Understanding the religious behaviour of Muslims in the Netherlands and the UK

1 Understanding the religious behaviour of Muslims in the Netherlands and the UK Dr Ayşe Güveli (Department of Sociology, University of Essex) Dr Lucinda Platt* (Insitut d’Anàlisi Econòmica, Barcelona and Institute for Social and Economic Research, University of Essex) Abstract The position of Muslims in Western societies is the subject of intense study and debate. However, remarkably little attention has been paid to the practice of European Muslims and how Muslim religiosity relates to conventional measures of social and economic integration. In this paper we draw on theories of secularization, assimilation, revitalization and integration to explore the correlates of attendance at religious meetings for Muslims of different backgrounds in the Netherlands and the UK. We conclude that patterns of religiosity and secularisation cannot be generalised across national contexts. Keywords: Religion, Muslim, Religiosity, Immigrant, Netherlands, UK *Platt undertook the work on this paper while she was a Marie Curie Fellow linked to the INSIDE at the IAE - CSIC Barcelona. She is grateful to the European Commission for funding the Fellowship and to the IAE and its director Clara Ponsati for their hospitality and for providing a supportive research environment. 2 1. Introduction The position of Muslims in Western societies is the subject of intense study and debate (Modood 1998). In particular, multiculturalism as both policy and concept has come under attack (Joppke 2004), though there have been attempts to defend – and reformulate – it (Parekh 2000; Modood 2007). At the same time there is increasing recognition of the economic disadvantage experienced by Muslims in many European countries (Ansari 2002; Platt 2007; Vasta 2007). However, while Islam is regarded as an obstacle to incorporation within European societies, little attention has been paid to the religious practice of European Muslim immigrants and their descendants and to how Muslim religiosity relates to conventional measures of social and economic integration. Phalet and Haker 2004; and Maliepaard et al. 2010 provide exceptions for the Netherlands. This contrasts with the strong US tradition of analysing religious practice as a route to integration for immigrants and their children (Foner and Alba 2008). Current debates about Muslims in Europe are typically underpinned – implicitly or explicitly – by a universalising frame of reference that assumes transnational continuity in religious beliefs and practices, and their correlates. However, the extent to which such assumptions are valid and are a persistent phenomenon of Muslim identity and practice requires interrogation in more than one national context and ideally at periods both prior to and subsequent to the recent debates surrounding multiculturalism. In this paper, therefore, we investigate Muslim religiosity in the 1990s, exploring similarities and differences in the correlates of religiosity in different national contexts, and in the period before 2001 and the enhanced politicisation of the discourse surround Islam that has marked the 21st century. Specifically, we investigate attendance at religious 3 meetings in two increasingly secular societies, the UK and the Netherlands. These two countries are both comparable and distinct and allow for investigation of differentiation (Foner and Alba 2008; Kelly and De Graaf 1997). By detailed case study comparison we can explore which secularising processes impact different Muslim populations, and whether differences in national context or national origin of Muslims remain pertinent to a consideration of Muslim religiosity. We draw on theories of secularization, assimilation, revitalization and integration, to identify potential correlates of religiosity. 2. Background and key concepts We frame our discussion and analysis in relation to theories of secularization (Durkheim 1995 [1912]; Berger 1967; Norris and Inglehart 2004), assimilation (Park and Burgess 1969 [1921]; Gordon 1964) revitalization (Laurence and Vaisse 2006; Statham 2004; Jacobsen 1997), and integration (Durkheim 1963 [1897]) and identity (Tajfel 1981). We use the term revitalisation to refer to both the institutionalisation of religious practice within country of migration, through, for example the construction of mosques and the provision of religious education, and to re-Islamization (Laurence and Vaisse 2006), or the assertion of overarching transnational religious identities, often replacing strong ethnic affiliation among the second generation of minorities (Alam and Husband 2006). Between them, these theories imply a role for educational and economic attainment, length of time in the destination country, generation, majority exposure and identification for the intensity of religious practice in highly secular societies. 