Islam, opinion climates, and immigrant party loyalties in Western Europe

Abstract This article examines how religion – particularly Islam – and anti-immigrant opinion climates influence the patterns of partisanship among first-generation immigrants in Western Europe. It suggests that Muslim propensity to become partisans and identify with Christian democratic parties depends on anti-immigrant opinion climate in their host country. The analyses based on individual-level data from the European Social Survey (ESS) 2002–2019 in 19 West European democracies reveal that while Muslims are indeed less likely to become partisans in anti-immigrant host societies, the opposite is true in hospitable opinion climates. Moreover, compared to other immigrants, Muslims are less likely to identify with Christian democratic parties, but this relationship is substantively small and limited to highly anti-immigrant countries. In contrast, Muslims are more likely to align with socialist parties at all levels of anti-immigrant sentiment. These findings have important implications for debates on immigrant political integration and the future of electoral alignments in Western Europe.

While there is evidence that religion influences party allegiances among ordinary citizens, we know much less about this relationship among immigrants.Most previous studies report that newcomers favour left-wing parties but offer little insight into the role of faith in shaping these loyalties (e.g.Bergh and Bjørklund 2011;Bird et al. 2011;Heath et al. 2013;Teney et al. 2010).It may be that while traditional morality associated with religion predisposes immigrants to support religious or conservative parties, other values -such as charity towards the poor promoted by some faiths, including Islam -motivate new arrivals to favour socialist or social democratic parties.Moreover, assessing how religious beliefs and practices affect party allegiances among foreign-born individuals requires disentangling their effects from the consequences of low socio-economic status, perceived discrimination, and other immigrant-specific factors.
This study examines how religious affiliation -particularly with Islam -matters for partisanship among first-generation immigrants in Western Europe.Specifically, I ask whether being a Muslim influences partisanship acquisition among foreign-born individuals, and if so, in what way.Moreover, I investigate to what extent being a Muslim affects party choice among partisans.And how much do these relationships depend on anti-immigrant opinion climates?The analyses based on individual-level data from the European Social Survey (ESS) 2002-2019 in 19 West European democracies reveal that, on average, foreign-born Muslims are neither more nor less likely to become partisans in their host country than other foreign-born residents.Instead, the effects of belonging to Islam are strongly conditioned by anti-immigrant sentiment in one's adopted homeland.Specifically, while Muslims are less likely to become partisans in highly anti-immigrant societies, they are more likely to do so in pro-immigrant settings.Moreover, although Muslims are less likely to identify with Christian democratic parties than other immigrants, this relationship is substantively small and detectable only in highly anti-immigrant countries.In contrast, Muslim identity is a powerful predictor of socialist party identification at all levels of anti-immigrant opinion climates.
This study contributes to existing research in several ways.First, it adds to our understanding how people acquire party allegiances.Although this question has interested political scientists for many decades (Campbell et al. 1960), we still know relatively little about how and why people who relocate to another country develop party loyalties.One reason for this omission is that traditional theories emphasising parental socialisation into partisanship are not applicable to foreign-born individuals.Yet, developing a comprehensive model of the sources of party identification among newcomers is important not only because immigrant shares in Western Europe are rising (Czaika and de Haas 2014), but also because, once acquired, partisanship generally endures over one's lifetime and is transmitted to later generations (Campbell et al. 1960;Dalton 2021).Moreover, research shows that party attachments play a critical role in motivating voting turnout and vote choice in the electorate (e.g.Campbell et al. 1960;Green et al. 2002;Rau 2022) and in shaping how people receive and interpret politically relevant information (e.g.Bartels 2002;Jerit and Barabas 2012;Zaller 1992).Thus, because partisanship has lasting and wide-ranging downstream effects on people's political behaviour and attitudes, developing party loyalties constitutes a central aspect of immigrant political integration.This article also adds to scholarship on the consequences of religious beliefs for immigrant socio-political integration (e.g.Banfi et al. 2016;Maxwell and Bleich 2014;McAndrew and Voas 2014;Reitz et al. 2009;Statham and Tillie 2016).As such, it goes beyond extant studies on the determinants of immigrants' religious beliefs (e.g.Cadge and Ecklund 2006;Connor 2010;Smits et al. 2010;Torrekens and Jacobs 2016;van Tubergen 2007;van Tubergen and Sindradóttir 2011) and contributes to this literature by systematically analysing how and why Muslim identity influences party attachments among first-generation immigrants.Moreover, the article enhances research on anti-immigrant attitudes that so far has focussed largely on the determinants of these attitudes (e.g.Ben-Nun Bloom et al. 2015;Hainmueller and Hiscox 2007;McLaren 2003) rather than their consequences.
Finally, most current research on immigrant party preferences or attachments is based on a small number of country (or even city) cases and examines a limited number of immigrant groups (e.g.Bergh and Bjørklund 2011;Bird et al. 2011;Dancygier and Saunders 2006;Hajnal and Lee 2011;Heath et al. 2013;Sanders et al. 2014;Teney et al. 2010;Wong 2000).By considering a wider range of West European countries with diverse newcomer populations, this study offers a more generalisable perspective on how individual religious beliefs and a macro-level environment in the form of anti-immigrant opinion climates influence party attachments among first-generation immigrants in Western Europe.
The article proceeds as follows.The next section develops my argument how and why religion -in particular, Islam -influences partisanship acquisition and its nature among foreign-born individuals in Western Europe.I then explain how the patterns of partisanship among new arrivals are shaped by host countries' anti-immigrant opinion climates, alone and in interaction with Muslim identity.To test my theoretical propositions, the subsequent section introduces the data and measures, followed by a presentation of my empirical results.I conclude with a discussion of my findings and their implications for future research.

