How and why does religion matter for the integration of Muslim minorities? Differentiating religiosity from religious orientations among Turkish Muslims in Germany

ABSTRACT

The high level of religiosity of immigrants in general and Muslims in particular is often conceived of as a barrier to integration in European public and scholarly debates. However, existing survey research on the relation between religion and integration was limited to measures of religiosity and has yielded contradictory findings. For a broad range of outcomes spanning the domains of structural, social and cultural integration, there are studies showing positive, negative or no associations with immigrants’ religiosity. Arguing that religious cognitions are independent of religiosity, we examine whether the way immigrants reason about religion explains additional variation in immigrant integration. Drawing on original survey data of Turkish-origin Muslims in Germany (N = 500), we examine how religious literalism, symbolism and theological exclusivism relate to a broad range of integration outcomes (educational attainment, labour market participation, intergroup contact, civic and political participation, volunteering, sense of belonging to Germany and social distance), over and above levels of religiosity. Factor analyses reveal that a symbolic approach to religion (i.e. emphasizing the need to take the historical context into account when interpreting religious texts) is conceptually distinct from literalism and exclusivism and can explain variation beyond religiosity, particularly in the domain of civic integration.

KEYWORDS:

• Religion
• religious cognition
• migration
• integration
• Turkish Muslims

Introduction

Religion has made a forceful re-entry into public and scholarly debates and is currently considered the most contentious issue in dealing with diversity (Brubaker 2015). Large-scale research on differences between migrants and natives in levels and trends of religiosity has flourished over the past 20 years. This has established that immigrants in general, and Muslims in particular, are more religious than non-migrants in Europe (e.g. Van Tubergen and Sindradóttir 2011). Moreover, immigrant families are motivated to pass on their religion across generations (Phalet, Fleischmann, and Hillekens 2018; Suárez-Orozco et al. 2011), and the intergenerational transmission of religion is comparatively more successful in Muslim migrant families (e.g. Şimşek et al. 2018). There is thus little doubt about the importance of religion for immigrants to Europe. However, there are conflicting and inconclusive findings on how immigrants’ religious involvement relates to their position in traditionally Christian but increasingly secularised societies (Fleischmann 2022). A reason for these non-conclusive findings may be that existing surveys on immigrants’ religion have only captured a part of what religion means to individuals. Available measures most frequently assess the importance of religion and the frequency of participating in religious practices, but largely ignore individuals’ reasoning about religion. Consequently, survey-based research has missed out on changes in the meaning and implications of religion that could occur with immigrant adaptation.

This is an important omission when considering research among non-migrants that has convincingly shown that the ways in which individuals reason about religion is a more important predictor of a range of attitudes, values and behaviours than religiosity (e.g. Duriez et al. 2007; Pickel et al. 2020). Yet this research has been limited to non-migrants with Christian heritage, raising the question whether religious cognition or reasoning can also be distinguished from religiosity among migrants with other religious backgrounds, and whether both aspects of religion relate to outcomes such as intergroup attitudes and contact similarly in non-Christian populations.

This research therefore aims to examine whether religious cognitions, specifically religious literalism, symbolism and theological exclusivism, can explain immigrant integration beyond levels of religiosity. We take a comprehensive approach to the multidimensional integration concept and include a number of indicators capturing structural, social, political and cultural domains of integration, which we consider as dependent variables (cf. Beek and Fleischmann 2020). For all indicators, we examine whether religious cognitions add to the explanatory power after religiosity and socio-demographic controls have been taken into account. Empirically, we test our model based on original survey data among Turkish-origin Muslims living in Germany from the KONID-study (Liedhegener et al. 2021), which contains a broad range of measures of religious cognitions and religiosity, in addition to commonly used measures of immigrant integration.

With this research, we aim to contribute to the scholarly and societal debate on the role of religion for immigrant integration by focusing on less frequently studied aspects of religion, among an important minority group. Our novel approach of relating several aspects of religious cognitions to a broad range of integration outcomes does not only have the potential to move scholarship on the role of religion for immigrant integration forward, it can also provide a cross-cultural extension of research in the psychology of religion by examining how religious cognitions relate to religiosity in non-Christian samples.

Theory & hypotheses

Religion and immigrant integration

Scholars have argued that religion acts as a barrier to immigrant integration in Europe for three reasons: the discrepancy in religious affiliations between immigrants and hosts (due to the large-scale migration from Muslim-majority countries), immigrants’ high levels of religiosity that contrast with prevalent secularism among their hosts, and European institutional approaches to the governance of religion (Foner and Alba 2008). In this literature and the current study, immigrant integration is conceptualised as a two-sided process of boundary dynamics where two initially not overlapping groups – i.e. immigrants and non-migrants – negotiate their group boundaries (Alba 2005; Wimmer 2008). These boundary dynamics occur in all domains of social life and encompass socio-economic, social and cultural aspects (Gordon 1964), with socio-economic or structural integration being considered a prerequisite for integration in other domains (Alba and Nee 2003). The key question is then how religion affects boundary dynamics, and specifically, how religious affiliations and expressions of religion create new boundaries, reinforce existing ones, or contribute to narrowing gaps in socio-economic attainment, inter-group contacts, attitudes, identification and political behaviour (Hirschman 2004).

Two research lines developed during the last 20 years to empirically address this question in the context of European post-WWII immigration. The first studied how immigrant religion changes due to migration. Different research approaches to studying religious change, ranging from the comparison of immigrant generations (e.g. Diehl and Koenig 2009), parent–child dyads (e.g. Van de Pol and van Tubergen 2014) and within-person trends (e.g. Fleischmann and Khoudja 2023) have yielded contradictory results. Some studies find intergenerational stability (e.g. Beek and Fleischmann 2020) or even increasing religiosity (e.g. Şimşek et al. 2018), while others suggest that religion declines due to migration (e.g. Khoudja 2022). The second research line examined the role of religion for immigrant integration, and likewise yielded contradictory findings. Regarding socio-economic integration, some studies find positive associations of religiosity with educational and labour market attainment (e.g. Carol and Schulz 2018), others negative (e.g. Güveli and Platt 2011) or non-significant ones (e.g. Koenig, Maliepaard, and Güveli 2016). Similarly mixed results have been found for social integration, with negative (e.g. Beek and Fleischmann 2020) and non-significant (e.g. Maliepaard and Schacht 2018) associations for inter-group contact, but positive relations with participation in civic associations (e.g. McAndrew and Voas 2014). Similarly, for cultural integration, some studies reveal that greater religiosity goes together with more conservative attitudes regarding sexuality and gender relations (e.g. Kogan and Weißmann 2020) whereas others find no association (e.g. Scheible and Fleischmann 2013). Also for national identification, the findings range from positive (e.g. Fleischmann, Leszczensky, and Pink 2019) via non-significant (e.g. Beek and Fleischmann 2020) to negative associations (e.g. Fleischmann and Phalet 2018). These discrepant results cannot be systematically related to differences in the destination countries, immigrant groups or generations under study. The upshot of two decades of large-scale survey research on the role of religion for immigrant integration in Europe is therefore rather puzzling: on the one hand, there is strong evidence for the continued importance of religion among immigrants, but on the other, it is still unclear how and why this matters for their integration into European societies.