4 Secularization: the role of education and economic security The founders of sociology claimed that not only the belief in religion among people but also the central role of religious institutions would decrease as economic security increased and the state took on the functions of religious institutions (Durkheim 1995 [1912]). At the same time, standard theories based on enlightenment perspectives have suggested that with education people learn to see religions as forms of superstition (Weber 1930 [1904]; Berger 1967). This view of the gradual disappearance of religion through security and education has dominated the sociological debate on religion in the 20th century. The tenets of secularism have more recently been subject to re-evaluation (Smith 2008), but it is clear that affiliation and traditional forms of observance are on the decline in Western European countries. We might anticipate, therefore, that both the acquisition of qualifications and economic security would lead towards a decline in religiosity for Muslim minorities faced with life in a secular society, as Kelly and De Graaf 1997 have found. However, it is helpful to distinguish education as personal learning or exposure to ideas, from educational contexts as providing contact or exposure to secular individuals. Education as a form of contact with non-religious, secular individuals or those of another religious affiliation may challenge beliefs or lead to convergence. General integration theory proposes that the more people integrate in social groups the more likely they are to follow the rules of these groups (Durkheim 1963 [1897]). In educational institutions, such ‘integration’ depends on whether individuals are enabled to foster their religious identity, and the extent to which that is congruent with academic success. 5 Participation in schooling among the second generation is a reflection of such potential contact, but may vary according to the national structure of educational institutions. Education can also lead to greater understanding and intellectual commitment to one’s religion. Contrary to stereotypes, education may not be regarded as a threat to faith: for example there has been a rapid expansion in the number of women from Muslim families in the UK entering higher education (Georgiadis and Manning forthcoming; Hussain and Bagguley 2007); and there is substantial evidence of the positive relationship between religious observance and educational attainment (Lehrer 2008) as well as on the increased income associated with religiosity in high income countries (Bettendorf and Dijkgraaf 2007; Portes and Rumbaut 2006). Even as secularization would suggest a negative relationship between qualifications and religiosity, the story may be more complex. Related to the concept of education as insight leading to rejection of religious attachment, Muslims with good host country language proficiency are more likely to come in contact with secular worldviews in the destination society. Indeed, host language fluency may both be a precondition for adopting a more secular outlook and, conversely, may be less necessary to those who are embedded in ethnically specific religious practices and institutions. Socio-economic position also may impact on religiosity both via social contact as well as via the security it offers. Labour market participation typically involves contact with majority society members, and has also been seen as representing a measure of engagement with or integration into the host society. According to Logan and Alba (1993), labour market integration is likely to lead to similar levels of secularisation to 6 non-immigrant counterparts. Conversely, the low participation rates of Muslim women have been related to religious norms and proscriptions, though there is some evidence of change across generations (Dale et al. 2006; Georgiadis and Manning forthcoming). However, such a notion of the ‘well-integrated’ being less secular is open to contestation. Ethnic attachment does not necessarily decline with resources; in fact greater economic assimilation will increase the relative power and position of minority groups and allow them more choices, including the choice to associate with those felt to be similar (Dorsett 1998). On the other hand, religion and religious institutions can act as a ‘resource’ in times of difficulty and therefore may be most important to those who are not in work and with lower levels of income. Assimilation and re-Islamization: religious expression over time and in the second generation For assimilation theories time is a crucial factor: as immigrants spend longer in a country, they are assumed to be more likely to take on board the values, norms and expectations of their country of residence (Gordon 1964). This would suggest that Muslims entering a secular country would be more likely to become secular themselves over time and across generations. However, the functions of religious participation, in terms of providing meaning in adversity and community support may in fact become more important over time, and the passage of time can allow greater institutionalisation of religious practices and settings. There is also increasing recognition that for second and subsequent generations, Islam can offer a meaningful source of identification for those at odds 7 between the homeland identification of their parents and their exclusion within their country of birth (Statham 2004, Alam and Husband 2006). Moreover, it is important to recognise that even apparently secular Western states are often imbued with specific religious symbolism, assumptions and privileges (Foner and Alba 2008). Convergence therefore represents not the gradual abandonment of a particular belief system, but by implication the tacit acceptance of an alternative one. In the face of such a ‘choice’ more powerful adherence to religion of birth and upbringing over time and into the second generation starts to seem more likely. Integration: religion as a resource and intergroup relations Religious identification can be regarded as a resource, where identity is used to pursue interests (Barth 2010). At the same time ascription to particular identity categories can inform behaviour and group attachment (Tajfel 1981). While there have been challenges to the concept of 