Islam and partisanship acquisition
At the level of individuals, religion has two central features that may influence the development of partisanship: belonging to a religious group and religiosity, usually expressed as the intensity of one's religious beliefs and/or the frequency of religious practices.Regardless of one's level of religiosity, belonging to a religious group may be a powerful marker of one's identity because it provides an 'eternal' group membership, unmatched by belonging to any other social group (Ysseldyk et al. 2010; see also Kinnvall 2004;Verkuyten and Yildiz 2007).Religious affiliation also offers a compelling and comforting worldview that enables individuals to deal with uncertainty and hardship in their daily lives.Thus, as a social identity and a worldview with set values -guided and reinforced by religious institutions and their leaders -belonging to a religion may shape the way individuals relate to the political world, their party loyalties, policy preferences, and political engagement (Knutsen 2004: 99; see also Alex-Assensoh and Assensoh 2001;Jones-Correa and Leal 2001;Verba et al. 1995).
In this article, I am particularly interested in how belonging to Islam influences partisanship among foreign-born individuals in Western Europe.Existing research suggests that Muslim immigrants integrate into the civic and political life of their host societies to a lesser extent and more slowly than non-Muslims (Aleksynska 2011: 569;Bisin et al. 2008).Moreover, a considerable share of Muslims in western democracies remains without party preferences (Minkenberg 2010: 410-1).One obstacle to Muslim political integration is that Islam has been viewed as a source of conflict in secularised yet culturally deeply Christian Western Europe (Foner and Alba 2008).Although such context could in principle encourage more political engagement among Muslims -to achieve, for example, more favourable conditions for practicing their religion -in reality, many Muslims feel excluded and discriminated against, and these perceptions often motivate them to turn to religion to enhance their self-esteem (Alba and Foner 2015: 123;Voas and Fleischmann 2012: 530).Consistent with this perspective, research finds that perceived rejection or discrimination by host society not only bolsters religious identity among Muslims but also undermines attachment to their new homeland (Fleischmann and Phalet 2016;Verkuyten and Yildiz 2007).If religion indeed provides a refuge for many newcomers -as this 'reactive religion' perspective suggests (Voas and Fleischmann 2012: 537) -then Muslims should be less politically engaged, and hence, less likely to adopt partisanship compared to non-Muslims among foreign-born individuals in their host country (Hypothesis 1).