Beyond large-scale surveys, qualitative studies of immigrant religion have already pointed out that the ways in which (children of) immigrants relate to their religion differ within and across congregations (e.g. Ebaugh and Chafetz 2000). Much of the ethnographic work in the European context has focused on Muslims, e.g. studying differences in how European Muslims conceptualise Muslim identity (e.g. De Koning 2008). Other work has documented how different religious organisations within the European Muslim community take distinct positions in theological debates and on social issues (e.g. Sunier and Landman 2015). While this work jointly documents the importance of internal diversity and individual differences in how members of the same religious community interpret and derive meaning from their religion, and thus stands in stark contrast to the tendency of ‘Muslimification’ of much survey-based research (Statham 2024), the depth of these investigations has necessarily limited their scope, and the results have so far not found their way into survey-based investigations among larger samples. The present study attempts to take a first step in this direction by examining specific religious cognitions in addition to immigrants’ religiosity as potentially complementary explanations for integration in multiple domains, among a specific group that has been extensively studied previously: Turkish-origin Muslims in Germany. Before we specify the context of the present research, we will argue why it is important to distinguish between religiosity and religious cognitions in studying the role of religion for immigrant integration.

Religiosity and religious cognitions

Psychologists of religion caution against a simplistic and static interpretation of religious affiliations and instead emphasise different understandings of religion (Allport 1979). Following Wulff (1997), individual differences in meaning-making or reasoning about religion should be considered as one out of two dimensions of individuals’ relation with religion. The other dimension is religiosity (also referred to as affirmation vs. disaffirmation) and captures the extent to which individuals include the possibility of transcendence, attach importance to religion and participate in religious rituals. This corresponds to existing survey-based measures of immigrant religion which typically include assessments of religious identification and the frequency of participation in religious rituals, and more rarely the strength of religious beliefs (cf. Pickel et al. 2020; Voas 2007). The content or meaning of these beliefs is more appropriately captured by the dimension of religious cognition, which can range from literal to symbolic (Fontaine et al. 2003; Wulff 1997). Literalists insist that there is only one correct interpretation of religious scripture and that every religious question has a single, immutable answer. Symbolists, in contrast, emphasise the need to (re-)interpret religious messages and acknowledge the value of multiple worldviews. The notion of symbolic religious cognition goes back to Ricoeur’s work on modern hermeneutics and the idea that restorative interpretation is necessary to surpass initial suspicion and uncritical faith in order to achieve a reflective religious faith (Fontaine et al. 2003).

The religious cognitions of literalism and symbolism thus resemble (and have been empirically shown to overlap with) religious fundamentalism and religious quest (cf. Batson, Schoenrade, and Ventis 1993), respectively, in that they emphasise distinct individual tendencies to approach religious matters in a more narrow-minded versus more searching manner. Individuals with more literal or fundamentalist religious cognitions tend to hold more stereotypical worldviews and avoid questioning their convictions, whereas those with more symbolic or questing cognitions are more open to challenging their worldview and adapting their attitudes and behaviours based on new information (Hunsberger et al. 1996). A questing religious cognition thus reflects ‘a tendency for people […] to think complexly both about religion and about people and diversity’ (Hunsberger and Jackson 2005, 816). Similarly, measures of symbolic religious cognitions emphasise the need to engage in personal reflection to uncover the deeper truth of religious texts, and the irrelevance of historical accuracy to the benefit of the underlying messages and guiding principles communicated through them (Fontaine et al. 2003).

Importantly, a symbolic religious cognition can be combined with high levels of religiosity, but amounts to relativism when combined with low religiosity. Relativism is diametrically opposed to religious orthodoxy (high literalism and high religiosity). While the positions of orthodoxy and relativism are familiar tropes in public debates, the positions of literal disbelief (also described as enlightenment fundamentalism) as well as symbolic religiosity are much less known. Yet they are potentially fruitful avenues to understand how immigrant religion can positively contribute to immigrant integration, just as the renouncing of religion as such can be a potential hindrance to the blurring of boundaries between (religiously different) immigrants and non-migrants. Because immigrant integration essentially requires individuals from different (religious) groups to come to terms with the diversity of their surroundings (Grunow et al. 2023; Pickel 2019), the theoretical distinction between religiosity and religious cognition as independent dimensions of religious orientations is of paramount importance. However, unlike religiosity, the religious cognitions of literal vs. symbolic reasoning have not yet been systematically related to immigrant integration.

Outside migration studies, the notion that religious cognition is important beyond individual differences in religious involvement is already more established. When the two dimensions are jointly considered, literalism-symbolism is typically found to be more predictive than religiosity of inter-group prejudice, universalist values and political party choice (Duriez et al. 2007). Similarly, dogmatic and exclusivist approaches to religion were found to be more relevant for understanding prejudice than religious affiliation and religiosity among representative samples in Germany and Switzerland (Pickel et al. 2020). Beyond Western countries predominated by Christians and non-believers with a Christian heritage, analyses of the 2006 World Values Survey reveal that across seven world religions an exclusivist understanding of religion (i.e. agreeing that one’s own religion is the only true religion) is consistently related to higher levels of distrusting religious others, whereas religiosity was not consistently related to inter-religious distrust (Kollar and Fleischmann 2022). Similarly, Kanol (2021) found that religious fundamentalism is consistently related to out-group prejudice beyond religiosity across three world religions. There is related research among migrant samples, including Muslims of Turkish origin, which shows that also in the context of migration, religious fundamentalism predicts out-group hostility beyond differences in religious identification (Koopmans 2015). Yet this research is limited both in the measures of religious attitudes, as well as the outcomes under study. Regarding religious attitudes, a focus on fundamentalism is problematic as it always presupposes a certain level of religiosity (Altemeyer and Hunsberger 2004; Moaddel and Karabenick 2021), hence the distinction from religiosity that is theoretically expected is harder to assess empirically. Moreover, it is an open empirical question whether the absence of a fundamentalist or literalist religious cognition implies the presence of a symbolic or questing orientation, which has been shown to go together with greater intergroup tolerance (Hunsberger and Jackson 2005). And while out-group prejudice is certainly important for the process of immigrant integration as an expression of the individually perceived strength of group boundaries, the singular focus on this outcome does not provide an answer to the question if and how religious orientations also affect immigrant integration in the domains of social relations, identification, participation and socio-economic positions.