'identity' and the meaning of 'ethnic group' within social science (Brubaker and Cooper 2000, Carter and Fenton 2010), identification and sense of ingroup belonging is recognised as being both consequential for well-being (Thoits 1983) and contributing to understanding of inter-group relations, however loosely the term group is conceived (Jenkins 2008). While much of the discussion of identity has focused on ethnic identification, the pertinence of religious identities is increasingly being recognised as significant for our understanding of social identity (Ysseldyk et al 2010) and its consequences (Modood 2010), as well as inter-group relations (Verkuyten 2005). At the same time, the nature of overlapping social identities (Roccas and Brewer 2002), 8 in particular ethnic and religious identities, are drawing increasing empirical and conceptual attention in Western Europe. Religious identification and accordance with norms associated with that identity can therefore both stem from patterns of social contact and engagement and contribute to shaping them. Thus we would expect to see greater religiosity paralleled by both greater in-group identification and by other demonstrations of group related behaviour, such as greater in group and less out group contact. This latter also accords with the more general expression of integration theory (Durkheim 1963 [1897]), whereby greater in-group contact is likely to be associated with greater religiosity and greater out-group contact to be associated with lower religiosity. Moreover, ethnically based association and clubs build community resources and create opportunities for integration (Putnam, 1995; Portes, Fernández-Kelly and Haller 2005; Zhou and Xiong 2005). Their function as a resource can be particularly important in less receptive contexts (Connor 2010). Such associations often have their basis in organised religious activities and therefore involvement in them implies greater levels of attendance at religious meetings. Conversely, involvement in interethnic associations may offer a substitute to association based on religious practice. Social identities conceived in terms of interest, following Barth (1969), may thus link patterns of association with religiosity. Finally, the geographical concentration of groups has been linked to group-related preferences or the enforcement of group norms. Exclusion and self-separation are debated in literatures on ethnic enclaves (Borjas 1998; Clark and Drinkwater 2002; Cutler and Glaeser 1997), ‘oppositional identities’ (Battu et al. 2005), and the marginalization of 9 minorities in general and Muslims in particular (Ansari 2002; Platt 2007; Vasta 2007). Whether the emphasis is on allocation or on preferences, neighbourhood concentration is likely to be consequential for behaviour, with concentrations of residents of the same minority faith or religious identity associated with greater religious participation. Furthermore, people tend to establish their religious institutions and social contacts in places where their religious groups are concentrated (Ebaugh, O’Brien and Chafetz 2000), facilitating access in such areas. Ethnic concentration seems to be more powerful than specifically religious concentration in fostering shared association (Peach 2006); but given the substantial overlap between religious and ethnic identities for both the UK’s and the Netherlands’ largest Muslim minorities, the level of concentration of Muslim minorities’ own ethnic group can be expected to play a part in reinforcing Muslim religious involvement (Van Tubergen 2007). The importance of context – origin and destination factors The majority of Muslims in Europe are either immigrants or the children of immigrants who have typically emigrated from countries where Islam is the majority religion and where participation in religious activities is higher than in the destination countries. However, within this general pattern there is great diversity in terms of both origin societies and destination countries. We therefore explore two broadly comparable but nevertheless distinct ‘destination’ contexts and examine the contribution of ethnic origins. These ethnicities reflect complex histories of migration, and settlement as well as diverse historical and cultural heritages with different religious interpretations and practices. Muslims in Britain 16 weekly. For robustness we also estimated models on men only. Although smaller sample sizes rendered it harder to attain statistical significance, the results were largely consistent in size and direction with those presented here from the pooled model. (Tables available from authors on request.) In cross-sectional surveys, it is problematic to disentangle cohort, period and age effects. In fact, those studies which have been able to separate the effects of age from those of cohort do not find aging influences religiosity (Tilley 2003; Crockett and Voas 2006; Norris and Inglehart 2004). We control for age as an important demographic characteristic, but recognise that we cannot distinguish it from cohort; and it is potentially confounded with duration. Method We estimated logistic regression models. We estimated six models for each country to separately identify the different sets of potential associations, including a basic model which includes only sex, age and partnership status, and a seventh model that includes the full set of independent variables. 