May the lord protect our country: ethnic relations as a moderator between religiosity and radical right vote

ABSTRACT While we know a lot about the typical type of radical right-wing (RRWP) voter, individual religiosity in explaining support for RRWPs has eluded consistent scholarly attention. The mixed results available from the scarce literature find both positive and negative associations between religiosity and RRWPs vote. The variation in these relationships is puzzling, especially if we consider how RRWPs often present themselves as guardians of native ethnic and religious identity. In this paper we argue that religiosity increases the chance of voting for RRWPs when ethnic relations are a salient issue in the political system. We test our theory using multilevel regression modeling on the European Social Survey, specifically Rounds 7, 8 and 9 and replicate our results based on the European Values Study from 2017. We find that religiosity is a significant predictor of the RRWP vote when there are salient ethnic relations in the political system, proxied by the presence of an ethnic minority party. On the other hand, in countries without minority parties, non-religious individuals are more likely to vote for RRWPs.


Introduction
For God and Country could easily describe the majority of radical right-wing politics (RRWPs), yet despite the apparent and prominent role of religion, voter religiosity in explaining support for RRWPs has eluded consistent scholarly attention. While we know a lot about the typical type of RRWP voter, his (Givens 2004) education (Betz 1994), age (Kitschelt and McGann 1997), attitudes towards immigrants (Van Der Brug, Fennema, and Tillie 2000) and have clear data on various other topics, the relationship between religiosity and RRWP support has not been fully investigated so far. The scarce research on the topic has produced confusing evidence in this regard. On the one hand, it has been argued that religiosity leads to a decrease in the likelihood of voting for the radical right (Lubbers and Scheepers 2000). The main reason for this lies in the assumption that 'religious people are integrated in religious communities that are likely to vote for Christian party 1 ; this makes them more likely to vote for traditional Christian parties, rather than RRPs' (Immerzeel, Jaspers, and Lubbers 2013, 946).
On the other hand, one of the key items on the agenda of radical right-wing parties is to 'warn of European civilization's destruction at the hands of non-Christian elites' (Montgomery and Winter 2015, 380). This can make people feel religiously threatened by others (i.e. non-native citizens of the country), which has the potential to contribute to them voting for an RRWP (Raiya et al. 2008). In the seminal work on the topic, Arzheimer and Carter found some evidence to support a positive relationship between the two elements in a sample of Western European countries and argued that religious voters would become increasingly available to radical right appeals as time passes (Arzheimer 2009). A decade later, Marcinkiewicz and Dassonneville (2021) took another look at this question and found that in some countries there is clear evidence, of a positive relationship between religiosity and voting for RRWPs, while in other countries there is none.
It seems that there is confusion on the exact nature of the relationship and its prominence. The variation in these relationships is puzzling, especially if we consider the fact that RRWPs often present themselves as the guardians of ethnic (Koev 2015) and religious identity 2 (Froio 2018;Montgomery and Winter 2015) of the nativist population. In other words, if an RRWP calls for the protection of ethnic and especially religious identity, it is unclear how individual-level religiosity can be associated with either support for or opposition to those parties.
Since the most recent work on the topic has shown that both negative and positive associations are empirically observed (Marcinkiewicz and Dassonneville 2021), we aim to extend our knowledge of this relationship and try to answer the question -What role do ethnic relations play as a moderator between religiosity and radical right vote? Are ethnic relations the factor that foster the activation of religiosity and result in support by religious voters for RRWP ? We argue that religiosity increases the chance of voting for RRWPs in countries with salient ethnic relations, where ethnic relations are one of the 'live' issues of political contestation. We argue in this paper that the presence of ethnic minority parties' 3 proxies for a political context in which ethnic relations are politicized. Building on that assumption, a recent study by Koev (2015) provides insight as to when that might be the case. By looking at the aggregate levels of electoral success of both minority and radical right parties in CEE Europe, Koev (2015) argues that the electoral success and/or the presence of minority parties in the system or in government, increases the fear associated with cultural purity and the threat to ethno-nationalist identity, leading to higher levels of RRWP electoral support. We expand on this argument and claim that under such conditions, religious individuals are the ones who will be more likely to fear for the preservation of their cultural identity and to opt for RRWP.
