Do Religion Clauses in Constitutions Predict Government-Based Discrimination against Religious Minorities?

: This study examines whether religion clauses in countries’ constitutions predict levels of government-based discrimination (GRD) against religious minorities. Using the Religion and State (RAS) and Religion and State-Constitutions (RAS-Constitutions) datasets, I ﬁnd that clauses declaring ofﬁcial religion, separation of religion and state (SRAS), and religious freedom have no signiﬁcant inﬂuence on GRD. In fact, 152 of 154 types of religion clauses found in constitutions measured by the RAS-Constitutions dataset do not signiﬁcantly predict GRD. However, constitutional clauses banning religious hate speech and protecting the right to not be religious are associated with higher levels of GRD. I theorize that these causes likely represent an anti-religious form of secularism, which can be hostile to religious minorities.


Introduction
This study has two purposes.First, it introduces the updated Religion and State Constitutions (RAS-Constitutions) dataset, which includes information on 154 types of religion clauses in 176 constitutions from 1990 to 2022.Second, it asks whether these constitutional clauses provide a good method to predict whether a state will engage in government-based religious discrimination (GRD) using data from the Religion and State (RAS) dataset.This question of constitutional clauses' influence on GRD provides a test of the usefulness of the RAS-Constitutions dataset and looks at issues that are in and of themselves important, including the question of whether constitutional declarations are good indicators of policy, especially religion policy.
I find that, for the most part, constitutional clauses do not significantly influence GRD.Specifically, when controlling for other factors, constitutional clauses declaring an official religion, separation of religion and state (SRAS), and religious freedom clauses, as well as most other clauses that deal with religion, do not significantly 1 predict levels of GRD.However, statements in constitutions banning religious hate speech and protecting the right not to be religious are statistically significantly associated with higher levels of GRD.I hypothesize that this is because they represent an anti-religious form of secular ideology among governments, which can result in GRD.This is consistent with the growing literature that argues secularism can be a source of intolerance toward religion.
This study is important for three reasons.First, the question of whether countries, particularly democracies, follow their constitutions is important for understanding governance.Second, the causes of GRD are important because it is an important measure of religious freedom (Fox 2020(Fox , 2021) ) and the presence or lack of religious discrimination impacts a wide variety of factors, including religious minority conflict behavior (Deitch 2022;Basedau et al. 2017Basedau et al. , 2019Basedau et al. , 2022Basedau et al. , 2023;;Finke et al. 2017), ethnic conflict (Akbaba and Tydas 2011;Fox 2002Fox , 2004)), terrorism (Saiya 2019), the spread of foreign fighters (Mishali-Ram and Fox 2021), the legitimacy of governments (Fox et al. 2019), foreign policy (Haykin et al. 2021;Zellman and Fox 2022), and international conflict (Zellman and Fox 2020).Finally, it introduces a new and important dataset.
This study proceeds as follows.First, I examine the questions of why religion clauses in constitutions might impact GRD and why states would or would not follow their constitutions.Second, I introduce the RAS-Constitutions dataset.Third, I examine the impact of constitutional clauses on GRD.This includes a discussion of why religious and secular ideologies might influence GRD.Even though the RAS-Constitutions dataset has data through 2022, the RAS dataset that contains the GRD variable is current only through 2014.For this reason, the empirical portion of this study focuses on 2014.

Why Would Constitutions Impact Government-Based Discrimination against Religious Minorities?
It is reasonable to assume that constitutional religion clauses and freedom of religion are linked because constitutions, at least in theory, reflect a society's ideals and choices (Murphey 1993;Keith 2002).Constitutions are statements of how the framers represent societal attitudes as well as the views of political leaders at the time of the constitution's writing.As constitutions can be amended or replaced, it is fair to assume that major changes in these attitudes and views would eventually be reflected in amendments to constitutions.Nevertheless, this expectation that countries follow their constitutions is countered by evidence that this does not always occur.
The expectation that countries generally follow their constitutions is particularly strong for democracies.Many Western social scientists assume that democratic states, especially those with SRAS and religious freedom clauses in their constitutions, protect religious freedom (e.g., Madeley 2003, pp. 4-5;Philpott 2007, pp. 509-10;Stepan 2000).Yet, most democracies, including Western democracies, engage in GRD (Fox 2016(Fox , 2020;;Fox and Akbaba 2015;Grim and Finke 2011;Peretz and Fox 2021).The RAS project defines GRD as restrictions on the religious institutions or practices of religious minorities that are not placed on the majority religion (Fox 2020).This is a clear violation of both religious freedom and SRAS.Thus, a systematic examination of whether and to what extent constitutional religion clauses influence GRD is warranted.
This opens the question of why would a country not follow its constitution?The general literature on constitutions provides several reasons for a potential disconnect between constitutional clauses and practice.Not all governments are able and willing to translate constitutional precepts into policy.They can lack the political will or even the desire to do so, an issue that is particularly prominent in authoritarian states (Davenport 1996, p. 629).In these states, constitutional clauses do not reflect the ideals of those who run the government but are nevertheless written into constitutions because the rulers wish to appear to be supporting these ideals both domestically and internationally (Davenport 1996).
Even democracies do not always fully observe their constitutions for several reasons.First, for new democracies, constitutional changes often require implementing legislation that is not always forthcoming (Finkel 2004, pp. 58-59;Goodliffe and Hawkins 2006, pp. 363-64;Van Cott 2000).Second, functioning democracies require a strong independent judiciary, protections for minority political parties, an independent media, a politically neutral and professional civil service, and both the ability and political will to fight corruption (Bugaric 2008;Fombad 2007).In fact, Finke and Martin (2014) found that an independent judiciary increases levels of religious freedom.
Third, sometimes national governments respect their constitutions, but local governments do not.Nigeria presents an extreme example of this phenomenon.Article 10 of Nigeria's 1999 constitution bans the adoption of any religion as the state religion by both the national and regional governments.While the national government follows this provision, many of Nigeria's Muslim-majority states began adopting aspects of Sharia (Muslim religious) law as state law beginning in 2000 (Fox 2008).Thus, federal states can present complicated cases.
Fourth, courts often allow governments to ignore or circumvent their constitutions.Sometimes, this is necessary when different constitutional clauses are in tension with each other, and courts need to decide which is given priority (Beatty 2001;McConnell 1992).While this is not necessarily an issue of disregarding constitutional ideals, it does mean that specific constitutional clauses examined outside of the context of other relevant clauses may not necessarily constitute reliable indicators of government policies.As is discussed below, constitutions themselves often qualify religious freedom clauses, allowing governments to limit religious freedom for reasons such as public health and national security, among many others.Morrison (2006) argued that, in extreme cases, some courts allow democracies to follow a doctrine of constitutional avoidance.In addition, courts are often influenced by the norms and ideologies of their members, which can allow a wide range of interpretations (Bader 1999, pp. 601-2;McGraw and Richardson 2019).
Given all of this, the question of whether constitutional clauses on religion influence religious freedom in general and GRD specifically is an open one.

Introducing Religion and State-Constitutions (RAS-Constitutions) Dataset Round 2
The RAS-Constitutions dataset examines the constitutions of 176 countries worldwide, covering from 1990 to 2022.The previous version covered from 1990 to 2008 for 169 constitutions.All variables are coded yearly, though they only change when constitutions themselves change.The dataset includes 154 variables that examine:

•
Whether the constitution declares either an official religion SRAS or does not address the issue as well as a categorization of the specific wording of these clauses.

