed against religious minority institutions, that garner political leaders their greatest legitimacy gains among members of the religious-majority public.
These patterns are well illustrated by the cases of Hungary and Belarus, two Eastern European states with divergent levels of anti-religious minority sentiment, governmental discrimination against religious minorities, and confidence in government. In Hungary in 2009, 5.6% were opposed to having an other religion neighbor (34th percentile of all included country-surveys), the country had a discrimination score of 2 (31st percentile), and respondents had remarkably low levels of confidence in government (3rd percentile). By contrast, in Belarus in 2011, 17.7% expressed intolerance of religious others (78th percentile), it had the highest observed GRD score (57), and respondents expressed high confidence in government (87th percentile).
Conclusions
The argument that discrimination is popular is not new. However, this potential cause of discrimination is rarely the focus of academic discussions. It is also seldom theorized as a result of politicians following popular sentiment. For example, Grim and Finke (Reference Grim and Finke2011: 9) state that “legal restrictions on religion as well as the easing of legal restrictions arise from social origins.” However, they focus on social institutions and factors as the mechanism for this relationship rather than the interests of politicians arguing that “popular religious movements, religious plurality, immigration patterns, political stability, and economic interests have all driven changes in the legal regulations placed on religion around the globe” (Grim and Finke, Reference Grim and Finke2011: 9).
Similarly, the religious monopoly literature focuses on the interests of politicians in supporting a religious monopoly, yet it pays far less attention to restricting religious minorities (Gill, Reference Gill2005, Reference Gill2008; Koesel, Reference Koesel2014; Sarkissian, Reference Sarkissian2015). When it does focus on religious minorities, rather than addressing the popularity of discrimination, it discusses the necessity of repressing minority religions to maintain a religious monopoly. Our argument is broader in that it posits that the popularity of discrimination can be a motive for engaging in GRD, even for governments that do not find it in their interests to support a religious monopoly.
We find that GRD is higher in Christian-majority countries with higher levels of dislike for religious others and that governments, legislatures, and political parties experience higher levels of confidence when they engage in higher levels of GRD generally as well as in higher levels of restrictions on religious institutions and practices. This, we argue, demonstrates that the popularity of religious discrimination is likely among its multiple and often cross-cutting causes.
Even more importantly, we find that the results for GRD diverge from those of supporting religion. Replicating previous results (Fox and Breslawski, Reference Fox and Breslawski2023), we find that state support for religion is associated with a decrease in trust in government, legislatures, and political parties, at the same time as GRD correlates with an increase in trust in these institutions. This demonstrates our argument that the popularity of GRD is independent of any motives politicians might have for supporting a religious monopoly. That is, our results are inconsistent with theorization in the religious monopoly literature, which views restrictions on religious minorities as an inevitable requirement for politicians interested in gaining legitimacy through supporting a religious monopoly. If this were the case, both GRD and state support for religion should be associated with an increase in legitimacy.
Our finding that GRD is higher where dislike of religious minorities is higher suggests that GRD is higher where it is popular. Our finding that only GRD is associated with an increase in trust in government, legislatures, and political parties while support for religion is associated with a decrease in trust in these institutions indicates that it is indeed the popularity of GRD driving this result. This also suggests that this popularity derives more from majority–minority dynamics than government religion policy. While our results are specific to GRD against religious minorities, they imply that this demonstrated political utility is likely also an important motivation for discrimination against a broader set of minorities, particularly ethnic ones (e.g. Mansfield and Snyder, Reference Mansfield and Snyder2005; McGauvran and Stewart, Reference McGauvran and Stewart2021).
As the popularity of GRD constitutes a significant temptation for politicians, it is imperative that religious freedoms be enshrined in constitutions and other difficult to amend legal frameworks, combined with strong independent judiciaries to preserve these protections against opportunistic political meddling. Such an approach may also be beneficial to politicians themselves, insofar as governmental discrimination against minorities has also been shown to contribute to higher levels of societal violence and unrest (Henne et al., Reference Henne, Saiya and Hand2020; Saiya, Reference Saiya2019). Accordingly, the linkage between popularity of discrimination and the societal instability it tends to produce are a yet further important agenda for future research.
Data availability statement
The dataset, do-files, and online supplementary appendix for the empirical analysis of this article can be found at the corresponding author’s Dataverse at https://doi.org/10.7910/DVN/OHG5FW.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S175577392400033X.
Funding declaration
This research was supported by the Israel Science Foundation [Grant 23/14], the German-Israeli Foundation for Scientific Research and Development [Grant 1291-119.4/2015], and the John Templeton Foundation.
Competing interests
The authors have no competing nor conflicting interests related to this article, the data analyzed, nor the results of this study.
Open access