Apart from in Koev's study, minority parties have been unfairly neglected as a driver of RRWP success and as a contextual moderating factor that shapes individual-level relationships. The latter part of this gap is what we address and is the main contribution of the study. We demonstrate how and why the salience of ethnic relations proxied by minority parties accounts for religious voters' support for RRWPs. The first value of this approach should be in providing an insight into when we can expect religiosity to be related to support for RRWPs. Furthermore, as the presence of minority parties reflects the fact that ethnic relations are salient in a given political system, it is crucial to understand the profile of voters that might respond to such circumstances by opting for RRWPs.
Our assumptions are tested using multilevel regression modeling based on the European Social Survey Rounds 7, 8 and 9 and replicated on the European Values Study of 2017 (EVS, 2020). We find significant evidence to support our claim across multiple estimation procedures and model specifications. In particular, we find that religiosity is not a significant predictor of RRWP vote and can even be a detrimental factor in countries with no minority parties. When minority parties are present, indicating the salience of ethnic relations, religiosity is positively related to voting for RRWPs. Additionally, this pattern persists when ethnic relations are proxied by levels of ethnic fractionalization. Here, religious individuals tend to support RRWPs disproportionally more in countries with medium levels of fractionalization, compared to countries with low or high levels of ethnic fractionalization.
The article proceeds as follows. First, we conceptualize RRWPs and explain the dominant ideological dimensions. Then we outline how these dimensions are relevant to disentangling the story of the literature of RRWPs and religiosity. Furthermore, we outline the various mechanisms between religion and RRWPs and point out how and why ethnic issues, operationalized through the presence of ethnic minority parties should moderate that relationship. The subsequent sections describe our data, methods, and the results of the analysis. Finally, we put our findings into a broader perspective.

What are RRW parties and who votes for them?
While our goal is to provide evidence for the contextual factor that accounts for religiosity as a predictor of RRWP vote, we first need to identify the ideological components that constitute such parties. As such, in this section we present previous findings on the typical characteristics of RRWP voters, which help us contextualize our findings and place them within the broader scope of research on the radical right in Europe.
Previous literature has suggested that radical right-wing organizations are borderline cases between movements and parties (Gunther and Diamond 2003), sometimes perceived by the general public as Nazis and associated with those types of ideas (Rydgren 2005). These parties are at their core ethno-nationalist and xenophobic, complemented by anti-political-establishment populism (Rydgren 2005). In Western Europe, their central feature is their strong anti-immigrant platform (Arzheimer 2009;Lubbers and Scheepers 2000;Van Der Brug, Fennema, and Tillie 2000). This created the crucial distinctiveness of what is a diverse party family and can be organized along three dimensions -nativism (promoting the economic and cultural interests of native inhabitants), authoritarianism (severe punishment for norm violators and the infringement of authority) and populism (the clash between the people and a corrupt elite) (Mudde 2007).
Regarding voting for this diverse party group, a large body of research has produced consistent profiles across several demographic and attitudinal characteristics. The typical RRWP voter is usually male (Givens 2004), young (Kitschelt and McGann 1997), with a lower or middle level of education (Betz 1994). Additionally, these people usually hold negative attitudes towards immigrants (Van Der Brug, Fennema, and Tillie 2000), show higher levels of political distrust and euroskepticism (Werts, Scheepers, and Lubbers 2013), and are more likely to show general dissatisfaction with representative democracy (Rico, Guinjoan, and Anduiza 2017) accompanied by a perception of the decline of society (Elchardus and Spruyt 2016). In terms of the political and societal context at the country level, previous research suggested that immigration and the unemployment rate, among others, are the most important contextual factors in explaining the rise of the extreme right vote (Arzheimer 2009).
Taking these findings into account, how does religiosity fit in to the story of the RRWP vote? Previous findings on the nature of this relationship have been inconclusive, with some finding a positive relationship between religiosity and RRWP vote, while others found no relationship whatsoever. We assume this discrepancy arises from three separate issues: (a) the broad definition of the radical right party family; (b) the diverse conceptualizations of religiosity resulting in various types of religious sentiments as vote predictors; and (c) the lack of salient country-level factors that account for diverging relationship between religiosity and the RRWP vote. While the third factor is the focus of this article, we will also briefly describe the first two.
First, the literature does not clearly differentiate between the various elements of right-wing ideology when classifying political parties as RRWPs. Instead, the dominant conclusion is that these three dimensions are enough of a unifying thread to bind these diverse parties together (Koev 2015). We argue that there is both a theoretical and empirical confusion between RRWPs and right-wing populist parties. We base this observation on theoretical and empirical arguments.