Islam and party choice among partisans
While believers usually align themselves with religious parties on the right of the ideological continuum (Botterman and Hooghe 2012;Elff 2007;Knutsen 2004;Raymond 2011;van der Brug et al. 2009), it is less clear whether this pattern applies to Muslims in Western Europe.On the one hand, Muslims -generally more religious and socially conservative than their host societies -may indeed favour parties with traditional value orientations.In the absence of Islamic parties, many Muslims may prefer Christian democratic parties because they represent Muslim preferences on social issues, such as abortion, gay marriage, and traditional family values more closely than other parties.That Muslims in Western Europe are socially conservative has been well documented (e.g.Bird et al. 2011: 71;Dancygier 2017: Ch. 3;Fish 2011;Klausen 2005a).For example, survey research reveals that Muslims consistently express less favourable attitudes towards gender equality in the labour market, university education, and political leadership than Christians (Fish 2011: Ch. 6), and that these attitudes translate into actual discriminatory behaviour (Adida et al. 2016: 89).Muslims are also more opposed to homosexuality -an attitude particularly strong among first-generation immigrants (Koopmans 2015: 47) -and to a lesser extent to abortion and divorce compared to Christians or non-believers (Fish 2011: Ch. 3). 2  Conversely, because socialist and liberal parties have been historically more critical of religious institutions and more committed to issues such as gender equality, women's reproductive rights, and secular education, they may be less appealing to Muslims.For example, scholars note that while most immigrants in the U.K. have supported the Labour Party, Muslims were attracted to the Conservative Party's rhetoric on family values in the 1980s (Otterbeck and Nielsen 2016: 56).The Christian Democratic Union in Germany similarly sought to appeal to Muslim voters by nominating Turkish-origin candidates committed to traditional values (Dancygier 2017: 98) and becoming the first party to appoint a Muslim minister to a state legislature (Dancygier 2017: 29).Cooperation between Muslims and the Christian Democrats based on shared ethical and religious principles was also documented in Denmark (Otterbeck and Nielsen 2016: 85).
We can therefore expect that, faced with a choice between secular and religious parties, Muslims may align with Christian democratic parties because of shared religious beliefs.Even though these parties are associated with Christianity, the religious values they espouse are closer to Islam than secularism.This expectation is consistent with existing evidence from public opinion surveys showing that Muslims are averse to non-believers in public office (Fish 2011: Ch. 2).As Fish (2011: 61) puts it, for many Muslims, politicians do not have to be Muslim, their faith does not have to be particularly pronounced, but they must not renounce belief in God.Consistent with this perspective, other scholars note that religious Muslims generally prefer a Christian state to one with no public religion (Klausen 2005b: 556) and that they often support Christian democratic parties because they believe that secularism alone cannot provide moral government (Klausen 2005a: 206). 3Finally, there is evidence that Muslims support religious parties because they expect that doing so will bring them divine rewards in the afterlife -a consideration particularly important for poor individuals (Grewal et al. 2019), such as immigrants.
On the other hand, rising tensions between Christianity and Islam in Western Europe may motivate Muslim immigrants to seek allies among secular rather than religious political actors.Existing research suggests that many Europeans consider Christianity as a basis for their national identity and its underlying values, especially in countries with sizable Muslim populations (Kunovich 2006;Zolberg and Woon 1999).There is also evidence that, with the rise of religious nationalism to counteract globalisation threats to self-identity (Kinnvall 2004), anti-immigrant attitudes have become more strongly linked to public concerns over cultural unity and homogeneity in Europe than to economic grievances (Sides and Citrin 2007;Sniderman et al. 2004;Storm 2011).In this environment, Muslims may be reluctant to side with religious parties, even when they feel close to these parties on social issues that are important to them (Fish 2011).Moreover, Muslims may be discouraged by the proclivity of many Christian democratic parties to focus on a culturally narrow, nativist conception of European Christianity instead of promoting faith and religious values more generally (Casanova 2007: 64;Klausen 2005a: 27).Finally, Muslims may also consider right-wing parties' efforts to appeal to them as not particularly credible because these parties have traditionally relied on ethno-centric voters opposed to Muslim presence in Europe (Dancygier 2017: 6).
While left-wing parties -especially socialist or social-democratic parties -differ markedly from Muslims in their social orientations, they do share views on some other issues.With respect to economic policies, Muslims generally support income redistribution more than non-Muslims.In the Islamic world, there is a long-standing annual tradition for Muslims to give 2.5% of their wealth to the poor (Fish 2011: 221), and for Islamist political actors to focus on social welfare provision as one of their core duties when in power (Cammett and Luong 2014: 198).Empirical evidence confirms that Muslims share their wealth more than Christians, and that income inequality in Muslim countries is significantly lower than in non-Muslim nations, especially when controlling for other factors such as economic development (Fish 2011: Ch. 6).
Beside economic preferences, Muslims may be attracted by left-wing parties' views on immigration.In-depth interviews with members of national parliaments in Western Europe reveal that left-wing politicians are more likely to see illegal immigration from Muslim countries as resulting from poverty, whereas those on the right attribute it to poor law enforcement and emphasise the need for strengthening border control (Cherribi 2007: 121).And while one of the core values of Christian democratic parties is compassion and charity for the disadvantaged (van Kersbergen and Krouwel 2008: 399), it is left-wing parties that have been historically more supportive of the socio-economic protection and anti-discrimination of immigrant minorities (Dancygier 2017: 83-4;Klausen 2005a: 21).Finally, left-wing parties nominate more candidates of immigrant background to winnable positions and pay more attention to the demands of immigrant voters than other parties (Alba and Foner 2015: 163), especially when electoral incentives for doing so are strong (Dancygier 2017).If this perspective is correct, then foreign-born Muslims should be less likely to identify with Christian democratic parties (Hypothesis 2a) and more likely to align with socialist parties (Hypothesis 2b) than other foreign-born residents.

Islam, opinion climates, and partisanship
Beside the role of Muslim identity, I am also interested in how the patterns of partisanship among first-generation immigrants vary in response to anti-immigrant opinion climate in their host country.I expect that anti-immigrant sentiment not only reduces the propensity of all foreign-born individuals to identify with a political party but also amplifies the effects of being a Muslim on partisanship acquisition and party choice among partisans.
Following previous research, I conceptualise anti-immigrant opinion climate as a constraint within a country's political opportunity structure (Kriesi et al. 1995;Tarrow 1998) that increases the costs and reduces the opportunities for immigrants' political engagement (Just and Anderson 2014: 938).Because anti-immigrant sentiment signals to newcomers that their presence is unwelcome, foreign-born residents are less likely to consider the expression of their political demands as legitimate and less likely to expect to find political allies willing to address these demands.In socially hostile settings, migrants may also worry that their political engagement will trigger a backlash from their host society, such as increased ethnic or religious harassment or even violence against their group members.In addition, they may fear a political pushback from the natives in a form of increased electoral support for far-right parties that may further restrict immigrant political and socio-economic rights.Conversely, highly pro-immigrant opinion climates should encourage newcomers to see fewer obstacles to their political participation and representation, and thus facilitate the emergence of their party allegiances.I therefore hypothesise that anti-immigrant opinion climates should be negatively linked to partisanship acquisition among foreign-born individuals in their host country (Hypothesis 3).
Beside inhibiting party loyalties among all first-generation immigrants, anti-immigrant opinion climates may also magnify the effects of Muslim identity on both partisanship acquisition and party choice among partisans.In Western Europe where collective identity is deeply rooted in Christian values and traditions, newcomers of other faiths, such as Islam, may be particularly sensitive to their social environment.Social psychology studies suggest that people who belong to subordinate or stigmatised groups are considerably more attuned to their context and pay attention even to nonverbal or affective signs from dominant group members (e.g.Frable 1997;Oyserman and Swim 2001).
In line with this view, scholars find that Muslim immigrants are less likely to become attached to their host country, trust its institutions, or adopt its values when they perceive their new homeland to be hostile towards them (Adida et al. 2016: Ch. 9;Verkuyten and Yildiz 2007).Moreover, claims against Muslims' rights in the national media have been found to reduce voting turnout among Muslims (Cinalli and Giugni 2016).Some scholars also suggest that anti-immigrant opinion climate and prejudice against Muslims motivate immigrants from Islamic countries to develop a 'reactive' religious identity (Connor 2010;Voas and Fleischmann 2012) and subsequently disengage from their host society (Verkuyten and Yildiz 2007).Other studies similarly conclude that anti-immigrant sentiment reduces immigrant political representation by undermining Muslim candidates' confidence to run for public office and discouraging political parties from nominating such candidates to winnable positions (Michon and Vermeulen 2013: 610;Schönwälder 2013: 646).
In short, because perceived hostility towards Muslims is negatively linked to their socio-political integration in western democracies (Adida et al. 2016: Ch. 9), Muslims should be less likely to develop party loyalties than other newcomers in more anti-immigrant opinion climates.Hostility towards new arrivals should also exacerbate Muslim reluctance to identify with Christian democratic parties because a society that is welcoming to immigrants is less likely to perpetuate a public perception of Islam as a 'problematic' religion.And so, I hypothesise that the negative effect of being a Muslim on partisanship acquisition among foreign-born individuals should be stronger in more anti-immigrant opinion climates (Hypothesis 4a).Hostile attitudes towards newcomers should also magnify the negative impact of being a Muslim on identification with Christian democratic parties among foreign-born individuals (Hypothesis 4b).