The present research

The current research aims to extend existing work with a more extensive measurement of religious cognitions that capture religious literalism, symbolism and theological exclusivism in addition to individual differences in religiosity. Moreover, we purposefully include multiple integration outcomes to capture the domains of socio-economic, social and civic integration, including national identification. Our overarching hypotheses are that religious cognitions will be more strongly related to immigrant integration than religiosity (H1), and that symbolic cognitions will be more positively related to immigrant integration than literal and exclusivist cognitions (H2).

More specifically, with regard to religious cognitions, we expect that our measures form a single underlying dimension with literalism and exclusivism as one pole and symbolic reasoning as its opposite (H3). Given the relation between symbolic religious cognitions and openness to diversity, we further expect that symbolic religious cognitions will be related to lower levels of social distance towards religious others (H4a) and higher levels of intergroup contact (H4b) as most proximal outcomes that have been previously related to religious cognitions. This expectation is based on the notion that the more symbolically and inclusively immigrants reason about religion, the more willing they are to engage in contact with religious others, whereas those who reason in more literal and exclusivist ways are more likely to withdraw from, or even actively oppose, cross-religious contact and be more negative towards religious others. Pro-diversity attitudes and greater intergroup contact, in turn, are expected to benefit immigrants’ participation in and identification with European societies. This expectation is based on the mutually reinforcing relations between cross-group contacts with various aspects of socio-economic integration (e.g. Kanas et al. 2012), national identification (e.g. Leszczensky et al. 2016) and civic and political participation (e.g. De Rooij 2012). We therefore expect positive associations of the more distal outcomes level of education (H4c), socio-economic participation (H4d), national identification (H4e) and civic participation, volunteering and political participation (H4f, H4 g and H4 h) with symbolic as opposed to literal and exclusivist religious orientations.

We test these hypotheses on a sub-sample of the KONID-survey of Turkish-origin Muslims living in Germany. The focus on (prevalently Sunni) Muslims of Turkish origin in Germany is partly pragmatic: we are not aware of alternative data sources that contain an equally rich measurement of religious cognitions, religiosity and immigrant integration among a sufficiently large sample to allow the disentanglement of religious cognitions and religiosity. At the same time, this focus is also societally relevant as the divide between Muslims and non-Muslims is the most prominent fault line in the polarised public opinion landscape regarding religion and immigrant integration (e.g. Foner and Alba 2008). At the same time, the focus on a single ethnonational origin group in a single destination country limits our ability to draw conclusions on the more general theoretical arguments outlined so far, and in the worst case contributes to the reification of problematic simplifications, such as the notion that all Turkish Germans (or even all immigrants in Europe) are Muslims, that all Muslims are highly religious (or even fundamentalist) and that this is the key to understanding why immigrant integration has ostensibly failed (cf. Statham 2024). We strongly reject such reductionist interpretation of this and related research. Instead, our study aims to make a contribution to unpacking the analytical category of ‘Muslims’ by introducing measures of individual differences that have previously been understudied, in order to see whether they help us understand the complex role that religion can potentially – but does not necessarily – play in the multifaceted process of immigrant integration. We thus examine for a list of specific outcomes which jointly provide a rich account of immigrant integration, whether they can be independently explained by individual differences in religious cognitions after religiosity and socio-demographic controls have been taken into account, and we document the additional explanatory power of religious cognitions by using stepwise regression models.

Materials & methods

Participants & procedure

We draw on a sub-sample of the KONID-survey, collected in 2019 (Liedhegener et al. 2021). KONID focuses on religious and social identities and aims to explain their impact on civil society, political participation, and social cohesion. In order to gain a better understanding of religious and social identities within certain minority groups, a variety of subsamples were included in the data collection process. Among them is the oversample of Turkish Muslims in Germany that our analysis is based on. People with a Turkish background were chosen for pragmatic reasons: they can be identified from telephone listings through onomastic procedures, and they represent the largest group of Muslims in Germany (45% of the Muslim and 3,4% of the total German population) according to the 2016 micro-census as best statistical source available (Pfündel, Stichs, and Tanis 2021, 21, 42).

We rely solely on the data of the Turkish Muslim sample in Germany (N = 500).¹ A quota sample was used and 500 was adopted as target, to balance financial constraints with the comparatively massive over-quotation of the Turkish population in Germany in the KONID-survey. The data were obtained with Computer-Assisted Telephone Interviews after onomastic procedures to identify individuals with Turkish heritage. A screening question at the start of the interview established whether potential participants belonged to an Islamic denomination. Bi-lingual interviewers were used and interviewees could choose to be interviewed in German or Turkish. Because the study was non-interventional, no ethical approval was required according to German law.

The response rate to the telephone survey was low at 2.3% (based on the net sample after removing incorrect phone numbers). No successful contact could be achieved for almost half of the numbers in the net sample (45.0%), and an addition 52.3% of potential participants refused to participate (this includes persons who did not answer the screening questions). The low response rate is not unusual given the target group being hard to reach. Since there is no reliable data on the number of Turkish Muslims in Germany, it is not possible to describe any corresponding data set as representative. The most comprehensive study conducted in Germany by the Federal Office for Migration and Refugees in 2020 among 5,194 Muslims (‘Muslim Life in Germany’, Pfündel, Stichs, and Tanis 2021) achieved a response rate of 15.5% among Turkish Muslims. Despite our lower response rate, our sample is highly similar to the Turkish Muslim sample in the Muslim Life in Germany study: it is characterised by 50% women, an average age of 44 years, and 43% of Turkish Muslims with a German passport, compared to 49% women, average age of 41 years, and a proportion of Turkish Muslims with German citizenship at 45% in Muslim Life in Germany (Pfündel, Stichs, and Tanis 2021, 60, 62). Given these similarities and our research focus on associations between measures rather than point estimates, we think that we can reliably analyse the relation between religiosity, religious cognition and integration of Turkish Muslims in Germany.