Theoretically, each dimension of the RRWP ideology carries (or not) a specific appeal to religious voters. As Montgomery and Winter argued (Montgomery and Winter 2015), 'Church attending Christians, for instance, might be less likely to hold populist (elite challenging) values but more likely to hold authoritarian (strong law and order) preferences'. It seems that a clear link between religiosity and nativism and authoritarianism can be drawn, while we do not see such a clear line of relationship with populism. Furthermore, treating populism as a sufficient condition for classifying a party as radical right-wing, might lead to empirical inconsistencies when considering the link between religiosity and RRWPs.
We expect that strong religious sentiment and church-going provides deep community relations and that consequently religious voters will genuinely care about the interests of their (cultural/ nativist) group. This makes them much more responsive to nativist (identity-based) messages. Under such conditions, the perceived norm violating threat posed by migrants will be viewed as much more important than the perceived threat from the alienated political elite in the country. Lucassen and Lubbers (2012), suggests that concerns for cultural identity are the basis of support for radical parties. For this reason, we argue that the appeal of radical-right parties for religious voters comes for their nativist and authoritarian rather than their populist dimension. Therefore, when exploring the role of religion, RRWPs must have a nativist (native born) ideological outlook.
While this distinction might feel redundant, since nativism is an ideological dimension of all RRWPs, we find it necessary to point it out. We argue that inconsistencies regarding the relationship between RRWP vote and religiosity arise when populism alone is considered a sufficient condition to label a party 'radical right'. Following this view, studies such as the one by Montgomery and Winter (2015) that include a variety of distinct parties in their classification of the radical right and which conflate RRWPs and right-wing populist parties, not surprisingly report a range of positive, negative and neutral effects of religion on the propensity to vote for such parties. To illustrate, the study reports a negative relationship between religiosity and voting for Public Affairs (Věci veřejné) in Czech Republic, a party that is classified by the authors as radical right-wing but which was antiestablishment populist without a clear anti-immigration or nativist ideology. Similarly, Guth and Nelsen (2019) find that right-wing populists attract individuals detached from religious institutions.
On the other hand, studies that limit their classification of the radical right to anti-immigrant politics (ethnocentrism or nativism respectively) find positive effects of religion on RRWP vote (Van Der Brug & Fennema, 2003;Van Der Brug, Fennema, and Tillie 2000;Molle 2019). Similarly, parties that dominantly monopolize nativist issues (such as Fidesz in Hungary, and Law and Justice in Poland) manage to capitalize on the religiosity of the electorate (see Montgomery and Winter 2015). This observation gives merit to the claim that it is the nativist message that appeals to religious voters. Following this statement, we classify parties as RRWP only if a clear nativist ideology is a significant part of their political program.
Second, the confusion on the issue may arise from conflating the relationship between various types of religious sentiments and how each dimension of religiousness can relate to support for RRWPs (Arzheimer and Carter 2009;Esmer and Pettersson 2007). We refer to two particular types of religiousness: religious practices and religious beliefs. While not many studies analyzing religiosity and voting make this distinction, some have argued that these dimensions could have a different effect on RRWP vote (Immerzeel, Jaspers, and Lubbers 2013). Despite this observation, Montgomery and Winter (2015) argued recently that all dimensions of religiosity need to be accounted for when estimating its effect on RRWP vote. Simple conceptualizations of religiosity focused on one dimension not only omit parts of religiosity, but may also underestimate the total effect religiosity might have on RRWP vote (Montgomery and Winter 2015). For this reason, we argue that only focusing on all aspects of religiosity, beliefs and practices alike can portray an accurate picture of the voting tendency for RRWPs.
Third, and most importantly for this research endeavor, the extant literature fails to distinguish between salient county-level factors that account for the diverging effects of religiosity. Instead, authors usually look at the strategies of specific RRWPs in attracting religious voters. However, party strategies tell us only about the core party ideology, and not about when or indeed whether the ideology is salient and important to religious voters.