Data and methods
My empirical analyses employ individual-level data from the European Social Survey (ESS) 1-9 round (2002-2019) cumulative file.Conducted every two years since 2002, this project is renowned for its high standards in cross-country survey design and data collection (Kittilson 2009).It relies on strict random sampling of individuals regardless of nationality, language, or legal status to generate nationally representative samples.Moreover, the data include samples of foreign-born respondents that closely track their host nation's official statistics regarding migrants' origin countries (de Rooij 2012; Just and Anderson 2012). 4Beside standard questions about partisanship, the ESS contains items related to respondents' foreign-born status as well as individual religious beliefs and practices.Furthermore, the ESS data have been used to analyse Muslim immigrant political or socio-economic integration in previous research (e.g.Adida et al. 2016: Ch. 9;Connor 2010;Connor and Koenig 2013;Just et al. 2014).Information on relevant variables is available for 19 advanced industrialised democracies from Western Europe with substantial immigrant populations: Austria, Belgium, Cyprus, Denmark, Germany, Great Britain, Greece, Finland, France, Iceland, Ireland, Italy, Luxembourg, Netherlands, Norway, Portugal, Spain, Sweden and Switzerland. 5

Dependent variables
My empirical analyses proceed in several steps.First, I analyse whether Muslims are less likely to be partisans than other foreign-born individuals.To measure partisanship, I employ two ESS questions: 'Is there a particular party you feel closer to than all the other parties?'If so, 'Which one?'These survey items have been widely used in cross-national research because they provide a clear opportunity for respondents to choose a 'non-identity' , and have been shown to offer a valid indicator of partisanship across countries (Blais et al. 2001;Sanders et al. 2002).My initial measure of partisanship is dichotomous, where one indicates feeling close to a party, and zero -otherwise. 6 Second, I focus on foreign-born respondents who identify with a political party and examine whether Muslims are more likely to report attachment to some parties than other immigrants.To do so, I follow existing cross-national research (van der Brug et al. 2009;Knutsen 2004) and group parties into party families: Christian democratic, conservative, socialist, liberal, green, radical right, radical left, and agrarian/centre parties. 7I then use identification with a party in each party family as my dependent variables.To create these measures, I match individual responses to survey questions about partisanship with information on party families in the Chapel Hill Expert Survey (CHES) data (Jolly et al. 2022).The resulting variables are dichotomous, where one indicates feeling close to a party from a selected party family, and zero -any other party. 8

Key independent variables
In order to assess who is foreign-born in my data, I rely on the following survey question: 'Were you born in this country?' Respondents who said 'no' were classified as foreign-born, while those who said 'yes' or did not answer the question were dropped from the sample.In addition, I use questions 'Was your mother born in this country?' and 'Was your father born in this country' to identify respondents who are foreign-born but whose both parents are native-born and exclude them from my analyses.Subsequently pooling observations across countries and nine survey rounds generates a sample of 24,019 foreign-born individuals (with at least one foreign-born parent), comprising 9.21% of all respondents in the ESS data.
To examine the consequences of religion at the level of individuals, I distinguish between religious affiliation, religiosity, and attendance of religious services.Religious affiliation is based on the survey question: 'Do you consider yourself as belonging to any particular religion or denomination?'If so, 'Which one?' Using responses to these questions, I identify Christians, Muslims, and other believers, as well as those who do not belong to any religion.In my sample of first-generation immigrants, there are 3,448 Muslims (14.36%), 10,741 Christians (44.72%), 927 other believers (3.86%), and 8,130 non-believers (33.85%). 9 Beside religious affiliation, religiosity is measured with the question: 'Regardless of whether you belong to particular religion, how religious would you say you are?' Individual responses were coded on a scale from zero to ten, with higher values indicating more religiosity.To capture the frequency of religious services attendance, respondents were asked: ' Apart from special occasions such as weddings and funerals, about how often do you attend religious services nowadays?'Individual answers range on a scale from zero to six, with higher values indicating more frequent attendance. 10 Finally, following previous research (Connor 2010; Just and Anderson 2014), I measure anti-immigrant opinion climate using the following survey questions: 1) whether immigration is bad or good for their country's economy, 2) whether immigrants undermine or enrich the country's cultural life, and 3) whether immigrants make the country a worse or better place to live.Individual answers to these questions (captured on a scale from zero to three, with higher values indicating more anti-immigrant views) were first used to compute an average score for each respondent in the ESS data.These scores were then employed to calculate country means in every ESS round. 11The data reveal that this variable ranges from .680(Sweden, 2014-5) to 2.122 (Cyprus, 2012-3), with an overall mean of 1.365.