•
SRAS clauses that focus on more specific issues or policy areas where the government is separate from religion or religion separate from the government.

•
Clauses protecting the freedom of religion, including the enumeration of specific freedoms that are encompassed within that right.

•
Clauses qualifying religious freedom under certain circumstances.

•
Clauses banning specific government actions that may impinge on religious freedom.

•
Other freedoms that can be limited by religious concerns.

•
Types of laws that are subject to or associated with religious precepts.

•
Symbolic references to religion.

•
Bans on discrimination on the basis of religion.

•
Other religion clauses that do not fit into the above categories.
I list all of these clauses and discuss this categorization in more detail in the data analysis sections below.In that section, I also test the impact of each of these variables' impact on GRD for variables which are sufficiently common for the assumptions of statistics to hold.
The RAS-Constitutions data were collected by obtaining English-language translations of constitutions from academic sites such as www.religlaw.org(accessed on 30 November 2022) and https://www.constituteproject.org(accessed on 30 November 2022).When these sites did not provide copies, they were obtained from other academic sources or official government websites.In cases where they were not available in English, we used Google Translate.
Each constitution was searched for multiple religion keywords, including but not limited to religion, religious, the names of all religions present in the country, God (and the specific names of other relevant deities), conscience, Bible, Koran, holy, sacred, etc. Coders identified all clauses relevant to religion and used this as a basis for the codings.All codings are 1 if a clause is present in a constitution and 0 if it is not.
Both the RAS-Constitutions and the RAS dataset use the country-year as the unit of analysis.All constitutions included in the RAS-Constitutions dataset are national constitutions, and the dataset does not include information from sub-units of federal governments such as provinces in Canada or individual states within the United States of America.
The RAS dataset's GRD variables include restrictions on religious institutions or practices of minority religions that are not placed on the majority religion.RAS codes 36 types of GRD, each on a scale of 0 to 3, which includes restrictions placed by both national and local governments that are combined into a scale that runs from 0 to 108.In cases where only some local governments (such as some states within the United States) engage in a form of GRD, the commonality of this type of GRD across states within the United States is taken into account in the scale for that sub-variable.For more on how the RAS dataset was coded as well as reliability tests for the data, see (Fox 2011(Fox , 2016(Fox , 2020;;Fox et al. 2018).
All multivariate tests and significance scores in this study were based on OLS regression using SPSS version 28.0.1.1.Both datasets are available at www.religionandstate.org(accessed on 30 November 2022).

Official Religions, Separation of Religion and State, and Religious Discrimination
Perhaps the most basic distinction that a constitution can make with regard to religion is what is the larger role of religion in government?There are three approaches in the world's constitutions, declaring a religion the official state religion, declaring separation of religion and state (SRAS), or simply failing to address the issue.As shown in Table 1, each option is present in a good portion of the world's constitutions.

Total N 176
Note: Many constitutions contain multiple wordings.This table shows all constitutions that contain a particular wording.
These clauses are potentially important because they are, at least in theory, fundamental statements about a governments policy and attitude toward religion.Studies have shown that both a strong connection with religion and anti-religious secularism can cause higher levels of GRD (Fox 2016(Fox , 2020)).In the below sections, I discuss why these factors might be expected to cause GRD.

The Potential Influence of Official Religion Clauses on GRD
The official religion and SRAS options can be written into constitutions using different types of wording that, I argue, are potentially of theoretical import when examining the impact of constitutional clauses on GRD.This is because both religious and secular ideolo-gies are predicted to cause GRD.Official religions, at least in theory, imply a commitment to support that religion and give it preference over all others.While this tends to be true in practice, there are exceptions and a considerable amount of variation in actual levels of support and commitment among states that declare official religions and those that declare SRAS (Fox 2015).
Why would this close relationship between a state and a single religion result in GRD?There are two arguments in the literature.First, Rodney Stark (2003, p. 32) argues that "those who believe there is only One True God are offended by worship directed toward other Gods".That is, believers in a religion deny the truth, authenticity, and legitimacy of religious beliefs and practices other than their own.This type of argument can be found across the social sciences.Sociologists Grim and Finke (2011, p. 46) argue that "exclusive religious beliefs provide motives for promoting the 'one true faith.'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".Political scientists such as Jelen and Wilcox (1990, p. 69) argue that "religion is often thought to inhibit the development of the tolerance for unorthodox beliefs and practices . . .Religion is accused of inculcating ultimate values in its adherents-values which do not lend themselves to compromise or accommodation".International relations theorists such as Laustsen and Waever (2000, p. 719) argue that "religion deals with the constitution of being as such.Hence, one cannot be pragmatic on concerns challenging this being".Finally, psychologists such as Silberman (2005, p. 649) argue that "once they are constructed, collective meaning systems tend to be viewed within a given group as basic undisputable truths.Accordingly, they are usually held with confidence, and their change or redirection can be very challenging".
Given this, we would expect that governments that are linked to a religious belief system are more likely to be intolerant of and restrict the religious practices of other religions.As the RAS' GRD variable precisely measures restrictions on religious practices and institutions of minority religions, we expect governments that declare an official religion to engage in higher levels of GRD than governments which do not.
The second argument is based on Gill's (2008) rational choice approach to religion, which, in turn, is based on the economic approach to religion developed by sociologists seeking to understand the impact of religious "monopolies" (their term for a state that is in some manner religiously exclusive) on religiosity 2 (Barro and McCleary 2003;Finke 2013;Finke and Iannaccone 1993;Fox and Tabory 2008;Iannaccone 1995aIannaccone , 1995b;;Koev 2022;Stark and Finke 2000;Stark and Iannaccone 1994).Gill (2008) focuses on the motivations for a religious monopoly and its impact on religious freedom.He argues that when governments support a religion, it is because they believe they will gain important benefits.These include increased legitimacy, which reduces opposition and increases compliance with the government, and increased morality, which reduces law enforcement costs.Most scholars who address the topic argue that an element of supporting a state religion is repressing competing religions (Casanova 2009;Froese 2004, p. 36;Grim and Finke 2011, p. 70;Stark and Finke 2000, p. 199).Thus, based on this, we expect official state religions to be associated with higher levels of GRD.
A growing literature that builds on this theory dirties the water a bit on this issue.While few critics question the motivations that Gill (2008) attributes to governments, they argue that he misses many additional motivations that would both increase and decrease the benefits of supporting a religion as opposed to supporting religious freedom.Thus, this literature discusses additional factors that make governments either more or less likely to support a religion.Koesel (2014) argues that religious organizations provide social welfare services at less cost to governments through their charity work.The literature also posits several factors that make supporting a religion less desirable as opposed to supporting religious freedom.Repression has material costs that might outweigh any legitimacy benefits (Sarkissian 2015).State religions can also build social trust (Fox et al. 2021).Saiya (2019), Saiya andManchanda (2019), andHenne et al. (2020) argue that state support for religion increases religious extremism and terrorism by members of the majority group against both religious minorities and members of the majority perceived as insufficiently religious.Grim and Finke (2011) similarly argue that religious freedom creates a level playing field between religions that reduces conflict.Some factors are more complex.For example, supporting religious institutions can result in coopting them, which makes them less likely to be a basis for opposition (Fox 2015), but this support can also make these institutions stronger and more capable of supporting the opposition (Sarkissian 2015).Finke (2013, p. 300) argues that even atheist governments can gain legitimacy from supporting religion; thus, a religous monopoly may not be necessary to gain this benefit.Some argue that other factors mediate this relationship, including the state's bureaucratic structure (Mayrl 2015), the population's religiosity (Buckley and Mantilla 2013, p. 345), and regime (Schleutker 2018).Nevertheless, despite all these complex and cross-cutting political incentives, according to the rational choice literature, if politicians decide to support a religion, then one expects that this could increase levels of GRD.
The exact wording of constitutions that establish a state religion can be important in this context because they can convey the intentions of the constitution's writers.There is little variation in official religion clauses because, as shown in Table 1, the overwhelming majority of constitutions that declare an official religion simply declare that there is an established, official, or state religion.However, Afghanistan, Bahrain, Iran, Mauritania, Morocco, and Oman declare the country a religious state.Bahrain actually has multiple clauses.Article 1 states "The Kingdom of Bahrain is a fully sovereign, independent Islamic Arab State . . .". Article 2 states "The religion of the State is Islam".Thus, it has both types of clauses.Declaring a country a religious state has different implications than declaring that the state has a religion.The former, at least semantically, implies that the state is beholden or subject to the religion in some manner, while that latter says the state has a religion.Put differently, at least semantically, religious state clauses place religion over the state, and established religion clauses have the state as the prime actor, which makes a religion its official religion.This conjecture is confirmed because states with religious state clauses have a higher mean level of GRD of 34.50 compared to a mean of 30.14 for states where a religion is declared the state religion, official religion, or established religion.Unfortunately, this cannot be tested in the multivariate analysis later in this section due to the low number of cases in the religious state category.