In this research, we focus on examining the favorable contextual conditions that would direct voters towards these parties (see Arzheimer and Carter 2009). The most recent research findings in the field show us that context matters. Marcinkiewicz and Dassonneville (2021), revisited Arzheimer and Carter (2009), and found that religion plays a role as a predictor of RRWP vote in some European countries, while in others it does not. We try to build on their paper and to provide an explanation for this difference across countries. Relying on Koev's work (Koev 2015) we posit that the salience of ethnic relations in the country activates religiosity as a predictor of RRWP vote. The following section elaborates on this idea.

Support for RRWP parties -religiosity in a specific political context
Previous research suggests that typical RRWP voters are more alienated people, individuals who are living in atomized societies, without meaningful connections with others (Rydgren 2008). RRWPs usually appeal to these voters through nationalist or populist agenda, promoting the ideas of 'us versus them' (others, such as migrants and/or the elite, whether real or imagined). This way, they provide a sense of security through belonging to a broader group for those that are socially alienated (Fontana, Sidler, and Hardmeier 2006). For most Western European countries which have experienced the rise of RRWP popularity them were portrayed as immigrants. Furthermore, Ivarsflaten (2008) found that no RRWP has managed to mobilize support in West without mobilizing the antiimmigrant vote. Additionally, as Arzheimer andCarter (2009) previously, andMarcinkiewicz andDassonneville (2021) recently showed, established Christian-Democratic parties in the West still manage to mobilize almost all religious voters.
On the other hand, in Central Eastern Europe (CEE), these Christian-Democratic parties have not taken root for a variety of reasons. Thus, religious voters are more prone to giving their support to various different political options. This leaves room for RRWPs to try and mobilize them. The problem which occurred is that the CEE region has not faced that much pressure in terms of cultural heterogenization during the immigration crisis in Europe. That did not stop RRWPs, and it is our assumption that they used ethnic differences within CEE countries to mobilize voters, and to present ethnic minority groups as them, the ones which will try to become newly empowered and to 'use their influence to improve their standing at the expense of the majority' (Koev 2015, 651).
With us vs. them mobilization strategies used in both Western and Eastern Europe, the question is thus why religious voters are, or appear to be, most susceptible to the RRWPs pleas. Two comparative studies by Van Der Brug, Fennema, and Tillie (2000) and van der Brug and Fennema (2003) show that church attendance was predictive of voting for anti-immigrant parties as well. It is quite possible that because church-attending individuals are deeply integrated into their culturally like-minded communities, they feel much more threatened by outgroup members and cultural diversity. RRWPs tend to capitalize on these fears. Furthermore, previous research has indicated a clear relationship between religiosity and ethnocentrism, or religiosity, and ethnic prejudice (Altemeyer 2003). Scheepers et al. (2002) found that religious service attendance and religious particularism increases prejudice towards ethnic minorities. This is not without contention as some scholars are showing the opposite effect, that religiosity can 'immunize' voters against prejudice, and thus against voting for RRWP (Siegers and Jedinger 2021). However, the development of ethnocentric attitudes has been linked to religious service attendance (Adorno et al. 1950) and religious beliefs (see Arzheimer and Carter 2009). For the latter, church-attending individuals are more likely to hold authoritarian attitudes (Altemeyer and Hunsberger 1992) and consequently respect and conform to religious authority. Respect for religious authority has been linked with adopting conventional norms and values related to the non-critical acceptance of discriminatory beliefs about outgroup members (Eisinga, Lammers, and Peters 1990). It is the conventionalism of religious voters that makes RRWP nativism appealing and electorally viable.
From a slightly different approach that focuses on beliefs not religious attendance, an overview of the literature suggests that religious beliefs can have a more direct relationship than attendance on voting for RRWPs. Religious extremists are almost natural voters for radical right parties (Camus 2013). However, these individuals cannot be found at religious services, because they have a more extreme position regarding theological teachings (Raiya et al. 2008). The more orthodox people are, literature shows, the more likely they are to feel threats from other religions. In the European context, orthodox religious believers tend to find the presence of Muslims dangerous for their group, neighborhood, city or whole society/country (Immerzeel, Jaspers, and Lubbers 2013). Furthermore, religious individuals are more likely to develop a 'closed-belief system' and dogmatic beliefs (Arzheimer and Carter 2009), often closely connected with ethnocentrism and authoritarianism. This type of 'closed belief system' contributes to developing prejudice towards outgroup individuals and can function as the basis for the justification of nativist beliefs.