4. Results The Netherlands The first model of Table I is our baseline model which includes basic demographic data on sex, age and partnership status. Women are less likely to attend regularly than men. This is a specific feature of Muslim public religious participation (see also Van Tubergen 2006). Those with a partner are also more likely to attend, consistent with existing 17 research (Becker and Hofmeister 2001). Age is associated with increased attendance. We explored whether there was any evidence for a non-linear relationship between age and attendance, but found no support for such a relationship. All three basic demographic factors persist across all models including other variables and in the full model. In Models 2 and 3, we estimate the relationship between educational and economic indicators and religious attendance, thus testing hypotheses relating to secularization and security. Being educated is associated with a lower propensity to attend, in line with the standard secularisation thesis, and there is some indication that this effect increases with educational level. Religious attendance among Muslims with good language skills is significantly lower than among those with poor Dutch language skills, according with the hypothesis that competence in Dutch is a precondition for absorbing a more secular worldview and conversely more necessary for those who are less religious, who will have more incentive to draw on alternative forms of social support and contact. Consistent with theories relating to both security and contact, and the potential of religious institutions as a resource, Muslims with a paid job attend religious meetings less often than unemployed Muslims (Model 3). All these effects persist in the final model. [Table I about here] Our discussion indicated that longer duration would be associated with assimilative processes and therefore lower attendance, especially for those socialised from an early age into Dutch society. However, we also discussed the view that revitalization and the 18 substitution of religious for more narrowly ‘ethnic’ identities could be associated with higher levels of attendance among the second generation. In fact, in Model 4, we find no support for either assimilation over time or for a second generation effect, once we have taken age into account. Younger minority group Muslims are less likely to attend, but that is the case regardless of generation. Being born or brought up in the Netherlands, over and above its potential for increasing probability of Dutch fluency, does not of itself enhance the tendency to assimilate to secular norms. Interacting ethnic group with duration (not illustrated) we found that there is a slight, but significant, tendency for Moroccan attendance to increase with duration, once we have controlled for age, suggesting that those groups with fewer infrastructural resources require time to establish the institutional arrangements to enable attendance. For Turks the probability of attendance remained constant with duration. Therefore even for particular groups, there is no evidence of assimilation; though it may be our cross-sectional data disguises cohort differences with earlier, more religious cohorts converging to the levels of later less religious ones. Without longitudinal data, this must remain at the level of conjecture, however. Forms of contact with the majority and identification with the majority were anticipated to reduce Muslim religiosity; and such integrative effects are partly found in the fifth model. It is worth noting the strong negative correlation between Dutch identification and attendance. This would appear to suggest that Muslim piety and Dutchness are in some ways at odds, as is often posited in current debates. We return to this issue when exploring the UK results, below. Interestingly, however, there is no evidence that living in a neighbourhood with a higher concentration of own group is 19 associated with greater religious attendance. It is worth noting that, unlike previous studies, we have used a measure of own group in the locality rather than simply the concentration of “non-Western” individuals. Our measure is clearly more relevant for hypotheses relating to group norms and contact. For the Netherlands at least, Muslim religiosity would not appear to be structured by local opportunities or group norms. Finally, origin effects partly accord with our expectations deriving from Norris and Inglehart’s classification of societies (Model 6), since Surinamese and Antilleans who come from countries with minority Muslim populations attend less frequently. But Turks, despite coming from an industrial country, attend significantly more frequently than the reference category of Moroccans. We speculated that there were two possible reasons for this apparently surprising outcome. First, the greater capacity among Turks to develop group level institutions and resources may encourage and support attendance Canatan (2001). This is supported by analysis exploring the interaction of duration with ethnicity, discussed above, which shows that Moroccan attendance increases with duration, whereas Turkish attendance rates remain largely constant. Second, and potentially related to the issue of institutional structures, we identified that greater attendance among Turks was largely driven by differences between Turkish and Moroccan women, with insignificant differences between Turkish and Moroccan men. The fact that Turks are, contrary to our hypothesis, no less likely to attend, may accord with the institutional infrastructure, while the greater attendance of Turkish women may be influenced both by infrastructure which facilitates their attendance and by origin country differences in gendered patterns of