Measures

Religious cognitions were assessed with four statements, with which participants could indicate their agreement on a 4-point Likert-scale (1 = do not agree at all, 4 = completely agree). The statements were ‘There is only one true religion’ (exclusivism), ‘Qur’an has to be interpreted literally’ (literalism), ‘Qur’an contains old texts that are no longer relevant for my life’ (Qur’an old) and ‘Qur’an can only be understood and interpreted in light of its historical context’ (context).

Religiosity was assessed with five items that capture religious identity, practices and beliefs. Religious identification was operationalised as importance of religion and assessed as part of a battery of social identification questions, on a 6-point scale with 1 = not important at all, and 6 = very important. Subjective religiosity was assessed on a 0–10 scale with 0  = not religious at all, and 10 = very religious. The frequency of prayer was assessed on a 1–7 scale for communal prayer (1 = never, 7 = daily) and on a 1–8 scale (1 = never, 8 = several times a day) for private prayer. Finally, the strength of belief in God (or something godly) was assessed on a 5-point scale with 1 = not at all, and 5 = strongly.

Immigrant integration was assessed with 8 indicators related to different positions of immigrants in the socio-economic domain (education, (labour market) participation), the social and political domain (intergroup contact, civic participation, volunteering, political participation) and the domain of identification and intergroup relations (sense of national belonging, social distance). Specifically, level of education was assessed with the highest level completed or currently attended and we distinguish four categories: 1  = lower secondary education or less, 2 = upper secondary general education, 3 = vocational education or training, and 4 = tertiary education. Regarding socio-economic participation, we distinguish those who are active on the labour market (working part-time or full-time in employment or self-employed) and those who are still in education from those who are unemployed, homemakers, retired, permanently sick or disabled or otherwise inactive. Intergroup contact was assessed with four items measuring the frequency of contact with Christians, non-religious persons, persons with another religion and persons with another nationality, with 1 = never and 4 = very often. Because factor analysis of the four items extracted a single factor on which all four items loaded highly, we used the average. To measure civic participation, a count variable was created assessing participation in a list of 11 different types of organisations (e.g. sports club, teacher-parent association, religious community).² Because participation in more than 7 kinds of associations was rare, values higher than 7 were merged with the maximum to create a 0–7 scale. Similarly, to assess volunteering, we counted how often participants volunteered in one or several of the 11 organisations on the list. Because volunteering in more than 3 organisations was rare, we merged values higher than 3 with the maximum to create a 0–3 scale. Political participation was coded as 1 for participants who indicated that they either have run for a political office, participated in a lawful demonstration, contacted a politician and/or signed a petition in the past, or that they are currently involved in one or more of these activities, and 0 if they indicated not to have participated or currently participate in any of these political activities. National identification was assessed by asking to what extent participants experienced that they belong in Germany (1 = do not belong at all, 4 = belong). Finally, social distance was measured on a 4-point scale with the willingness to marry (or enter a permanent relationship) with someone of a different religion (1 = yes, 4 = no).

In addition to religiosity and religious cognitions, socio-demographic controls were considered in all multivariate analyses. We contrasted female gender with male and other genders. Age was measured with 8 categories (1 = 16-18 years, to 8 = above 75 years) and treated as a continuous variable for simplicity. We control for generational status by distinguishing participants who are born in Germany (2nd or 3^rd generation) from those who are born abroad. In terms of marital status, we distinguish participants who are married (with or without children) or widowed, from those who are single or divorced (with or without children).

Finally, we examine whether our results hold equally for participants who are officially affiliated to Islam (N = 500) and those who also feel subjectively connected to their religious community (N = 366). The former was asked by letting participants select from a list if they are affiliated with a religious community. Consistent with the screening question at the start of the telephone interview, all 500 participants declared their affiliation with Islam. The more subjective self-identification was established by asking participants ‘Regardless of whether you are a member or affiliate of a church or religious community, do you feel you belong to a religious community? If so, which one?’. 366 participants chose either Sunni, Shi’a, Alevi, Ahmadiyya, Hanafi, Maliki, Schafii, Hanbali or ‘Muslim: general’ as their subjective religious identity and are jointly described here as self-identified Muslims. The remaining participants either indicated not to experience a sense of belonging to any religious group (despite being officially affiliated with Islam, N = 68), or expressed their belonging to a religious group other than Islam (e.g. Christianity, Buddhism, unspecified other religion, N = 12), or did not answer this survey question (N = 54).

Analytical strategy

We first conduct exploratory and confirmatory factor analyses in R’s lavaan package to examine whether religious cognitions can be empirically separated from religiosity, and whether literalism and symbolism constitute opposite ends of this dimension, thus testing H3. We select factor solutions based on the recommendations for model fit indices by Kline (2015). Subsequently, we use the thus established measures of religious cognitions and religiosity as predictors in a series of structural equation models with the 8 integration outcomes as dependent variables, while controlling for socio-demographic characteristics, to test H4a to H4h. Constructing latent variables to measure religiosity and religious cognitions allows us to take measurement error into account and thus increases the statistical power to estimate their relations with immigrant integration (Kline 2015). We first estimate models that relate religiosity to integration after including controls, and subsequently add the measures of religious cognitions to be able to assess how much additional variance in the dependent variable is accounted for by religious cognitions. We conclude the results section with a reflection on the overarching hypotheses H1 and H2.

Results

Descriptives

Before turning to the factor analyses, we first describe the data in terms of the patterns of means (see Table 1) and bivariate correlations (Table 2). Both sets of descriptives are shown separately for the full sample (N = 500) and the subsample of self-identified Muslims (N = 366). The differences between these results are negligible, and we therefore present our main analyses based on the full sample to increase statistical power. Yet we replicated all results with the sub-sample of self-identified Muslims only and find consistent results, though due to the lower power, not all coefficients reach conventional levels of significance in these analyses (results available upon request).

Table 1. Descriptive statistics (Mean/% (S.D.)).