From the surveyed literature above, we might simply conclude that religious voters are an easy target for RRWP mobilization efforts. However, that is not always the case. As previously mentioned, the relationship is not entirely straightforward, and we must account for contextual factors as well. Thus, here we examine the political context -arguing that the salience of ethnic relations in society, which makes cultural treats salient, ensures a positive relationship between religiosity and RRWP vote.
We build on Koev's (2015) work and operationalize the salience of ethnic relations through the existence of ethnic minority parties in the system. Koev argues that 'when ethnic minority demands for equal rights, have themselves mobilized in the formation of an ethnic minority parties' and further adds that 'successful ethnic minority parties heighten the salience of ethnic-nationalist divisions within a state, creating electoral demand for parties of the populist right'. (Koev 2015, 649).
There are several reasons why we would expect any minority party to have an effect on the vote share of RRWPs, and specifically on the role of religiosity as the driver of that support. First, ethnic relations are more salient as it is much easier to mobilize against out-group members if they have political representation -that is, a minority party in the political system. Once ethnic minorities are politically organized, the party becomes perceived as a legitimate threat to national virtues (Koev 2015, 651). The more serious the cultural 'threat' from others becomes, the more likely it is that 'nativist' groups will counter-mobilize against the specific threat. Small parties, especially after an electoral success, tend to make particular topics very salient in public discourse (Abou-Chadi 2016; Wagner and Meyer 2017). Building on these studies, we assume that minority parties increase the salience of ethnic relations, stressing their cultural difference. Increasing the salience of nativist issues offers favorable conditions for religiosity to be linked to RRWP voting. When threatened by them and their different cultural standards, the RRWP message about the preservation of cultural unity should find fertile ground among religious individuals. In a way, minority parties provide RRWP parties with significant mobilization potential against them, especially among cultural purists that perceive the minority parties as a threat to national identity.
We argue that religious voters are the primary group to be mobilized in these conditions. Previous research has shown that Christian affiliation and Church attendance are associated with delineating and constituting what it means to be ethnically British (Storm 2011). Similar ideas are found in the strategies of Law and Justice in Poland, Jobbik in Hungary, Attaka in Bulgaria, and the Slovak National Party in Slovakia, where being Catholic (or Orthodox, as appropriate) is a prerequisite for being truly Polish, Hungarian, Bulgarian or Slovak, respectively. Where minority parties represent a particular minority interest, RRWPs and their emphasis on (nativist) religious purity become salient and resonate with religious voters. Note that it is not any one specific action that these outgroups perform that drives this, but rather the mere fact that they are culturally different and politically present, thus contributing to the ethnicization of politics and increasing the perceived threat to the nativist identity, which seems to push religious people towards RRWPs. In turn, religious individuals opt for RRWPs in order to protect what they see as the dominant culture against the perceived lurking threats of minorities.
Therefore, taking everything into consideration, we expect: H1 -Religiosity is positively related to the RRWP vote when ethnic relations are salient (i.e. a minority party is present) in the political system.

Data and methods
We use the European Social Survey Rounds 7 (2014), 8 (2016) and 9 (2018) with a reference point to the most recent parliamentary elections held for questions on electoral behavior. The final sample used in the analysis totals 118,019 observations across 26 countries that have at least one RRWP in the system. For the dependent variable, we used respondents' vote choice in the last election and coded RRWP choices across countries as 1, and all other party responses as 0. This amounted to a total of 13,323 respondents classified as RRWP voters. The list of parties considered to fit the profile of RRWP described in the theoretical chapter can be found in Online Appendix. Our strategy here was to look at party leaders and their dominant political messages at the time of the election, the party manifestos, and the Party Manifesto coding scheme (using code 70 for nationalist parties) to help resolve certain borderline cases (Volkens et al. 2020).
The main independent variable was created from three questions about religious beliefs and religious practices. The frequency of religious service attendance apart from weddings and funerals were coded from 1 (Every day) to 7 (Never). In terms of religious beliefs, the question asked how religious an individual was on an 11-point scale, ranging from 0 (Not at all) to 10 (Very religious) and how often an individual prayed, scaled as 1 (Pray every day) to 7 (Never pray). Based on the argument from Montgomery and Winter (2015) that stresses the importance of studying the uniform effect of religiosity on political behavior, we summed the responses to the three variables and divided the result by 3 to create an additive index of Religiosity ranging from 1 to 8, where higher numbers represent higher levels of religiosity. To retain to the maximum possible extent the logic of a uniform measurement, we decided to use original scaling in the creation of the index, rather than standardized scores, as standardization would give much higher weight to the two behavioral measures as opposed to one that measures religious beliefs.