The Potential Influence of SRAS Clauses on GRD
The potential influence of SRAS on GRD is also complex.While classical liberal arguments hold that a secular government that maintains SRAS provides the highest level of religious freedom, this argument has come under increasing challenge.Stepan (2000) provides one of the most cited versions of the classic argument that SRAS is an essential element of liberal democracy that guarantees religious freedom: Religious institutions should not have constitutionally privileged prerogatives that allow them to mandate public policy to democratically elected governments.At the same time, individual religious communities . . .must have complete freedom to worship privately.In addition, as individuals and groups they must be able to advance their values publicly in civil society and to sponsor organizations and movements in political society, as long as their actions do not impinge negatively on the liberties of other citizens or violate democracy and law.(Stepan 2000, pp. 39-40) He follows in the tradition of Rawls ' (1993, p. 151) normative injunction to "take the truths of religion off the political agenda".This is expected to reduce religious conflict and increase religious freedom, particularly for religious minorities.
Thus, these arguments posit that an absence of state support for religion is necessary for religious freedom for precisely the same reasons that theorists expect state support for religion to reduce religious freedom.This is a common argument.For example, Breskaya et al. (2018, p. 423) argues that "minority faiths can function more openly in a society that has a constitutional and statutory structure offering at least minimal guarantees for religious freedom".Brathwaite and Bramsen (2011) elaborate that established religions create a perception of superior and inferior citizenship, and legislation favoring one religion puts other religions at a disadvantage.Thomas (2005, p. 216) points out that SRAS does not mean that society is secular and can in fact buttress religious freedom.
Many also argue that SRAS creates an environment where religious radicals are less likely to flourish and encourage violence.This is because favoring a single religion provides a safe environment for the growth of religious radicalism and violence.In contrast, a society where the government endorses no religion allows moderates to compete effectively with radicals and deprives fringe groups of their legitimacy (De Juan 2008;Henne et al. 2020;Saiya 2019).
In opposition to the classical argument, there is an increasingly large literature that argues that secularism and SRAS policies can undermine religious freedom and have results detrimental to religious minorities.Many of them center on the argument that secularism as an ideology can be anti-religious, and this can result in anti-religious policies by governments as well as anti-religious attitudes and actions in society.It is inaccurate to consider secularism a monolithic ideology.Accordingly, this discussion focuses on trends present within a complex and multifaceted secular movement and should not be taken as a statement that any of these factors are true of all versions of secularism.
I identify eight interrelated arguments in the literature that in some manner posit that secularism and secular political actors are anti-religious, undermine religious freedom, or instigate discrimination against religious minorities.
First, many secular political actors see religion as irrational, violent, and dangerous and, therefore, feel it should be restricted.This type of secularism "presuppose[s] that religion is either an irrational force or a non-rational form of discourse that should be banished from the democratic public sphere" (Casanova 2009(Casanova , p. 1052)).This argument that religion is the source of violence and conflict has its roots in a European view of religion that evolved after the Thirty Years War and the Treaty of Westphalia (Casanova 2012, pp. 79-80).This creates a good and evil-like dichotomy where "religion is aligned with tradition, superstition, and supernaturalism and kindred categories, whereas secularity is aligned with modernity, rationality, and science" (Gorski and Altinordu 2008, p. 61).In fact, "religion is thought to be a regressive irrational and force and individuals would be better off if they left it behind entirely.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 2012, p. 2).Henne and Klocek (2017, p. 8) call this "the myth of religious violence".Cavenaugh (2007) claims that this type of argument is used, particularly in the West, to silence and restrict members of "certain faiths".All of this is part of a larger phenomenon where secularism is an ideology that can motivate political actions."Being secular is not just the absence of religion: they embrace secularism as a non-religious world-view.In the broadest possible sense, secularism is an idea that seeks to restrict religious elements" (Arzheimer 2022, p. 2; see also , Cesari 2021;Farr 2008;Kettell and Djupe 2020;McAnulla et al. 2018;Pabst 2012, p. 38;Stark and Finke 2000;Troy 2015).
However, it should be noted that some argue that the actual percentage of militant secular individuals is low even in the West.Branas-Garza and Solano (2010, p. 347) argue that, in the West, "the proportion of clearly religious-averse citizens is very small and never larger than 6%".Thus, this phenomenon may be driven by a small number of partisans (Buckley and Wilcox 2017, p. 5).Although, these partisans may be influential because "a purely secular view of reality has its principal social location in an elite culture" (Berger 1996(Berger /1997, p. 9, p. 9; see also Juergensmeyer 2012, p. 201;Sherkat and Ellison 1999, p. 364).
Second, the secular model of relegating religion to the private sphere is inherently restrictive of religion and gives preference to the secular over the religious.This in and of itself constitutes a form of discrimination.Philosophers such as Bader (1999) argue that the secular goal of relegating religion to private life is unfair, exclusionist, and not possible.In fact, he argues that the concept of neutrality is really a cover for a bias in favor of dominant religions or anti-religious secular ideologies.
On a more practical level, relegating religion to the private sphere decreases the political influence of religious actors, undermining their ability to protect their interests, including their ability to protest and prevent discrimination against them (Gorski 2016).Joustra (2018) similarly argues that political secularism does not work well in pluralist societies because it essentially makes second-class citizens of religious individuals or members of minority religions.That is, "by insisting that religion is a matter of individual belief, secularism undercuts ritual life and communal solidarity" (Gorski 2016, p. 53;see also, Gill 2004).This trend may increase in its restrictiveness of religion because "the privatization of religion and its exclusion from the public sphere is no longer considered by many to be sufficient" (McAnulla et al. 2018, p. 52; see also, Lerner 2013).
Third, at least in part due to the above factors, many see religious and secular political actors as competing for influence in politics and society, with each seeking to restrict the other.Calhoun et al. (2012, p. 16) argues that "so much of contemporary social conflict is linked to religion-or, rather the notion that religion and secularism are in opposition".This competition also includes a competition over the nature of morality and which morality should be dominant in society and politics (Euchner and Preidel 2018).This competition can lead to increased restrictions on religion because the secular side of this competition is "aimed at diminishing religion's potential influence on politics and society" (Sarkissian 2015, p. 52).This includes an explicit secular-based challenge to "the public role of religion through promoting secularist arguments, in particular the separation of church and state alongside the removal of perceived unfair privileges granted to religious groups . . .[and challenges to] the social customs perceived to protect religious beliefs from scrutiny and questioning both in public and private life" (McAnulla et al. 2018, p. 154; see also , Bush 2007, Bush , p. 1645;;Calhoun et al. 2012;Finke and Mataic 2021;Fisher 2016Fisher , 2020;;Fox 2015Fox , 2019c)).
Fourth, there is a growing literature focusing on anti-religious states.A central argument in this literature is that these anti-religious policies are driven explicitly by antireligious secular ideologies.For example, Philpott (2019) argues that these states' ideologies "emanate 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" (Philpott 2019, p. 77).That is, in practice, these states fully buy into the argument that religion is irrational, dangerous, and violent.These regimes tend to be authoritarian and explicitly secular with secularism declared in their constitutions.While they may allow some expression of the majority religion under state supervision, they tend to be restrictive of any religious expression outside of this supervision.They are often even more restrictive of minority religions (Fox 2020; see also, Bramsen and Vermeer 2019;Durham 1996;Finke and Mataic 2021;Froese 2004;Muller and Neundorf 2012).
Fifth, many secular political actors see religion as an explicit challenge to the power of the secular state.This is important because "when the state views religions as offering ideological or political alternatives, the state's restrictions of religious freedoms will increase" (Finke 2013, p. 300).It is often explicitly religious minorities that are seen as threatening secular institutions and ideologies (Finke and Mataic 2021, p. 4;Sarkissian 2015).
Sixth, there is considerable secular hostility to religion because religion is seen as being in opposition to human rights.Milder forms of this critique of secularism argue that liberalism may limit religion when religious practices or beliefs are seen as negatively impacting others (e.g., Ahdar and Leigh 2013).Others call human rights a secular religion (e.g., Rogobete 2014; Malachuk 2012).For example, Freeman (2004) argues that human rights are often presented as universal principles, but some aspects of human rights can clash with some aspects of major religious traditions.This can result in a situation where religious groups, most often religious minorities, are asked to give up their religious principles or cease certain religious practices in order to accommodate this universal value system considered by its advocates to be superior to religion and to supersede religion.It is for this reason that many religious critiques of international human rights characterize it as the promotion of secularism, which they often classify as a form of religion (Martin 2005, p. 834;Joustra 2018).
Seventh, some secular advocates essentially claim a secular veto over religious practices.Proponents of this veto see themselves as enlightened and superior to the irrational and primitive religious believers.While they see religious freedom as a secular value, they believe that only manifestations of religion that are in accordance with their secular ideology may benefit from this freedom.In cases where religious values and practices clash with secular morality, religion must be altered in order to conform (Sweetman 2015).That is, "secularism is not merely being defined by engagement with religion.Secularism also intellectually and politically redefines religion to suit secularist values and purposes" (Triansafyllidoum and Modood 2017, p. 53).Put differently, proponents of secular ideology demand that religions change their beliefs and practices whenever those practices are found to be in contradiction to the secular beliefs or ideology in question.In this sense, secularism can be seen as a totalitarian ideology that is no more tolerant of dissent than many manifestations of traditional religion.When given the power to enforce this program, secularism is structurally no different than a state-supported religion imposing its doctrines and practices on non-believers (Keane 2000;Fox 2020).
Finally, while this argument is inherent in the above arguments, it is important to emphasize that all of this has explicit implications for both religious freedom in general and GRD specifically.Glendon (2019, p. 6) argued that secular hostility to religion caused "waning consensus on the importance of religious freedom, plus a good deal of open hostility to religion among opinion leaders in the media, the world of entertainment, and the academy.Some legal scholars now maintain that religious freedom is an unnecessary right since everything worth protecting is covered by freedom of speech and association".That is, secularism has, in an increasing number of instances, sparked explicit opposition to religious freedom and tolerance (Keane 2000).This also resulted in the targeting of religious practices of minority groups, which are seen as abhorrent to some secular ideologies.For example, Fox (2020) documented that, in Western countries, secular-based opposition to the Jewish and Muslim practices of ritual slaughter, infant male circumcision, and female modesty have resulted in restrictions on these practices.Thus, while dominant anti-religious secular ideologies can be a threat to the religious freedom of even members of the majority religion, religious minorities are particularly vulnerable.
Yet, despite all of this, one could ask why GRD? GRD, as measured in this study, constitutes restrictions on the religious institutions or practices of religious minorities that are not placed on the majority religion.More specifically, to discriminate is to treat differently.One would expect anti-religious governments to restrict all religions, which would certainly violate religious freedom but would not constitute GRD.Fox (2020) uncovers two inter-related reasons for the link between secularism and GRD.First, majority's religious practices tend to be embedded in the culture, which makes them less likely to be considered abhorrent to secular ideologies and individuals.Minority religious practices, especially those of minorities considered non-indigenous, are more likely to be deemed in conflict with secular ideologies and, therefore, restricted.Second, there are multiple reasons why a majority group may wish to restrict minority practices that are independent of secular and religious ideologies.When one of these reasons is present, anti-religious secular ideologies and attitudes can exacerbate levels of GRD.
It is important to note that I make no claim that these are the only motivations and causes for GRD.Other motivations and causes include nationalism, defense of culture, anti-cult sentiment, Antisemitism, Islamophobia, securitization, economic development, regime, religious tradition, and religiosity (Ettensperger and Schleutker 2022;Fox 2016Fox , 2020;;Fox and Akbaba 2015;Fox and Topor 2021;Fox et al. 2021;Haykin et al. 2021;Koev 2022;Peretz and Fox 2021).