	Range	Full sample (N = 500)	Self-identified Muslims (N = 366)
Religiosity
Religious identification	1–6	4.49 (1.50)	4.63 (1.42)
Subjective religiosity	1–10	6.28 (2.44)	6.40 (2.27)
Communal prayer	1–7	2.91 (1.66)	2.99 (1.67)
Private prayer	1–8	4.72 (2.61)	4.84 (2.61)
Belief in God	1–5	4.40 (1.03)	4.46 (0.98)
Religious orientations
There is only one true religion [exclusivism]	1–4	2.42 (1.25)	2.46 (1.22)
Qur’an has to be interpreted literally [literalism]	1–4	2.33 (1.06)	2.33 (1.06)
Qur’an contains old texts that are not immediately relevant for my life anymore [Qur’an old]	1–4	1.98 (0.99)	1.96 (0.98)
Qur’an can only be interpreted and understood taking into account the historical context [context]	1–4	2.48 (1.03)	2.43 (1.04)
Integration measures
Level of education	1–4	2.11 (1.09)	2.14 (1.14)
Socio-economic participation	0/1	64.0	61.2
Intergroup contact	1–4	3.14 (0.57)	3.14 (0.54)
Civic participation	0–7	1.88 (2.14)	1.90 (2.16)
Volunteering	0–3	0.68 (1.08)	0.72 (1.08)
Political participation	0/1	49.8	48.1
Sense of belonging to Germany	1–4	3.31 (0.83)	3.32 (0.82)
Social distance other religion	1–4	2.59 (1.24)	2.72 (1.22)
Controls
Female	0/1	50.7	52.6
Age categories	1–8	3.90 (1.63)	3.80 (1.62)
2nd generation	0/1	54.4	54.6
Married	0/1	49.7	50.0

Table 2. Bivariate correlations between measures of religiosity and religious orientations.

	1	2	3	4	5	6	7	8	9
1 Religious identification	1	.399	.153	.261	.387	.368	.150	−.206	−.057
2 Subjective religiosity	.400	1	.351	.485	.526	.361	.274	−.137	.017
3 Communal prayer	.213	.342	1	.387	.180	.275	.211	.004	−.003
4 Private prayer	.264	.492	.383	1	.394	.276	.133	−.127	−.041
5 Belief in God	.345	.537	.198	.377	1	.190	.092	−.176	−.017
6 Exclusivism	.331	.382	.261	.306	.211	1	.365	−.056	−.042
7 Literalism	.139	.289	.186	.183	.085	.393	1	.264	.151
8 Qur’an old	−.209	−.087	−.049	−.110	−.171	−.012	.279	1	.551
9 Context	−.072	−.003	−.025	−.053	−.045	−.018	.087	.494	1

Note: Values below the diagonal refer to the full sample (N = 500), above the diagonal to the subsample of self-identified Muslims (N = 366). Correlations in bold font are significant at p < .05.

As Table 1 shows, participants display high levels of religious identification, subjective religiosity and belief in God. Participation in communal prayer is lower than in private prayer. Regarding religious cognitions, participants tend to agree rather than disagree that there is only one true religion (26.6% of participants fully and another 19.6% rather agreed), and that the historical context needs to be taken into account when interpreting Qur’an (15.8% fully and 30.8% rather agreed). The statements that Qur’an needs to be interpreted literally (16.2% fully and 23.4% rather agreed) and that Qur’an contains old texts that are no longer relevant for today’s life (37.4% did not agree at all and another 17.4% rather disagreed) find less support.

In terms of integration outcomes and socio-demographics, the sample is mostly economically active with mean levels of education between upper secondary general education and vocational training. Intergroup contacts are rather high, as is the sense of belonging to Germany. The sample is gender balanced and slightly more than half of the participants belong to the 2nd (or higher) immigrant generation.

Religiosity and religious cognitions

When it comes to the bivariate correlations between the five indicators of religiosity and the four measures of religious cognition, we observe (see Table 2) that all measures of religiosity are positively and significantly related with each other. Religious literalism and exclusivism are also positively and significantly associated with almost all measures of religiosity, and with each other. In contrast, the view that Qur’an contains old and irrelevant texts is negatively related to religiosity.

On the other hand this statement is positively associated with the statement that the historical context is important for the interpretation of Qur’an. Notably, the more participants agree that Qur’an needs to be interpreted literally, the more likely they are to agree that it contains old and no longer relevant texts. Moreover, emphasising the role of historical context is unrelated to religious literalism, against the expectation that these items would be negatively associated as they should represent the theoretically derived opposite ends of the continuum from symbolism to literalism. In fact, of the participants who (rather or fully) agree that Qur’an needs to be interpreted literally, 60.6% also (rather or fully) agree that one always has to take the historical context in which it was written into account. Conversely, when zooming in on participants who agree that this context needs to be taken into account in interpreting the holy scripture, there is almost an even split into those who agree with a literal interpretation (49.4%) and those who disagree that Qur’an needs to be interpreted literally (50.6%). Again, findings do not differ between the full sample and the sub-sample of self-identified Muslims.

We further investigated the factor structure of the 9 religion measures by conducting an exploratory factor analysis with full information maximum likelihood as extraction method and promax rotation, which is an oblique rotation allowing factors to be correlated. This extracted a 3-factor structure, where the five measures of religiosity load on the first factor above the threshold of .3 (the item emphasising historical context also loads, but the loading is below the threshold value), the items describing the Qur’an as an old text and emphasising the historical context load on a second factor, and literalism and exclusivism constitute the third factor. This three-factor model fits the data rather well: χ²(12) = 28.81, p = 0.004; CFI = 0.981; TLI = 0.942; RMSEA = 0.053. The distinction between religiosity as a separate factor from religious cognitions confirms our expectation that these are distinct dimensions of individuals’ orientations towards religion. However, against our expectation, the four measures of religious cognition did not emerge as a single dimension in the exploratory factor analysis. We therefore also estimated a confirmatory factor analysis to test whether forcing all four measures of religious cognition onto one single dimension, while retaining the five measures of religiosity as a separate dimension, would result in lower model fit. Indeed, this 2-factor model had a poor fit to the data: χ² (24) = 98.892, p < 0.001; CFI = 0.737; TLI = 0.635; RMSEA = 0.132; SRMR = 0.096, which confirms the need to separate the four measures of religious cognition into two distinct factors. When conducting an exploratory factor analysis on these four items only, two separate factors are extracted, one containing the measures of literalism and exclusivism, the other the measure of Qur’an being old and the importance of context. This two-factor model has good fit statistics: χ² (2) = 10.79, p = 0.005; CFI = 0.971; TLI = 0.913; RMSEA = 0.094.