Additional variables included at the individual level were demographics -age (years of age grand mean-centered to avoid convergence issues), the dummy variable male 1 and female 0, education measured in completed years of education, and income measured on a 10-point scale as income percentile distribution in the sample. Substantive controls related to the literature findings we outlined above -interest in politics, measured on a 4 -point scale from 1 (Not interested at all) to 4 (Very interested); and satisfaction with democracy, measured on an 11-point scale, with higher numbers indicating higher levels of satisfaction.
From one country to another, we see that radical right parties are averaging around 10% of vote share across Europe nowadays (Stockemer 2016), with an increasing trend in terms of electoral viability. Bohman and Hjerm test the assumed effect the electoral success of RRWP parties has on anti-immigrant attitudes over time, specifically -(a) that the peoples' attitudes towards migrants have become more negative, (b) that migrants' ethnicity plays a role and (c) that all of this has led to increasing polarization in European societies, finding no evidence in support of either (Bohman and Hjerm 2016). However, as we pointed out in the theoretical chapter, some have linked religious service attendance with prejudicial attitudes (Scheepers et al. 2002), while others show that religiosity can 'immunize' individuals against prejudicial attitudes and consequently against RRWP (Siegers and Jedinger 2021). To account for the possibility of a relationship between prejudice and RRWP vote, we included an item asking respondents how many migrants of different race or ethnic ancestry should be allowed to come and live in the country. The range was from 1 (Allow none) to 4 (Allow many to come and live here). Furthermore, we control for general ideological preferences measure on an 11point left-right scale, with 0 being left, and 10 being right.
For minority parties, we looked at whether parties that can be classified as such had an electoral presence in previous elections. Countries satisfying that condition were coded 1, with others coded 0. To classify minority parties, we used the Party Manifesto project as the first point of reference (using code 90 for Ethnic or Regional Parties in the Manifesto project), complemented by Koev's (2015) classification of minority parties. The list of countries that have a minority party is in Online Appendix.
We included several contextual variables as controls. We used GDP per capita from the World Bank; net migration data from Eurobarometer for the year of the election expressed in tens of thousands: the ethnic and religious fractionalization index from ESS multilevel dataset; dummy variable for the presence of Christian Democratic parties in the system based on Manifesto Project database; and to account for serial autocorrelation and time, two dummy variables were introduced, ESS Round 8 and ESS Round 9, with the reference category being ESS Round 7.
Finally, while anti-immigrant attitudes play a crucial role in a radical right vote in Western Europe, this topic has been present but not that salient in the agenda of their ideological comrades in the East. For these parties, the clash with ethnic minorities has been more significant. To account for these differences, we include a control for Western vs. Eastern Europe when evaluating the role of ethnic parties in support for RRWPs.

Results
Our results confirm the initial expectations from the theoretical chapter. We argued that religiosity as an RRWP predictor is activated by the salience of minority relations proxied by minority rights. For this reason, we fitted several multilevel logistic regression models through which we were able to account for country differences as well as individual-level characteristics. Our initial model resulted in a 0.312 intraclass correlation coefficient, meaning that around 31% of our variance in voting for RRWP is attributed to country-level variation. The next step in the analysis checked whether we should expect this variation in the success of RRWP parties across countries to be reflected in a different relationship between religiosity and vote choice. Figure 1 present this relationship across countries, clearly demonstrating that this relationship is context dependent. With the visual inspection confirming that we should expect diverging effects across countries, we now turn to formal hypothesis testing.
Moving on to the main argument of the paper, we hypothesized that the salience of ethnic relations would turn religious voters towards RRWPs (Table 1, Model 1). Here, the effect of religiosity on voting for RRWPs is negative and significant (−0.090***), indicating that more religious people are less likely to vote for RRWP parties. However, since the separate effect of religiosity is conditioned on the interaction, the direction of the relationship is negative under the condition that no minority party is present in the system. In situations where a minority party exists, the relationship moves in a different direction, meaning that more religious people are in fact more likely to vote for RRWPs. The logistic function reveals that the probability of casting a vote for an RRWP rises from around 30% for non-religious individuals to around 50% for religious individuals if a minority party is present in the system (Figure 2).