Analysis of the Influence of SRAS and Official Religion Clauses on GRD
I test the influence of these constitutional clauses on GRD using a multivariate analysis of the 176 countries included in the RAS-constitutions dataset.I use, as a dependent variable, the RAS dataset's GRD variable, which measures 36 ways a government might restrict the religious institutions or practices of minority religions. 3As the most recent data available in the RAS data is 2014, this is the year use for all variables in the analysis.The constitution variables are measured as 1 if a clause is present in a constitution and 0 if not.
Each test includes control variables that Fox (2008Fox ( , 2015Fox ( , 2016Fox ( , 2019bFox ( , 2020; Fox and Topor 2021) finds significant in predicting GRD.These include: (1) Whether the majority is Christian or Muslim.Each of these is measured separately as binary variables.Fox (2016Fox ( , 2020) ) finds that the majority of religious traditions of a country was a potential predictor of GRD.(2) The majority percent of the population because larger majorities may feel safer in engaging in GRD. 4 (3) The government's level of democracy in order to account for predictions that democracies engage in less discrimination. 5(4) Durability, which measures the number of years since the last change in regime characteristics. 6(5) The state's population and per-capita GDP. 7(6) Societal religious discrimination, which is taken from the RAS dataset and measures twenty seven types of action taken by members of society who do not represent the government against religious minorities. 8Fox (2020) and Grim and Finke (2011) find that societal animus to religious minorities can predict GRD.
The final control used in some tests in this study is state support for religion.This variable is from the RAS dataset and measures 52 way a government might in practice support religion.That is, states support religion through laws, practices, and court rulings in a manner far more diverse than the presence or absence of an official religion.Previous studies identify this as among the most important variables in predicting GRD (Fox 2008(Fox , 2016(Fox , 2019b(Fox , 2020)).This also accounts for the argument described earlier in this study that constitutions are not necessarily reflective of actual state practice, and it was intended to determine whether constitutional clauses, as opposed to day-to-day policy on supporting religion, are the best predictor of GRD.
As a general note, I run no tests using variables that were coded positive in less than ten cases because, with fewer cases coded, the results are too likely to be driven by the idiosyncrasies of the few states coded.As the RAS-Constitutions dataset includes 154 variables, I only include multivariate tests in tables if one of the following were true: (1) they involve the general official religion or SRAS variables because these are core constitutional clauses; (2) they are summary variables examining the impact of a category of variables; (3) the results for the constitution variable are at least marginally significant (sig < 0.1).
Overall, official religion and SRAS clauses have no significant influence on GRD.Table 2a,b shows that both clauses declaring a state religion and SRAS do not significantly predict GRD.Model 1 is a baseline with only controls.Models 2, 3, and 4 examine the general presence of SRAS and official religion clauses.Models 5 through 9 examine specific wording variations of these clauses.Besides clauses declaring a state religion, official religion, or established religion, no category among the official religion clauses had ten cases coded.I, therefore, combine all other categories into a single variable.There is a similar variable for SRAS wording categories that are present in less than ten constitutions.With the exception of "the state is separate/autonomous/independent from religion or religious organizations" in model 6, but not model 7 and 9, none of these variables prove significant.That is, this variable is significant but not when including all controls.