Control variables
Beside my key independent variables, I include controls found to be important determinants of partisanship in previous research.All models account for respondent's socio-economic status measured as income, education, unemployment, and manual skills.Higher socio-economic status usually enables and motivates people to engage politically in various countries (Jennings et al. 1989;Verba et al. 1995), and these findings extend to immigrants' partisanship acquisition (Wong 2000;Wong et al. 2011).Socio-economic status also matters for party choice among partisans, as left-wing parties have traditionally defended the interests of the poor (Cain et al. 1991;Zingher and Thomas 2012).
My models additionally control for gender because women have been less politically active than men due to socialisation into traditional gender roles and more limited resources available to them (Burns et al. 1997;Jennings 1983).Political engagement also increases with age (Niemi et al. 1985), and this finding applies to partisanship acquisition among foreign-born individuals, even when accounting for their duration of stay in host country (White et al. 2008).I control for gender and age also in my models of party choice among partisans because female and younger respondents among first-generation immigrants are more likely to identify with left-wing parties than male or older individuals (Just 2019: 671).
Beside standard socio-demographic controls, political engagement is usually higher among more socially connected individuals (Cain et al. 1991;Giugni and Grasso 2020).To gauge social connectedness, I rely on marital status, union membership, and how frequently one meets with others.Moreover, I include being a crime victim (Bateson 2012) and perceived discrimination, previously shown to influence immigrant socio-political integration (Fleischmann et al. 2011;Maxwell 2006) and party choice (Sanders et al. 2014).
Finally, with respect to immigrant-specific characteristics, I control for the duration of stay, citizenship, and proficiency in host country's official language (de Rooij 2012; Just 2019Just , 2021;;White et al. 2008;Wong 2000).To account for pre-migration experiences, I consider whether an individual came from a party autocracy -a variable taken from the Autocratic Regimes dataset (Geddes et al. 2014) -because arrivals from party autocracies are less likely to become partisans than newcomers from other political regimes (Just 2019).Another relevant control is whether one's origin country had a communist leader prior to arrival (Cheibub et al. 2010) because those who originate from communist countries are less likely to identify with left-wing parties than other immigrants (Just 2019).Finally, all models include ESS round fixed effects to account for the cumulative structure of the ESS data (For further details, see the Online appendices).