To conduct robustness checks, we used several approaches and model specifications. First, we limited the pool of respondents to only those who belong to a Christian denomination. Several prominent studies including Arzheimer and Carter (2009) and Montgomery and Winter (2015) have suggested such an approach. In our subsample of Christian respondents, we replicated the results obtained earlier, with a similar effect size (see Table 1, Model 2).
Second, some authors have suggested taking into consideration the unique relationship between religion and politics in Poland (Montgomery and Winter 2015). To account for the specificities of this case and test whether it might be driving or skewing the results of the analysis, in one of the models we omitted Polish respondents from the analysis ( Table 1, Model 3). The results were replicated, although the differences were not as sharply identifiable as before, with much greater uncertainty associated with religious individuals in countries with minority parties.    34,198.150 18,854.810 32,253.540 Note:*p < 0.05; **p < 0.01; ***p < 0.001 Third, to account for downward biased standard errors (Heisig and Schaeffer 2019) we tested a model with random slopes across countries. The random effects of religiosity did not reach statistical significance; however, the fixed effects part of the model still produced significant results, both for individual predictors (religiosity and minority party presence) and for the interaction effect (in Online Appendix). The results indicate that specific country level characteristics have to be taken into consideration so that the role of religiosity can be identified.
In the last stage of our robustness check, we replicated the model in as much detail as possible based on the European Values Study 2017 dataset consisting of 25 European countries with RRWPs. A few differences in the modeling procedure deserve clarification. Our dependent variable here was slightly different and is more of an attitudinal than behavioral measurement. To be clear, the EVS asked what party appeals to you the most, instead of asking about vote choice in the previous elections. We coded all those who choose RRWP parties as (1) and the remainder as (0). Second, the EVS used identical questions and scales for issues of religious attendance and prayers but implemented a nominal scale measuring religious beliefs essentially capturing whether or not a person is religious. Instead of this question, we used a 10-point item that measures how important God is in an individual's private life. Using these three questions we created an additive index of religiosity rescaled to a range from 1 to 8. Finally, we used a set of control variables (gender, age, education, left-right preference, satisfaction with democracy, interest in politics, and the impact of immigrants in the country). The model shows that the effects of religiosity on the appeal of RRWPs are indeed conditioned on the presence of a minority party. When a minority party is present in the system, religious individuals have around a 15% chance of being drawn towards RRWPs, compared to around 10% in systems without minority parties. Furthermore, the direction of the relationship is different, in that in systems with at least one minority party, more religious individuals are more often drawn towards RRWPs, while in other countries they are less attracted to RRWPs. The model results, their visual presentation as well as information on the parties classified as RRWPs and the countries with minority parties in the EVS are given in our Online Appendix.

Discussion
Our results illustrate the correlation between religiosity and RRWP vote when a minority party is present; however, this finding warrants further consideration. In particular, the presence or absence of minority parties closely follows the East-West Europe division, although this is not so evident in the European Values Study sample from 2017. Regardless, if minority parties are present primarily in Eastern Europe, is it the presence of minority parties or another particular East vs. West Europe difference that accounts for the role of religiosity in RRWP vote?
In the first step undertaken to account for this observation, we did include a control variable for East and West Europe and despite this control, the moderating effect was still significant and altered the relationship between religiosity and RRWP vote. 4 However, to be certain the validity of our findings, a more nuanced examination is required. A recent paper by Marcinkiewicz and Dassonneville (2021), provides a partial answer to this question. They argue that the effects of religiosity are conditioned on the East-West division, primarily as a function of the strength of Christian Democratic parties. Where Christian Democratic parties are strong, primarily in Western Europe, religiosity is negatively related to RRWP vote, while this relationship takes on a different direction where Christian Democratic parties are weak or nonexistent, as is primarily the case in Eastern Europe. We agree with this explanation insofar as it refers to the general East-West divide in Europe. However, we argue that our exploration has more validity. We base this argument on several empirical and theoretical observations. First, when we inspect the countries where RRWP parties are most successful, the pattern is nonlinear, in that those countries that have medium levels of ethnic heterogeneity have the most successful RRWP parties ( Figure 3). We argued earlier that the salience of ethnic relations is what pushes religious voters towards RRWPs, and we should expect that ethnic relations are salient in countries with medium levels of fractionalization. We base this argument on the observation made by Chandra (2007) in her analysis of patronage democracies and ethnic clientelist networks, where she stipulates that in homogeneous or highly heterogeneous countries, ethnic linkages are hard to monitor. A similar logic should apply in our case as well, where the salience of ethnic relations would be hard to establish where societies are either very homogeneous or very heterogenous. To test this assumption, we fitted a multilevel model with an interaction effect between religiosity and ethnic fractionalization, with the same control variables as presented in the main part of our analysis. As we expected a nonlinear relationship to exist, we used the squared values of the fractionalization index. The results confirmed our assumption, showing that the effects of religiosity on vote choice are strongest in the medium tercile of ethnic fractionalization, while its strength diminishes in the upper tercile and completely reverses in the lowest tercile ( Figure 4).