More Specific Types of SRAS Clauses
While the results show that the grand policy clauses generally declaring SRAS or an official religion do not predict GRD, there are many other types of religion-based clauses in constitutions.I argue that, in order to fully determine constitutional impact on GRD, it is also important to examine the impact of each of these clauses.For this reason, I divide these clauses into categories and examine each one individually.
The RAS-Constitutions dataset identifies 14 types of more specific SRAS clauses.These clauses do not generally declare SRAS but, rather, require a more specific type of SRAS where the government is banned from interfering in a specific aspect of religion or religious actors are banned from engaging in some form of politics or government.Table 3 shows the percentage of constitutions that have these types of clauses controlling for whether the constitution has a SRAS clause, official religion clause, or no clause.One interesting finding from this table is that while, as one would expect, constitutions that declare SRAS, on average, have more of these clauses, 18.7% of constitutions that declare an official religion have at least one of these SRAS clauses.Nearly all of these clauses found in states with official religions are intended to limit the role of religion in politics.For example, Article 75 of Costa Rica's constitution declares that "the Roman Catholic and Apostolic Religion is the religion of the State", but Article 28 states that "clergymen or secular individuals cannot make political propaganda in any way invoking religious motives or making use of religious beliefs".Article 131 states that Costa Rica's president must be "A layman or a laywoman".Articles 142 and 159 say the same for government ministers and justices.Fox (2015) argues that restricting religions' role in politics is common in states with official religions because, while these states do wish to support religion, few of them are theocracies, and these governments do not want to cede power to clergy.This is consistent with Gill's (2008) theory that governments seek to gain benefits from supporting religion but wish to minimize costs such as religious institutions and clergy usurping government power.
While I test all 14 variables that were coded as positive for at least ten cases, none of them were individually significant.As shown in Table 4, the number of these specific SRAS clauses, as well as this score interacted with the presence of a general SRAS clause are not significant.The interaction test is meant to see whether these clauses differentiate among constitutions that have general SRAS clauses.

Other Clauses Supporting Religion
As is the case with SRAS clauses, there are additional clauses supporting religion that are more specific than the more general clauses declaring an official religion.The RAS-Constitutions dataset includes three categories.First, rights and freedoms qualified by religion are rights such as freedom of expression, which can be restricted for religious reasons.For example, Article 19 of Pakistan's constitution states that "Every citizen shall have the right to freedom of speech and expression, and there shall be freedom of the press, subject to any reasonable restrictions imposed by law in the interest of the glory of Islam".Similarly, Article 15 of Sri Lanka's constitution states that many basic rights, including freedom of speech, expression, assembly, and association, "shall be subject to such restrictions as may be prescribed by law in the interests of racial and religious harmony".As shown in Table 5, these types of clauses are not particularly common and are found mostly in constitutions that declare official religions.It is reasonable to conjecture that the presence of clauses subjecting basic freedoms to religion might indicate a willingness to discriminate against religious minorities.Second, the RAS-Constitutions dataset identifies seven categories of laws that constitutions declare are subject to religious law or are otherwise associated with religion.
As shown in Table 6, the most common is the declaration that religion is a principle or the main source of legislation.All 16 constitutions that have such declarations refer to Islamic law.For example, Saudi Arabia's constitution has two such clauses.Article 1 states "God's Book and the Sunnah of His Prophet, God's prayers and peace be upon him, are its constitution".Article 8 states "Government in the Kingdom of Saudi Arabia is based on the premise of justice, consultation, and equality in accordance with the Islamic Shari'a".Article 48 states "The courts will apply the rules of the Islamic Shari'a in the cases that are brought before them, in accordance with what is indicated in the Book and the Sunnah, and statutes decreed by the Ruler which do not contradict the Book or the Sunnah".Article 55 states "the King oversees the implementation of the Islamic Shari'a".Similarly, Article 2 of Egypt's constitution states "Islam is the religion of the state and the Arabic language is its official language.Principles of Islamic law (Shari'a) are the principal source of legislation".I argue that these clauses show a higher level of commitment to a religion than a general declaration of an official religion and, therefore, may predict higher levels of GRD.Finally, as shown in Table 7, many constitutions make symbolic references to God and religion.These references, which are often in a constitution's preamble, are found even in many constitutions that declare SRAS.For example, Poland's constitution, which declares SRAS in Article 25, nevertheless in its preamble mentions the "Christian heritage of the Nation" and recognizes "our responsibility before God or our own consciences".While this type of statement likely has less of a commitment to religion than a formal declaration of an official religion, it is still worthwhile to test its impact on GRD.As shown in Table 8, when controlling for all relevant factors, none of these types of clauses predict GRD.None of the individual types of clauses proves significant (and were therefore not shown in the table).The number of freedom qualification clauses proves marginally significant in model 10 but not in model 12 when controlling for additional factors.The number of types of laws subject to religion proves significant in models 11 and 12 but not in model 13, which controls for state support for religion.Symbolic mentions of religion is also not a significant predictor of GRD.Thus, as is the case for clauses declaring an official religion, these types of clauses do not predict GRD when controlling for other factors.Thus, overall, constitutional clauses in support of both religion and SRAS do not predict GRD.This, combined with the finding that practical levels of support for religion do predict GRD, indicates that these types of constitutional clauses are likely a poor measure of a state's true commitment to a religion or secularism.