Empirical strategy and results
My empirical analyses rely on information collected at the levels of individuals and countries, often across multiple survey rounds.Thus, my data have a multi-level structure, where one unit of analysis -the individual -is nested within another unit -the country-round.Such data structure may cause statistical problems, including non-constant variance, clustering, and incorrect standard errors (Steenbergen and Jones 2002).The results reported below are therefore multilevel -random intercept -logistic regression estimates, listed as log odds with their standard errors in parentheses and odds ratios in italics.
Previous research suggests that thinking in partisan terms among immigrants is a step that is often separate from, or prior to, identification with a particular party (Hajnal and Lee 2011;Wong et al. 2011: 130).Hence, I examine party attachments among foreign-born individuals as a two-step process: (1) partisanship acquisition and (2) identification with a particular party type among partisans.
Focussing on partisanship acquisition as my dependent variable in Table 1, I first provide the results of my estimations using non-Muslims as the reference category for Muslims.The additive model in the first column reveals a statistically insignificant coefficient of being a Muslim.This means that, on average, foreign-born Muslims are neither more nor less likely to become partisans in their adopted homeland than foreign-born non-Muslims, failing to support Hypothesis 1.In contrast, the estimate for anti-immigrant opinion climate is negative and highly statistically significant, indicating that all foreign-born residents are less likely to become partisans in more anti-immigrant societies, in line with Hypothesis 3.
Looking at the results of the interaction model in the second column reveals a more nuanced story, however.Specifically, the additive term of Muslim identity becomes positive and highly statistically significant, while the multiplicative term between being a Muslim and anti-immigrant sentiment is also highly statistically significant but negative.This means that in opinion climates that are welcoming to immigrants, foreign-born Muslims are more likely to be partisans than foreign-born non-Muslims.However, in line with Hypothesis 4a, this relationship weakens and possibly reverses in highly anti-immigrant societies.To probe deeper into these results, the third and fourth columns in Table 1 report the estimates of my models using Christians (instead of non-Muslims more generally) as the reference category for Muslims. 12The results remain essentially the same: while Muslim identity is statistically insignificant in the additive model, its additive and multiplicative terms are highly statistically significant in the interaction model.In order to assess the substantive impact of these variables, Figure 1(a) plots the marginal effects (with 95% confidence intervals) of being a Muslim (vs.Christian) on the probability of partisanship acquisition among foreign-born individuals as we shift across the scale of anti-immigrant opinion climate in my sample of countries. 13In line with my expectations, the marginal effect of belonging to Islam is positive and statistically distinguishable from zero at low values of anti-immigrant opinion climate -that is, in host societies that are welcoming to newcomers.However, as we move from the minimum to the maximum value of anti-immigrant sentiment, the marginal effect of being a Muslim on the likelihood of having partisanship reverses entirely (.080 vs. −.089) and also becomes highly statistically significant. 14 Figure 1(b) provides additional support that Muslim identity and anti-immigrant sentiment interact in shaping partisanship acquisition among first-generation immigrants.It plots the marginal effect (with 95% confidence intervals) of anti-immigrant opinion climate on the probability of partisanship separately for Christians and Muslims.The results show that the marginal effect of anti-immigrant opinion climate is negative and statistically distinguishable from zero for both groups, but it is almost twice as large for Muslims than it is for Christians (-.14 vs. −.27).Taken together, the results confirm that there is a distinct pattern of partisanship acquisition among foreign-born individuals in Western Europe that is linked to their religious identity.However, this pattern emerges only when we account for the conditional effect of anti-immigrant opinion climate: while Muslims are less likely to report partisanship than other foreign-born individuals in highly anti-immigrant opinion climates, they are more likely to do so in pro-immigrant environments.
In order to understand how Muslim faith influences the nature of partisanship, the next step of my analyses focuses exclusively on partisans among foreign-born residents and examines their attachments to two party types -Christian democratic and socialist parties.The results presented in Table 2 reveal that, compared to Christians, Muslims are less likely to identify with Christian democratic parties, and more likely to align with socialist parties. 15Hypotheses 2a and 2b are therefore supported by the data.Moreover, in line with Hypothesis 4b, I find a statistically significant coefficient of the interaction between being a Muslim and anti-immigrant opinion climate in the model of Christian democratic but not socialist party identification.
To facilitate the interpretation of these results, Figure 2(a) and 2(b) plot the marginal effects (with 95% confidence intervals) of being a Muslim on the probability of identification with Christian democratic and socialist parties, respectively, at various values of anti-immigrant opinion climate. 16Figure 2(a) reveals that while Muslims are less likely to identify with Christian democratic parties in highly anti-immigrant  settings; this relationship is not statistically significant in more pro-immigrant environments.In comparison, Figure 2(b) shows that Muslims are significantly more likely to identify with socialist parties than Christians, and this positive effect is at least 3.6 times larger than the negative impact observed with respect to Christian democratic partisanship (.208 vs. −.058 at the highest value of anti-immigrant opinion climate in my sample of countries).In short, while there is a small negative effect of belonging to Islam on Christian democratic party identification, this effect is absent in pro-immigrant opinion climates.Moreover, Muslim immigrants are considerably more likely to align with socialist parties than Christian immigrants, regardless of how welcoming or unwelcoming their host society is to newcomers.Finally, Table 3 shows that Christian democratic and socialist parties are not the only party families where Muslim identity matters. 17Specifically, compared to other first-generation immigrants, foreign-born Muslims are less likely to identify with conservative parties, and are particularly repelled by far-right parties.However, the results in the models of liberal, green, or far-left party identification are less clear-cut.For example, while Muslims are less likely to align with liberal parties than Christians, there is no statistically significant difference between Muslims and secular individuals.In contrast, Muslims are neither more nor less likely to identify with green parties relative to Christians, but they are considerably less likely to do so than non-believers.Finally, Muslims are more likely to attach themselves to far-left parties than Christians, but not in comparison to secular individuals.
In order to test the robustness of my findings, I have performed several additional analyses.First, to ensure that anti-immigrant opinion climate is not a proxy for party discourse on immigration, I re-estimated my models while additionally controlling for the electoral strength of far-right parties.Far-right party vote share in national legislative elections preceding ESS interviews in each country-round was statistically insignificant in all my models, while the results with respect to my key variables of interest remained essentially the same (see Table B1a,b in the Online appendices).Second, to capture immigration policy stances of all parties in a system, I included party system mean and polarisation on immigration control (both weighted by party vote shares).I found that while party polarisation on immigration control contributes positively to partisanship acquisition among foreign-born individuals, my main findings did not change (Table B2a,b).Third, my results are also robust to the inclusion of foreign-born population share (Table B3a,b) and the migrant integration policy index (MIPEX) (Solano and Huddleston 2020) (Table B4a,b). 18 Furthermore, to ensure that the observed relationship between anti-immigrant opinion climate and immigrant partisanship is not causally reciprocal, I re-estimated my models using anti-immigrant opinion climate lagged by one survey round in the cumulative ESS 1-9 round data but found no change in my key results (Table B5a-c).In addition, following Baysu and Swyngedouw (2020), I examined whether the effects of my key explanatory variables depend on respondent's religiosity.One way to assess this possibility is to divide my sample between respondents with high and low levels of religiosity, and to re-analyse my models separately in each group. 19These split-sample estimations revealed that my key variables remained statistically significant and were substantively the same for both more and less religious respondents .
Taken together, the results confirm that Muslim identity systematically shapes the patterns of partisanship among first-generation immigrants in Western Europe.However, this pattern emerges only when we account for the conditional effects of anti-immigrant opinion climate.While foreign-born Muslims are less likely to report party allegiances than other foreign-born individuals in anti-immigrant host societies, the pattern is reversed in pro-immigrant opinion climates.Among newcomers who have acquired partisanship, being a Muslim is negatively linked to identification with Christian democratic parties, but this effect is substantively small and detectable only in highly anti-immigrant societies.In contrast, belonging to Islam is positively associated with socialist party identification and this relationship is highly pronounced at all levels of anti-immigrant attitudes.