These findings show that we cannot conceive of literalism and exclusivism as the opposite of symbolic approaches to religion on a single underlying dimension. Rather, in this sample, literal and exclusivist approaches to religion form a separate dimension from symbolic approaches. Importantly, this means that participants who are high on literalism and exclusivism do not necessarily negate symbolic approaches to religion, and vice-versa. Based on these factor analyses, we conduct a confirmatory factor analysis to construct the latent measures of religiosity and religious cognitions to be used as predictors in structural equation models of integration outcomes. The fit of a three-factor model is acceptable (χ² (24) = 98.892, p < 0.001; CFI = 0.913; TLI = 0.870; RMSEA = 0.079; SRMR = 0.054) but some estimated variances are negative, indicating model misspecification. Further inspection of the results shows that this is due to the low loading of the second item on the latent variable symbolism, which indicates that the two measures that are used to define this latent variable are not sufficiently strongly associated with each other to construct a reliable latent variable. We therefore retain religiosity and literalism/exclusivism as latent measures in our analyses but separately add the two measures of symbolic religious cognitions as observed variables. Our structural equation models of integration thus contain four predictors, which we add in three steps: we first estimate the association between the dependent variable with religiosity (a latent variable consisting of five items), while taking control variables into account, then add the latent measure literalism/exclusivism (consisting of two items) and finally add symbolism (two observed variables termed, for brevity, ‘Qur’an old’ and ‘context’).

The relation between religiosity, religious cognitions and immigrant integration

Table 3 shows that higher levels of religiosity go together with greater social distance towards religious others. When all religious cognition measures are included, literalism/exclusivism shows a similar positive and significant association with social distance, while the two measures of symbolic reasoning are both significantly and negatively related to social distance, thus providing support for H4a. Not supporting H4b, however, we find no significant associations of participants’ level of religiosity and any type of reasoning about religion with levels of intergroup contact.

Table 3. Structural equation model: Prediction of social distance and intergroup contact.

	Social distance			Intergroup contact
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Constant	2.188 (.280) ***	2.162 (.279) ***	2.993 (.325) ***	3.355 (.120) ***	3.370 (.120) ***	3.316 (.144) ***
Religiosity	3.137 (.526) ***	2.662 (.664) ***	1.962 (.676) **	−.186 (.200) n.s.	0.135 (.305) n.s.	.099 (.312) n.s.
Literalism/Exclusivism		.116 (.104) n.s.	.200 (.109) °		−.080 (.054) n.s.	−.075 (.055) n.s.
Qur’an old			−.220 (.072) **			−.025 (.032) n.s.
Context			−.119 (.070) °			.041 (.030) n.s.
Controls included	x	x	X	x	x	X
R^2	.215	.221	.254	.073	.084	.087
χ2 (df)	157.127 (59)	210.514 (88)	284.222 (102)	135.592 (59)	187.672 (88)	270.800 (102)
CFI	.844	.852	.799	.866	.870	.798
RMSEA	.058	.053	.060	.051	.048	.058
SRMR	.047	.046	.051	.043	.044	.049
N	500	500	500	500	500	500

Note: Cells show unstandardized regression coefficients and their standard errors between parentheses. *** = p < .001, ** = p < .01, * = p < .05, ° = p < .10, n.s. = p > = .10. Control variables included are gender, age, migrant generation and marital status.

When it comes to the more distal integration outcomes, Table 4 shows that there is a significant but weak negative association between participants’ level of religiosity and their level of education, but religious cognitions are unrelated to educational attainment, refuting H4c. Regarding participation in education or on the labour market, we initially observe a similar negative association with levels of religiosity, which however turns non-significant when literalism/exclusivism is added. Symbolic religion does not relate significantly to participation. Thus, those who subscribe to literal and exclusivist interpretations of religion less frequently participate in work or schooling, but individual differences in religiosity and symbolic interpretations do not make a difference, which provides only partial support for H4d. Regarding national identification, Table 4 reveals that a sense of belonging to Germany is particularly negatively related to literal and exclusivist understandings of religion, partly supporting H4e, but also negatively predicted by the assertion that Qur’an contains old texts that are no longer relevant, which runs against our hypothesis. When all measures of religion are included, levels of religiosity and an emphasis on the importance of context when interpreting Qur’an are unrelated to national identification.

Table 4. Structural equation model: Prediction of level of education, socio-economic participation and sense of belonging to Germany

	Level of education			Socio-economic participation			Sense of belonging
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Constant	1.136 (.213) ***	1.140 (.213) ***	1.061 (.257) ***	.312 (.088) ***	.322 (.087) ***	.322 (.087) ***	3.357 (.172) ***	3.410 (.166) ***	3.758 (.197) ***
Religiosity	−.918 (.384) *	−.902 (.555) n.s.	−.930 (.569) n.s.	−.319 (.153) *	−.005 (.229) n.s.	.057 (.240) n.s.	−.541 (.303) °	1.213(.524) *	.804 (.499) n.s.
Literalism/Exclusivism		−.000 (.093) n.s.	.004 (.094) n.s.		−.076 (.040) °	−.085 (.043) *		−.433 (.103) ***	−.363 (.094) ***
Qur’an old			−.024 (.059) n.s.			.016 (.024) n.s.			−.115 (.045) *
Context			.049 (.054) n.s.			.022 (.022) n.s.			−.032 (0.42) n.s.
Controls included	X	X	X	X	X	X	X	X	X
R^2	.141	.140	.142	.237	.249	.257	.049	.197	.193
χ2 (df)	136.942 (59)	190.453 (88)	273.332 (102)	134.605 (59)	187.207 (88)	269.555 (102)	143.419 (59)	191.287 (88)	272.403 (102)
CFI	.872	.872	.803	.887	.886	.821	.851	.869	.803
RMSEA	.051	.048	.058	.051	.047	.057	.053	.048	.058
SRMR	.043	.044	.049	.044	.045	.050	.044	.044	.049
N	500	500	500	500	500	500	500	500	500

Note: Cells show unstandardized path coefficients and their standard errors between parentheses. *** = p < .001, ** = p < .01, * = p < .05, ° = p < .10, n.s. = p > = .10. Control variables included are gender, age (in categories), migrant generation and marital status.

Finally, for the domain of civic and political engagement, we find partial support for H4f, H4g and H4h as our measures of symbolic religion are positively associated with civic and political participation and volunteering. Specifically, the more participants agree that Qur’an contains old texts that are no longer relevant, the more they participate in civic associations and political acts, and the more participants agree that the historical context needs to be taken into account when interpreting Qur’an, the more they volunteer for civic organisations. Individual differences in literalist and exclusivist interpretations of religion are unrelated to civic and political participation and volunteering, but more religious participants are less likely to participate politically, also after religious cognitions are taken into account (see Table 5).

Table 5. Structural equation model: Prediction of civic participation, volunteering and political participation.