Freedom of Religion Clauses
In theory, one would expect that a clause in a constitution guaranteeing freedom of religion would impact GRD.That is, if a constitution promises religious freedom, then one would expect lower levels of restrictions on minority religious institutions and practices compared to countries with no such protection in their constitutions.However, as is shown in Table 9, the vast majority of states promise freedom of religion, sometimes using other phrases such as freedom of worship or the freedom to practice religion, in their constitutions, and multiple studies show that a large majority of states, including Western liberal democracies, restrict the religious freedom of minorities (e.g., Fox 2016Fox , 2020;;Grim and Finke 2011).Thus, such protections clearly do not guarantee an absence of GRD, but it is possible they result in less GRD.Accordingly, an analysis of the many types of religious freedom clauses and their ability to predict levels of GRD is warranted.
As presented in Tables 9 and 10, all declarations of religious freedom are not the same for two reasons.First, many constitutions, in addition to a general promise of religious freedom, also promise specific types of religious freedom.Second, many constitutions list circumstances under which religious freedom may be restricted.It is reasonable to ask whether these extra protections for religious freedom actually increase religious freedom and whether these qualifications on religious freedom decrease it in practice.While discussing all 19 such protections for specific religious freedoms listed in Table 9 and all 13 qualifications on religious freedom listed in Table 10 is beyond the scope of this discussion, I discuss some of the more common types of clauses below.Twenty-seven constitutions include the right to change one's religion.This is an important right as many states, in practice, restrict missionaries, proselytizing, and conversion (Fox 2016(Fox , 2020)).The Czech Republic is an example of a state that promises this right and observes it.Article 15 of its constitution states "The freedom of thought, conscience, and religious conviction is guaranteed.Everyone has the right to change their religion or faith or to have no religious conviction".In contrast, Armenia is among the 12 states with this type of clause that nevertheless restrict proselytizing according to the RAS data.Article 26 of its constitution states: Everyone shall have the right to freedom of thought, conscience and religion.This right includes freedom to change the religion or belief and freedom to, either alone or in community with others manifest the religion or belief, through preaching, church ceremonies and other religious rites.The exercise of this right may be restricted only by law in the interests of the public security, health, morality or the protection of rights and freedoms of others.
Thus, this article includes several of the specific religious freedoms and qualifications in those freedoms included in the RAS-Constitutions dataset.While Armenia does not ban conversion, it does discourage it by banning proselytizing.Article 8 of the Law of Freedom of Conscience and on Religious Organizations prohibits proselytizing by all religions, including the country's majority religion, the Armenian Apostolic Church.Yet, the Armenian Church has access to media channels and educational settings not granted to minority religions, and according to the Law Regarding the Relationship between the Republic of Armenia and the Holy Armenian Apostolic Church, it may "contribute to the spiritual education of society".
There are two types of clauses promising the right to religious education.Fifteen constitutions promise the right to religious education in public schools and 45 promise the right to religious education in general, but according to the RAS data, ten of these countries nevertheless restrict religious education.Portugal is among the majority that observes this guarantee.Article 41 of its constitution provides extensive protections for religious freedom, including religious education as well as for SRAS: (1) Freedom of conscience, religion, and worship shall be inviolable.( 2) No one may be persecuted, deprived of rights, or exempted from civil obligations or duties because of his convictions or religious observance.(3) No authority shall question anyone in relation to his convictions or religious observance, save in order to gather statistical data that cannot be individually identified, nor shall anyone be prejudiced in any way for refusing to answer.(4) Churches and other religious communities shall be separate from the state and free to organize themselves and to perform their own ceremonies and their worship.(5) Freedom to teach any religion within the denomination in question and to use appropriate media for the pursuit of its activities shall be guaranteed.( 6) The right to be a conscientious objector, as laid down by law, shall be guaranteed.
Sudan has a similarly detailed religious freedom clause.Article 6 of its constitution states: The State shall respect the religious rights to: (a) worship or assemble in connection with any religion or belief and to establish and maintain places for these purposes, (b) establish and maintain appropriate charitable or humanitarian institutions, (c) acquire and possess movable and immovable property and make, acquire and use the necessary articles and materials related to the rites or customs of a religion or belief, (d) write, issue and disseminate religious publications, (e) teach religion or belief in places suitable for these purposes, (f) solicit and receive voluntary financial and other contributions from individuals, private and public institutions, (g) train, appoint, elect or designate by succession appropriate religious leaders called for by the requirements and standards of any religion or belief, (h) observe days of rest, celebrate holidays and ceremonies in accor-dance with the precepts of religious beliefs, (i) communicate with individuals and communities in matters of religion and belief at national and international levels.
Yet, the government regularly violates many of these rights, including the right to religious education.While it guarantees Christian education classes to Christians in public schools, in practice, many schools claim to be unable to find teachers, and only some schools excuse Christian students from Islamic education classes. 9Apostacy laws are strictly enforced, including on Christians who are descended from converts away from Islam. 10  The government limits entry into the country by non-Muslim religious groups, allowing only a limited number of Christian groups to enter for the purposes of humanitarian activities and to promote Muslim-Christian cooperation.Although proselytizing is not explicitly prohibited by law, non-Muslim groups are monitored and pressured, through deportations and financial means, not to proselytize. 11The 1992 Prisons and Treatment of Prisoners Law states that the minister of justice can release any prisoner who memorizes the Qur'an during his prison term, encouraging non-Muslims to convert to Islam to take advantage of the parole option. 12The government generally declines to allocate land for most non-Islamic places of worship, citing that churches must meet population density parameters that do not apply to mosques. 13In addition, the government has closed or destroyed numerous churches and schools (usually citing zoning restrictions or national security).There are also numerous reports of government officials harassing Christians and preventing them from engaging in religious activities. 14 There are fifty countries with religious freedom clauses that are not qualified.However, among these states, only Cameroon, Canada, the Philippines, South Korea, and Taiwan engaged in no GRD against religious minorities in 2014, according to the RAS data.Among the countries that do qualify religious freedom in some manner in their constitutions, fifteen countries also engaged in no GRD.Many of these countries include quite a few qualifications, such as Barbados.In fact, Article 19 of its constitution provides for both many specific freedoms and qualifications: (1) Except with his own consent, no person shall be hindered in the enjoyment of his freedom of conscience and for the purpose of this section the said freedom includes freedom of thought and of religion, freedom to change his religion or belief, and freedom, either alone or in community with others, and both in public and in private, to manifest and propagate his religion or belief in worship, teaching, practice and observance.