These findings add more validity to our assumption that the salience of ethnic issues, proxied by the presence of a minority party or ethnic fractionalization, creates a favorable context for RRWPs to recruit religious individuals, as opposed to a situation where traditional Christian Democratic parties are relatively strong.
Secondly, while we do concur that Christian Democratic parties are much less prominent and successful in Eastern Europe, there are still countries where they do play a major role in the political system. To point out a few successful cases: People's Movement Party in Romania (5.3% of votes in 2018); Cristian Democratic Movement in Slovakia (8.8% in 2012Slovakia (8.8% in , 4.9% in 2018; New Slovenian Christians Peoples Party (4.8% in 2012, 5.6% in 2019, 7.1% in 2020) (Manifesto Project). Nevertheless, to further establish the validity of our argument, we fitted an additional model with the control for the presence of Christian Democratic party included. The results previously outlined in the analysis chapter remain almost identical (model can be found in Online Appendix).
Thirdly, Marcinkiewicz and Dassonneville (2021) make the somewhat tautological argument that the success of radical right parties among the religious electorate depends on the extent to which other right-wing parties are successful among such individuals. We believe this reasoning requires much more of an explanation, and that these dynamics can also be explained by the salience of ethnic relations. This much is implied by Koev (2015) who links the success of RRWPs to the success of minority parties in the previous electoral cycle. Regarding our argument here, where ethnic issues are salient, RRWP parties have success in mobilizing the religious electorate and increase their vote share most often at the expense of Christian Democratic parties.

Conclusion
In this paper, we explore whether contextual factors, specifically ethnic relations within a society, proxied by the presence of ethnic minority parties, moderate the effects of religiosity, and contribute to the increased likelihood of voting for RRWPs. Our research interest was guided by the scarce literature on the topic, and the inconclusive results on the nature of the relationship between religion and RRWP voting. We imagined that these mixed results were a consequence of a failure to account for significant factors that might push religious voters towards or insulate them against RRWP parties. We posited that ethnic issues in the political system push religious voters towards RRWPs, as a contextual factor that had not previously been explored.
While Koev (2015) considered the success of minority parties to be the general driver of the success of RRWPs in CEE Europe, we expanded on this argument and regarded the presence of any ethnic minority party as an indicator of the salience of ethnic relations in the political system. Specifically, when an ethnic minority party is present, it organizes and represents minority interests, therefore threatening the dominant culture in the society as perceived by RRWP voters. In such conditions, religious voters integrated into their cultural communities might be particularly sensitive to cultural threats and diversity and as such be incentivized to vote for RRWP parties and their nativist ideological platforms. The findings presented here confirm that ethnic relations moderate the relationship between religiosity and RRWP vote by making religious voters more likely to be RRWP supporters. These findings stand the test of multiple model specifications and robustness analysis as well as the introduction of important control variables at both the individual and country level.
As such, we have demonstrated that religion is an important predictor of RRWP voting, on top of individual attitudes and other demographic characteristics, but only when the proper contextual factors that foster the religious link to RRWP are accounted for. Our primary contribution is in identifying the ethnic relations, i.e. the presence of an ethnic minority party, to be such a moderating factor in the political system.
We recognize the importance of studying further contextual factors that help us explain the differences across and between countries related to various predictors and RRWP vote. Future research should try to expand on this, and potentially investigate in more detail the role of Christian Democratic parties around Europe, their size, or their party strategies, such as their emphasis on nativist rhetoric. These topics, though very important, were outside of the scope of this work, but we certainly hope that this paper will stimulate more research in this direction.