	Civic participation			Volunteering			Political participation
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Constant	2.212 (.439) ***	2.184 (.439) ***	1.349 (.520) **	.616 (.221) *	.610 (.221) *	.180 (.262) n.s.	.342 (.102) **	.338 (.102) **	.315 (.121) **
Religiosity	.033 (.757) n.s.	−.838 (1.162) n.s.	.008 (1.151) n.s.	.322 (.384) n.s.	.182 (.565) n.s.	.346 (.575) n.s.	−.446 (.179) **	−.653 (.263) *	−.536(.266) *
Literalism/Exclusivism		.228 (.209) n.s.	.104 (.196) n.s.		.036 (.096) n.s.	.013 (.096) n.s.		.049 (.044) n.s.	.033 (.044) n.s.
Qur’an old			.314 (.120) **			.032 (.060) n.s.			.061 (.028) *
Context			.050 (.112) n.s.			.133 (.056) *			−.041 (.026) n.s.
Controls included	X	X	X	X	X	X	X	X	X
R^2	.038	.045	.065	.036	.038	.058	.052	.059	.064
χ2 (df)	148.783 (59)	203.372 (88)	283.281 (102)	145.275 (59)	197.862 (88)	280.522 (102)	140.007 (59)	193.573 (88)	277.903 (102)
CFI	.842	.849	.783	.847	.854	.785	.857	.861	.789
RMSEA	.055	.051	.060	.054	.050	.059	.052	.049	.059
SRMR	.045	.045	.050	.044	.045	.050	.044	.044	.049
N	500	500	500	500	500	500	500	500	500

Note: Cells show unstandardized regression coefficients and their standard errors between parentheses. *** = p < .001, ** = p < .01, * = p < .05, ° = p < .10, n.s. = p > = .10. Control variables included are gender, age (in categories), migrant generation and marital status.

Having thus tested our specific hypotheses, we now consider our overarching hypotheses that religious cognitions will be more strongly related to immigrant integration than religiosity (H1), and that symbolic cognitions will be more positively related to immigrant integration than literal and exclusivist cognitions (H2). Providing partial support for H1, religiosity is unrelated to five out of the eight measures of immigrant integration. Except for intergroup contact, which is also unrelated to religious cognitions, for the other outcomes where individual differences in religiosity do not matter, the way of reasoning about religion does make a difference. Statistically speaking, the share of additional variance in the outcomes under study explained by religious cognitions, after having taken religiosity (and controls) into account, is in the range of 1.5–5%, with a peak of about 15% for national identification. Regarding levels of education, however, religious cognitions do not explain additional variation beyond religiosity. The latter finding runs against H1, but educational attainment is the only of the eight outcomes under study where religious cognitions do not add to the explanatory power.

Also, with regard to H2, our findings provide only partial support. Our expectation is most clearly supported in the civic and political domain, where symbolic approaches to religion relate to greater engagement, whereas exclusivist and literalist cognitions are unrelated and religiosity is even negatively related to political participation. The finding that more religious participants and those who are higher in literalism and exclusivism feel a lower sense of belonging to Germany also supports our assertion that this particular way of being religious is at odds with integration in the domain of identification, though here we do not find a positive relation with symbolic religious cognitions, again only providing partial support for our expectation. Finally, with regard to the socio-economic and social domains, neither religiosity nor religious cognitions play a particularly influential role. While the findings from the eight structural equation models thus provide only partial support for our overarching hypotheses, none of the findings directly contradicts these expectations, as we do not find negative associations of symbolic religious cognitions, or positive associations of literal and exclusivist cognitions, with integration. Moreover, with the exception of levels of education and intergroup contact, adding religious cognitions to the analyses helps us to explain differences in integration between individual migrants beyond the extent of their religious involvement.

Discussion and conclusion

There is persistent societal and scholarly interest in the role of religion for immigrant integration in general (e.g. Kogan, Fong, and Reitz 2020), and of Islamic religion among migrants to Europe in particular (e.g Foner and Alba 2008), but previous survey research relating levels of religiosity to integration outcomes did not yield consistent results (Fleischmann 2022). Integrating this line of research with insights from the psychology of religion (Wulff 1997), this study sought to extend the conceptualisation of immigrant religion with different ways of reasoning about religion to provide a more nuanced understanding of the meaning of religion for immigrants and how it can potentially affect immigrant integration. To this purpose, we examined four statements that captured literalist, exclusivist and symbolic religious cognitions in addition to immigrants’ religiosity in a sample of Turkish-origin Muslims in Germany, and related them to eight distinct outcomes that gauge socio-economic, social and civic domains of integration.

In line with the notion that religious cognition is conceptually distinct from religiosity, agreement with these statements loaded on different factors than measures of religiosity, and religious cognitions had additional explanatory power for several dimensions of immigrant integration. Interestingly, and departing from previous research in the psychology of religion (e.g. Duriez et al. 2007), however, literalism and symbolism did not emerge as opposite ends of a single dimension. Instead, agreeing that Qur’an should be interpreted literally and that there is only one true religion turned out to be related in this sample to the dimension of religiosity. Importantly, this did not imply the negation of a symbolic approach as those who agreed more with the statement that Qur’an needs to be interpreted literally also agreed more that the historical context is important in this interpretation. These findings imply that a focus on literalism and exclusivism – as well as the related concept of religious fundamentalism (cf. Altemeyer and Hunsberger 2004; Moaddel and Karabenick 2021) – might be too limited in operationalising different religious cognitions, at least among Sunni Muslim migrants.

Given the Islamic doctrine of Qur’an being the literal word of God, and Islam emerging as historical successor of the preceding monotheistic religions of Christianity and Judaism, these specific measures of literalism and exclusivism might tap into forms of religiosity that are considered normative among this religious community, rather than capturing the individual differences in meaning-making they have found to underpin in Christian (heritage) samples. Operationalising symbolic approaches to religion thus seems to be particularly important to better understand how religion can benefit immigrant integration. This recommendation for future research rests particularly on our finding that symbolic cognitions were positively related to participation in civic associations and volunteering, which suggests that a more contextual or interpretative approach to religion has the potential to provide the ‘social glue’ that is needed in post-migration diverse societies (Putnam and Campbell 2010). Importantly, this positive association emerged after individual differences in religiosity were taken into account which did not – except for political participation – relate significantly to civic integration. This means that at the same levels of religious involvement, those who take a more symbolic approach are more civically engaged than those who score lower on symbolism. It is important to emphasise here again that symbolism and literalism/exclusivism were not inversely related, meaning that a literal approach to religion and an exclusivist understanding of it do not stand in the way of a more symbolic understanding of religion, and the civic participation that goes along with that. While the investigation of symbolic religion thus seems promising for future research on immigrant integration both based on theoretical reasons as well as the present findings, it should also be acknowledged that it is challenging to operationalise symbolic religion (as distinct from both literal religion and symbolic non-religion, cf. Fontaine et al. 2003). Moreover, our study is limited to two observed items, rendering it susceptible to measurement error. Future research should therefore develop more extensive – and ideally cross-culturally comparable – measures of symbolic religious reasoning to provide better empirical proof of the relevance of this religious cognition for immigrant integration.