(2) Every religious community shall be entitled, at its own expense, to establish and maintain places of education and to manage any place of education which it wholly maintains.(3) No religious community shall be prevented from providing religious instruction for persons of that community in the course of any education provided by that community whether or not that community is in receipt of any government subsidy, grant or other form of financial assistance designed to meet, in whole or in part, the cost of such course of education.(4) Except with his own consent (or, if he is a person who has not attained the age of twenty-one years, the consent of his guardian), no person attending any place of education shall be enquired to receive religious instruction or to take part in or attend any religious ceremony or observance if that instruction, ceremony or observance relates to a religion which is not his own.(5) No person shall be compelled to take any oath which is contrary to his religion or belief or to take any oath in a manner which is contrary to his religion or belief.(6) Nothing contained in or done under the authority of any law shall be held to be inconsistent with or in contravention of this section to the extent that the law in question makes provision-(a) which is reasonably required-(i) in the interests of defense, public safety, public order, public morality or public health; or (ii) for the purpose of protecting the rights and freedoms of other persons, including the right to observe and practice any religion without the unsolicited of members of any other religion; or (b) with respect to standards or qualifications to be required in relation to places of education including any instruction (not being religious instruction) given at such places.(7) References in this section to a religion shall be construed as including references to a religious denomination, and cognate expressions shall be construed accordingly.
Article 37 of Bulgaria's constitution similarly both promises and qualifies religious freedom: (1) The freedom of conscience, the freedom of thought, and the choice of religion and of religious or atheistic views shall be inviolable.The State shall assist the maintenance of tolerance and respect among the believers from different denominations, and among believers and non-believers.(2) The freedom of conscience and religion shall not be practiced to the detriment of national security, public order, public health and morals, or of the rights and freedoms of others.
Unlike Barbados, Bulgaria significantly restricts minority religions, particularly Muslims and minorities the government considers cults.National security services regularly harass and arrest Muslims for engaging in religious activities.This is usually targeted against Muslims that the government considers radical. 15There have been several reported instances where police disrupted Jehovah's Witness services and checked identity cards of all present or demanded proof of local registration. 16In 2011, the European Court of Human Rights ordered Bulgaria's government to compensate members of the Unification Church after a 1997 incident where police disrupted religious services in a private home and seized religious objects. 17Jehovah's Witness and Muslim communities face systematic problems in obtaining construction permits for new prayer houses and mosques. 18 Given this anecdotal inconsistency between religious freedom clauses and government practice, it is not surprising that, as is shown in Table 11, with one important exception, there is little correlation between these clauses and GRD.The number of specific religious freedom clauses and qualification clauses did not significantly predict GRD.While the presence of a general religious freedom clause did have marginal significance in model 17, as did clauses protecting the freedom of worship, observance, or to practice religious rituals or rites in model 18, neither of these clauses remained significant when controlling for additional factors in model 20.Interestingly, clauses including the right not to profess a religion or promising freedom from religion significantly increased levels of GRD in models 19, 20, and 21.Thus, a clause included in the list of potential religious freedom clauses is associated with an increase rather than a decrease in GRD.I posit that this is likely because clauses protecting the right to not be religious are an indicator of the government's secularism in the anti-religious meaning of the term, and as argued in this study, this form of secularism is associated with restrictions on religion.I discuss this finding in more detail later in this study.
Another form of protection for religious freedoms, presented in Table 12, are bans on government interference in religion.Rather than promising religious freedom, under certain circumstances or a specific type of freedom, these clauses specifically prohibit the government from engaging in a type of activity that might impinge on religious freedom or protect a specific religious activity.These types of protections may indicate a commitment to religious freedom.Yet, anecdotally, the presence of these types of clauses also does not guarantee a lack of GRD.For example, India's constitution includes five of these types of clauses.Article 27 of its constitution bans religious taxes.Article 28 bans required religious education in public schools.Article 30 protects the right of religious minorities to create religious educational institutions.It also bans the state from discriminating in aid to such institutions.
Additionally, Article 29 protects the general rights of religious minorities, including indigenous minorities.Despite these protections, RAS scores India as engaging in high levels of GRD against religious minorities, though none of these restrictions violate any of these specific clauses.India's GRD involves restrictions on places of worship, religious services, organizations, missionary work, and the arrest and harassment of religious minorities, among other forms of GRD.
In fact, of the ten countries that are coded as having five of these types of clauses-the most coded in the RAS-constitutions dataset in 2014-only Lesotho engages in no GRD.Slovenia, Fiji, Papua New Guinea, Belize, Botswana, and Gambia engage in low levels.Greek Cyprus Malaysia, Pakistan, and, as noted, India engage in moderate to high levels.
A final form of protection measured by the RAS-constitutions dataset, presented in Table 13, does not strictly measure religious freedom but, rather, measures bans on discrimination on the basis of religion.While most religious freedom variables in the dataset measure freedoms specific to religion, this category bans multiple types of discrimination, which in theory could be imposed on any type of minority.As shown in Table 13, these clauses are common even in countries with official religions.Table 14 shows the impact of bans on government interference in religion and discrimination on the basis of religion on GRD.The number of each of these types of clauses has no significant influence on GRD.Bans on religious taxes are marginally significant in model 23 but lose significance when controlling for other factors in models 25 and 26.The only specific clause (among those that are present in at least ten constitutions) that influences GRD is bans on religious hate speech, which interestingly, increases GRD rather than decreases it.As is the case with clauses protecting the right not to profess a religion, this type of clause is particularly common in countries with SRAS clauses and may reflect a stronger commitment to an anti-religious form of secularism.I discuss this phenomenon in more detail below.