Conclusions
Many public and scholarly debates on Muslim immigrants in Western Europe consider Islam as a source of conflict and an obstacle to immigrant integration (Casanova 2007).By challenging values such as secularism, individualism, and gender equality, Muslims are feared to introduce new socio-political divisions and thereby undermine democratic governance in their host countries.The relevance of Islam in western democracies is unlikely to diminish any time soon because Muslims have higher fertility rates and stronger religious commitment than Christians (Cesari 2013: Ch. 4;Fischer et al. 2007).Moreover, religiosity among Muslims is stable across generations (Jacob and Kalter 2013;Leszczensky et al. 2020;Soehl 2017), in large part because Muslims arrive from highly religious countries and because their faith is subsequently fostered through residential concentration, support from Islamic communities, and religious endogamy (Voas and Fleischmann 2012).
This study was designed to examine how Islam and anti-immigrant sentiment in host societies influence party loyalties among first-generation immigrants in Western Europe.Consistent with my expectations, the results reveal that anti-immigrant opinion climates not only reduce partisanship among all foreign-born respondents but also powerfully condition the relationship between Muslim identity and partisanship.Specifically, with respect to partisanship acquisition, I find that while foreign-born Muslims are less likely to become partisans than other foreign-born individuals in opinion climates that are hostile towards newcomers, the opposite is true in hospitable opinion climates.Furthermore, although Muslims are generally less likely to identify with Christian democratic parties than Christians, this difference is substantively small and absent in host societies that are welcoming to immigrants.At the same time, Muslim identity is strongly linked to socialist party identification at all levels of anti-immigrant sentiment.
These findings are in line with previous research suggesting that no single party can take Muslim support for granted (Dancygier 2017: 10).However, targeting Muslims electorally is not without dilemmas for West European parties.On the ideological left, nominating Muslim candidates and catering to Muslims electorally signals party commitment to minority inclusion and diversity.Yet because Muslims are on average more socially conservative than their host societies, Muslim incorporation presents a risk of alienating party supporters who care deeply about social issues, such as gender equality, gay rights, and secularism.On the ideological right, appeals to Muslims may be compatible with party commitment to social conservatism, but they often irritate ethno-centrists concerned with their country's cultural homogeneity (Dancygier 2017: 7).Thus, efforts of established political parties to transform Muslims into their partisans run the risk of undermining party ideological coherence, and, consequently, turning away some of party's traditional supporters.At the same time, while not always optimal as an electoral strategy, attracting partisan support from different social groups, including Muslims, may enable parties to mediate socio-political conflict and thereby enhance the stability and functioning of their country's political system (Lijphart 1968;Rose and Urwin 1969).
While my theoretical expectations are strongly supported by the available data, this study is not without limitations.Empirical analyses of party loyalties among first-generation immigrants would benefit from larger sub-samples of Muslims than currently offered in the ESS data (Connor and Koenig 2013: 32).Moreover, using cross-sectional data cannot fully rule out a potentially reciprocal causality between Muslim identity and partisanship.Future studies based on panel data, in-depth interviews with immigrants, and larger Muslim samples could help scholars to assess this relationship with more accuracy.
Future research on partisanship among Muslims may also delve deeper into the underlying reasons of party support.Strong attachment to socialist parties identified among foreign-born Muslims in this study may go beyond values or issue orientations.In Belgium, for example, trade unions that often helped newly arrived immigrants to get settled, played an important role in socialising them into long-term support for the Socialist party (Eelbode et al. 2013).Furthermore, while today both left-and right-wing parties nominate minority candidates (Dancygier 2017;Eelbode et al. 2013;Sobolewska 2013), 20 left-wing parties have a more extensive record of doing so than other parties.To what extent Muslim immigrants respond to party long-term policy reputation and recruitment practices as opposed to short-term electoral strategies is a question deserving further research.
Understanding the sources of immigrant party loyalties would also benefit from a more systematic account of party policy positions.For example, there is evidence that foreign-born individuals are more likely to become partisans when parties in their host country take more distinct positions on immigration control (Just 2021).Beside directly measuring party policy positions, scholars may also want to assess the degree to which political parties invoke social identities -such as religious affiliation or religiosity -in their electoral campaigns (Dickson and Scheve 2006) and analyse the consequences of these choices for immigrant party attachments.
Another potentially fruitful area for future research is to consider the consequences of diversity among Muslims.This study focussed on Muslims as a single group to assess whether there are systematic differences in the patterns of partisanship between Muslims on the one hand and non-Muslims or Christians on the other.However, many scholars note a large diversity among Muslims in Western Europe that fragments them politically (Klausen 2005a: 28;Statham and Tillie 2016: 184;Warner and Wenner 2006) and this fragmentation may have consequences for Muslim political behaviour in their host country.More research is needed to ascertain whether belonging to a particular Muslim group defined by religious doctrine, ethnicity or minority status in origin country offers a more nuanced perspective on immigrant party allegiances in Western Europe.