Our multidimensional approach to the concept of immigrant integration also allows us to provide more insights into the domains where immigrant religion matters, and then to specify which aspect of it matters. In line with previous findings showing inconsistent and small associations of immigrant religion with socio-economic integration (e.g. Carol and Schulz 2018), we found levels of religiosity and religious cognitions to be largely unrelated to differences in educational attainment and participation in education or on the labour market. We also found religion to be unrelated to levels of intergroup contact and the sense of belonging to Germany, which indicates that for many domains of immigrant integration, religion plays a minor role, also when going beyond counting practice frequencies or the subjective importance of religion. Adding our evidence to previous conflicting results in these domains suggests that scholars who are interested in explaining intergroup contact and a sense of belonging to migrants’ receiving societies, or practitioners who strive to foster these outcomes, are advised to look for other explanations and de-emphasise the role of religion. When it comes to intergroup prejudice, our regression results of social distance align with those from previous research (e.g. Pickel et al. 2020) showing that those with more exclusivist and literalist orientations are more negative towards religious others, whereas those with more symbolic cognitions are more open, but apparently these associations do not translate into downstream relations between religious orientations and realised contacts or identification with the wider society.

Strengths & limitations

This study was a first attempt to operationalise different ways of meaning-making among immigrants adhering to the same non-Christian minority religion, while taking into account different levels of religiosity. The four statements we incorporated provide a more extensive measurement of the religious cognition concept than previous research could offer, while allowing the differentiation of religious cognitions from religiosity, and assessing their relations empirically rather than imposing a factor structure based on theory alone. The latter is particularly problematic because most research on religion and intergroup relations has been conducted in Christian contexts and the generalisability of these findings as well as underlying theoretical arguments should be critically examined. While this study constitutes a step forward in this cross-cultural comparative research agenda, the measurement of religious cognitions remains limited to four statements with varying levels of complexity. We cannot be sure that participants understood these items in the way that theorists of religious cognition intended. This is particularly true for symbolic approaches to religion, where the two available measures did not correlate sufficiently strongly to form a well-functioning latent variable. It will be important in future research to provide a more comprehensive, clear and cross-culturally sensitive measurement of symbolic reasoning in order to further examine the notion that symbolic ways of reasoning about religion can facilitate immigrant integration.

A second limitation that needs to be addressed is the specific sample that was available for this research. Ideally, to test the broader theoretical notion that religious cognition is important for immigrant integration beyond religiosity, we would want to include a more diverse sample of Muslim minorities, as well as including immigrant-origin groups with other religious affiliations. However, survey data containing measures of both religiosity and religious cognition among migrant-origin samples are rare, and the oversample of Turkish-origin Muslims in the KONID-survey is a first opportunity to examine our research question empirically. Due to these practical constraints, our results only apply to a single origin group in a single destination. This prevents us from examining the role of ethnonational origins and receiving society contexts for the complex relation of immigrant religion and integration. Moreover, the focus on religion among one of the largest origin groups in one of the largest destination countries in Europe runs the risk of contributing to the reification of the simplistic notion that all immigrants are Muslims. While the available data do not allow us to draw on comparisons across origins and destinations to unpack this simplistic notion and increase complexity, similar to the empirical strategy of Statham (2024), our analysis has a similar aim in shedding light on individual differences within a religious community that is under much public and scholarly scrutiny and often unduly homogenised. By differentiating distinct ways of reasoning about religion, as well as distinguishing religious cognition from religiosity, we aim to show that there are different ways of being Muslim in Europe, with distinct repercussions for the boundary-blurring process that we call immigrant integration.

A final limitation worth pointing out is the cross-sectional nature of our data which prevents us from drawing conclusions about the direction of causality connecting immigrants’ religion and integration. In addition to building more comprehensive and cross-culturally comparable measures of religious cognition, future research should thus investigate the stability of these cognitions over time and assess whether they are affected by, and/or causally affect immigrant integration.

Conclusion

This study showed that differences in reasoning about religion can help understand integration among Turkish-origin Muslims in Germany. Literalist, exclusivist and symbolic understandings of religion add to the explanatory power of integration outcomes beyond the quantitative involvement with religion. This is particularly true in the domain of civic engagement but less so in the domain of structural integration and intergroup contact. Importantly, while literal and exclusive orientations to religion were closely associated with religiosity in this sample, and often related similarly to the broad range of integration outcomes under study, symbolic approaches to religion differed most clearly from these findings and were more positively related to immigrant integration, particularly in the civic domain. This suggests that religion might act as a bridge rather than a barrier to immigrant integration, regardless of migrants’ levels of religiosity, if religious orientations leave room for the role of historical context in interpreting religious scripture and caution against an unreflective application of centuries-old religious texts to contemporary issues.

Acknowledgments

This research was supported by a grant from the German Research Foundation (DFG) and the Swiss National Science Foundation (SNSF) for the project ‘Configurations of Individual and Collective Religious Identities and their Potential for Civil Society (KONID)’, awarded to Antonius Liedhegener and the fourth author, and by a grant from the Dutch Research Council (NWO VI.Vidi.211.125) for the project ‘The role of religious cognition for immigrant integration (RECOGNITION)’ to the first author.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Nederlandse Organisatie voor Wetenschappelijk Onderzoek [grant number NWO VI.Vidi.211.125].

Notes

1 We exclude the Muslim samples of the Swiss KONID-survey because Turkish Muslims only make up a small minority of the Muslim population in Switzerland, where the population of migrants consists of significantly different origin groups.

2 Including religious organisations in the construction of measures of civic participation might induce (some) participants to interpret these questions as measures of communal prayer or other religious activities taking place in religious community centres such as mosques. We therefore checked if our results are affected when excluding religious organisations from the measurement of civic participation and volunteering and found no differences (details available upon request).