Other Types of Religion Clauses
The RAS-Constitutions dataset identifies an additional 35 types of religion clauses not included in the above categories, which are presented in Table 15.These clauses tend to be measures that increase state connections with religion.Thailand's constitution, for example, has five such clauses.Article 8 states "The King shall be enthroned in a position of revered worship and shall not be violated".Article 9 declares the King the "upholder" of Buddhism, which the RAS-Constitutions dataset codes as making him the protector or defender of Buddhism.It also declares he must be Buddhist.Article 79 recognizes Buddhism as "the religion the majority of Thai people".It also states that "the State shall patronize and protect Buddhism".While, as shown in Table 15, few of these clauses are particularly common individually, most constitutions contain at least one of them.As shown in Table 16, the number of such clauses is significant in model 28, predicting more GRD.However, when controlling for general state support for religion in model 29, this drops to marginal significance.Additionally, while a requirement for supporting religious education is significant in model 29, it loses significance when controlling for state support for religion.

A Final Model and Its Implications
GRD is an important phenomenon both normatively and empirically.Normatively, it is a violation of religious freedom and human rights.Empirically, it is an important cause of religious violence, conflict, terrorism, and other types of oppositional behavior (Fox 2002(Fox , 2004;;Basedau et al. 2017Basedau et al. , 2023)).
Very few of the constitutional religion clauses prove to be strong predictors of GRD.In model 32 in Table 16, I present the final model, which includes all categories of clauses that remain significant when controlling for other factors.Specific clauses protecting the right not to profess a religion and banning religious hate speech both increase levels of GRD, as does the number of "other" religion clauses in a constitution.This has several implications.
Most importantly, for the most part, constitutions are poor predictors of GRD.The factors that most consistently predict GRD are autocracy, practical levels of state support for religion, and societal discrimination (SRD) against religious minorities, which all predict higher levels of GRD.These results are not unexpected, as they are similar to previous studies (Fox 2016(Fox , 2020;;Grim and Finke 2011).
This lack of correlation between most constitutional religion clauses and GRD combined with the fact that other religion variables do predict GRD implies that these clauses are a poor means to measure a country's true policy and attitude toward religion.The clauses we most expected to predict GRD do not.Constitutional support for an official religion does not predict GRD when controlling for other factors, though one of those factors is practical levels of state support for religion, but even when dropping that control from the test, constitutional official religion clauses predicted GRD with, at best, marginal significance.
In contrast, actual levels of state support for religion consistently, significantly, and strongly predict GRD.While the official religion clauses and practical support are correlated 19 , the practical levels of support for religion are the key measure of a state's relationship with religion that predicts GRD.Thus, it is government support for religion as reflected in laws, government practices, and court rulings that influences GRD.Official religion clauses in constitutions influence GRD only to the extent that they reflect a practical commitment of a state religion, and the evidence presented here shows that these clauses are at best weak indications of commitment.
The religion clauses that support religion that do predict higher levels of GRD are "other" religious clauses.More specifically, these clauses can be described as various clauses that tend to indicate a connection between religion and state in some specific area, such as government involvement in appointing religious officials or requirements that the government provide religious education.This is likely the case because these clauses require the government to support or engage with the majority religion in a more practical manner.In contrast, declaring that the state has an official religion can mean many things, ranging from a symbolic connection with no practical implications to a state governed by religious law.Thus, these "other" clauses reflect or perhaps even require a more practical commitment to religion than more general declarations of an official religion.
Religious freedom clauses do not even have an indirect influence on GRD.The utter failure of these clauses-including qualifications to these clauses and protection for more specific religious freedoms-to predict GRD demonstrates that these clauses are rhetoric common to constitutions but are poor reflections of intent.That is, they are general principles which may or may not be applied in practice.
The clauses that do influence GRD are not found among the more central declarations of a state's relationship to religion and its commitment to religious freedom.Rather, they focus on more specific religious issues that few would consider a priori as the most important indicators of religious freedom in constitutions.
Clauses protecting the right not to profess a religion and banning religious hate speech both predict significantly higher levels of GRD.I posit that this is because these clauses better reflect an anti-religious form of secularism than do constitutional declarations that a state is secular.Banning religious hate speech on the surface seems like a reasonable option.Yet, consider the second commandment "Thou shalt have no other Gods before me" among many bans in the Old Testament on worship of other Gods and idolatry.These biblical texts could be construed as hate for other religions.Similarly, recitations of texts from the Koran that are critical of other religions could be considered hate speech.For example, when discussing Jews, the Koran states "We cursed them and made their hearts hard; they altered the words from their places and they neglected a portion of what they were reminded of; and you shall always discover treachery in them excepting a few of them".On a more basic level, most religions, and certainly most monotheistic religions, claim to express a universal truth that implies that those who do not believe in these truths, at the very least, believe incorrectly.In many cases, condemnations of non-believers are harsher.Yet, any expression of this basic and core religious belief could be construed as hate speech.
In addition, any law that restricts speech has illiberal implications.Free speech is a cornerstone of liberal philosophy and government.Without it, true freedom cannot exist.While some limitations are inevitable, such as the notorious ban on shouting "fire" in a crowded theater, liberalism demands that restrictions on free speech be few and implemented only for vital causes.As noted above in my discussion of qualifications on religious freedom, the more reasons a freedom can be limited, the easier it is for governments to do so.Constitutional clauses banning hate speech tend to be vague and open to interpretation.For example, Article 44 of Bulgaria's constitution states that organizations and activity in Bulgaria may not be contrary to a list of principles including "religious enmity".Similarly, Article 39 of Croatia's constitution states "Any call for or incitement to war or use of violence, to national, racial or religious hatred, or any form of intolerance is prohibited and punishable by law".These ambiguous laws allow for broad interpretations.
For example, based on an anti-hate speech law (rather than a constitutional clause), prosecutors in the Netherlands filed a criminal charge of hate speech against the rector of Rotterdam Islamic University for criticizing non-believers and Alawites.In 2014, the courts ruled that the prosecutor's interpretation was incorrect because criticizing a religious ideology (as distinct from its followers) does not necessarily constitute a criminal offense. 20 On its surface, the right not to profess a religion seems neutral.Yet, it can also be seen as reflecting a desire to protect people who are not religious or are secular from religion.This could reflect a fear or distrust of religion.That this protection is present largely in states that maintain SRAS indicates that this fear is largely coming from a non-religious place.In addition, of the nineteen countries that have such a clause, two, Cuba and Laos, are Communist and an additional eight, Azerbaijan, Belarus, Bulgaria, the Czech Republic, Kyrgyzstan, Lithuania, Ukraine, and Uzbekistan, are former Communist states where the anti-religious aspects of Communism ideology are still influential.
This has some significant implications for measuring secularism.Secularism is notoriously difficult to measure (Fox 2019a).For example, Philpott (2009) lists nine different uses of the terms "secular", "secularism", and "secularization".The only commonality among them is that they all define something that is other than religion, the decline of religion, or anti-religious.Thus, he makes two arguments: that there is no agreement on the definition of secularism, and that whatever its definition, secularism is intrinsically linked to religion.However, if we want to focus on secularism as a political ideology, then we need to differentiate between ideologies that simply want the state to be separate from religion and those that are anti-religious.For example, Kuru (2006Kuru ( , 2009) ) distinguishes between two types of political secularism: "Passive secularism, which requires that the secular state play a 'passive' role in avoiding the establishment of any religions, allows for the public visibility of religion.Assertive secularism, by contrast, means that the state excludes religion from the public sphere and plays an 'assertive' role as the agent of a social engineering project that confines religion to the private domain" (Kuru 2006, p. 571).
This study indicates that, at least for constitutional measures, a desire to simply keep religion and state separate is not useful for understanding the causes of GRD.However, anti-religious forms of secularism seem to increase levels of GRD.Fox et al. (2022) makes this argument in the context of anti-religious ideologies that lead to restrictions on religious practices deemed contrary to secular beliefs in animal rights (restrictions on Halal and Kosher slaughter), women's rights (restrictions on head coverings and other modest dress by women), and an infant's right to bodily integrity (restrictions on male infant circumcision).Yet, measures for this anti-religious attitude are scarce.This study's findings that protections for the right to not profess a religion and bans on religious hate speech predict GRD indicate that these may provide a good starting point to develop practical measures of this anti-religious element of secularism.A crosstabulation between state support for religion and whether the state has an official religion produced a significance (Chi-Squared) of 0.001.As the constitutional variable is binary, I also used logistic regression with state support for religion predicting an official religion.The two were correlated but not identical.The significance was < 0.001.The Cox and Snell adjusted r-squared was 0.298, and state support for religion correctly predicted whether the constitution had an official religion clause 88.6% of the time.20 (7 March 2014) Islamic college chief did not incite hatred, public prosecutor says, https://www.dutchnews.nl/news/2014/03/islamic_college_chief_did_not/ Accessed on 30 November 2022. 19

Table 1 .
Official Religion and SRAS Clauses with Specific Wording in 2014.

Table 2 .
a: OLS regressions predicting GRD in 2014 examining the impact of official religion and SRAS constitutional clauses; b: the influence of specific SRAS clauses on GRD in 2014.

Table 4 .
OLS regressions predicting the influence of specific SRAS clauses on GRD in 2014.

Table 5 .
Rights and freedoms qualified by religion in 2014.

Table 6 .
Types of laws that are subject to religious law or associated with religion in 2014.

Table 7 .
Symbolic references to religion.

Table 8 .
OLS regressions predicting the influence of freedoms and laws subject to religion and symbolic mentions of religion on GRD.

Table 10 .
Qualifications on religious freedom (among countries with religious freedom clauses).

Table 11 .
OLS regressions predicting the influence of religious freedom clauses on GRD in 2014.

Table 12 .
Specific bans on government interference with religion in 2014.

Table 13 .
Bans on discrimination on the basis of religion in 2014.

Table 14 .
OLS regressions predicting the influence of bans on government interference in religion and discrimination on the basis of religion on GRD in 2014.

Table 16 .
OLS regressions predicting the influence of other types of religion clauses on GRD and final model in 2014.
Boychev et al. v. Bulgaria European Court of Human Rights (27 January 2011) Religion and Law Consortium.