Notes
1.In 2016, Europe -defined as 28 EU countries, plus Norway and Switzerland -was home to 25.8 million Muslims (4.9% of total population) (Pew Research Centre 2017).
2. The gap in social issue preferences between Muslims and non-Muslims in Western Europe is particularly pronounced in urban areas, in part because migrants sort themselves into conservative co-ethnic enclaves and also because preferences between Muslims and non-Muslims living nearby have become more polarized (Dancygier 2017: Ch. 3). 3.In Muslim countries, Islamists have often enjoyed an electoral advantage because they are automatically assumed to be more honest, fair, and incorruptible than other political actors (Cammett and Luong 2014: 201).4. Estimating my models on sub-samples of foreign-born individuals who are more likely to be under-sampled in the ESS data -that is, those who do not speak their host country's official language at home, non-citizens, and more recent arrivals (that is, those who arrived ten or fewer years ago) -reveals even stronger results with respect to my key variables (Table B7a,b in the Online appendices).5.The country and survey-round combination produces 134 macro-level units.6. 39% of foreign-born respondents (with at least one foreign-born parent) reported having partisanship compared to 54% of native-born individuals.7.For information on parties in each party family by country, see the Online appendices (Table A2).8.I focus on partisanship rather than voting in part because many first-generation immigrants are non-citizens and do not have a legal right to vote in their host country's national elections.Excluding foreign-born non-citizens (54% of my sample) would reduce the generalizability of my findings to first-generation immigrants.Moreover, partisanship has been shown to play a critical role in motivating electoral behavior, including voting turnout and vote choice (e.g.Campbell et al. 1960;Green et al. 2002) as well as in shaping how people receive and interpret politically relevant information (e.g.Bartels 2002;Jerit and Barabas 2012;Zaller 1992).Thus, because both foreign-born citizens and non-citizens may become partisans in their host country and because, once acquired, partisanship shapes people's subsequent political behavior and orientations, it constitutes a central aspect of immigrant political integration.9. See Table A1 in the Online appendices for information by country.10.I use attendance of religious services to capture the effect of religious institutions above and beyond individual religiosity.Research shows that mosques play an important role in mobilizing Muslims politically because they enable political candidates to reach a large audience quickly and efficiently, and religious leaders often instruct their followers which parties to support (Dancygier 2017: 89).11.My ideal measure would focus on people's attitudes towards Muslim immigrants rather than immigrants in general.However, such questions are not available in the ESS data.Moreover, existing research suggests that, as most migrants arriving to Western Europe are Muslim, immigration and Islam have become almost synonymous (Casanova 2007: 61;Cesari 2011;Zolberg and Woon 1999).12.The results using non-believers as the reference category for Muslims are available in the Online appendices (Table A4).13.Other variables are held at their means and dichotomous variables at their medians.14.In comparison, the marginal effect of being male (vs.female) on the probability of having partisanship is .054points, while moving from the minimum to the maximum value of income (on a scale from zero to three) yields an increase of .049points.15.The results using non-Muslims or non-believers as the reference category for Muslims are the same (Table A5a,b in the Online appendices).16.I focus on the range of pro-immigrant opinion climates where Muslims are more likely to become partisans than Christians, as shown above.17.These models also include all controls (not shown).18.Moreover, there is no evidence that being a Muslim interacts with immigrant integration policies in shaping the patterns of partisanship among foreign-born individuals (Table B4c,d).19.I used the median value among foreign-born respondents to divide the sample.20.Far-right parties are, of course, an exception.

Figure 1 .
Figure 1.(a) Marginal effects of being a Muslim (vs.christian) on partisanship probability among foreign-born individuals by anti-immigrant opinion climate in Western europe, 2002-2019.(b) Marginal effects of anti-immigrant opinion climate on partisanship probability among Muslim and christian foreign-born individuals in Western europe, 2002-2019.

Figure 2 .
Figure 2. (a) Marginal effects of being a Muslim (vs.christian) on the probability of christian Democratic party identification among foreign-born partisans by anti-immigrant opinion climate in Western europe, 2002-2019.(b) Marginal effects of being a Muslim (vs.christian) on the probability of socialist party identification among foreign-born partisans by anti-immigrant opinion climate in Western europe, 2002-2019.

Table 1 .
partisanship acquisition among foreign-born individuals inWestern europe, 2002-2019.Multilevel (random-intercept)logistic regression estimates obtained using the stata's xtlogit command; odd ratios are in italics and standard errors in parentheses.Having arrived to one's host country 6-10 years ago is the reference category for other periods of arrival, and non-party autocracy -for party autocracy and democracy in origin country.*p≤ .05,**p≤ .01, ***p≤ .001(two-tailed).

Table 2 .
identification with christian democratic or socialist parties among foreign-born partisans in Western europe, 2002-2019.Multilevel (random-intercept) logistic regression estimates obtained using the stata's xtlogit command; odd ratios are in italics and standard errors in parentheses.christian is the reference category for Muslim, other believer, and non-believer.*p≤ .05,**p≤ .01,***p≤ .001(two-tailed).