attendance that work in the opposite direction 20 to the industrialisation thesis. Future research with more origin countries and more extended measures of institutional provision may help to illuminate this question. The United Kingdom Table II shows that the UK reveals a rather different set of associations. While the smaller sample size reduces the chances of achieving statistical significance in some cases, an inspection of the sign and size of the coefficients show that there are differences between the two countries. Model 1 shows that, as with the Netherlands, being a woman is negatively associated with attendance, while age is positively associated and being partnered is positively though not significantly associated. In Model 2, we see that education is not clearly associated with attendance, as it was for Dutch Muslims. Closer inspection shows that part of any education effect comes through greater English language fluency (and thereby also partly by generation). The sign for some of the education dummies is in the expected direction and further aggregation could produce some significant differences in the reduced model, but overall, the secularising power of education for UK Muslims, over and above these exposure effects does not seem clear cut as it does for Dutch Muslims. Fluency in English is also associated with lower attendance, as Dutch fluency was for the Netherlands. This again emphsises the ways in which competence in English may be a precondition for absorbing a more secular world view and also be more necessary for those who are less religious. Economic security as measured by employment (Model 3) is again negatively but not significantly associated with attendance, and this association is weakened further in the full model, thus providing only limited support the relevance of security in the secularising process. Overall, then 21 the UK does not provide strong support for process of secularisation operating among British Muslims in the ways they do for Dutch Muslims. [Table II about here] Similarly, assimilation in terms of duration of stay (Model 4) is not associated with either greater or lesser religiosity, once age is taken into account; and in this model the age coefficient itself becomes non-significant. However, being second generation is highly significantly related to decreased attendance. Unlike in the Dutch data, where being second generation was only relevant in so far as it was associated with youth, either via an age or perhaps more probably a cohort effect (Crockett and Voas 2006), in the UK second generation and the implied greater assimilation to more secular norms stemming from that is significant in its own right. This indicates that religiosity is sensitive to local conditions and exposure and that is also supported by the integration model. Being active in an own group association is not significant in the simple model, but it becomes so in the final model when we control for other relevant factors. Test also confirm that participation in an own ethnic group organisation differences significantly from the effect of belonging to a majority organisation, relative to no organisation. The association between ethnicity and association may well imply reverse causation since religious participation and attendance facilitates organisational involvement (Ruiter and de Graaf 2006), but could also relate to mutually reinforcing group norms. Unlike in the Netherlands, the increased group norms or resources implied by own group area concentration have a positive relationship with 22 religiosity, and this association persists in the final model. Again we may be seeing reverse causation, in that religious people may choose to co-reside in places where they have access to religious institutions. More important, for our purposes, however, is the fact that this association is found here but not in the Netherlands, despite – or perhaps because of – the high degree of ethnic group concentration in the Netherlands. A particularly striking finding for the UK is the positive relation between selfidentification with British society and religious attendance, which remains in the full model. This is at odds with our expectations relating to integration and contrasts with the Dutch results. It is perhaps particularly worthy of note given that in the UK, by the date of our data, the Rushdie affair had already politicised perceptions of British Muslims and reshaped debates on ethnicity and integration within the UK (Ahmad and Evergeti 2010; Modood 2009). This association between British identification and higher rates of attendance is robust to the inclusion of identification with own ethnic group, which is also positive, in line with expectations. Thus, it is not simply that those more attached generally who are more religious, but among those with different levels of own group attachment British identity is still associated with greater religiosity. Despite debates about ‘oppositional’ cultures (Battu et al. 2005), religious practice in the UK would appear to have an integrative role in terms of destination country identification. Put otherwise, strong national identification is not felt to be at odds with devout religious practice (cf. Manning and Roy 2010). It may be that we are seeing the integrative role of religious institutions themselves, which can foster more general feelings of belonging and associated self-confidence (Foner and Alba 2008; Furbey et al. 2006). 23 Finally, Bangladeshis are significantly more likely to attend regularly than Pakistanis (Model 6). This is in line with our expectations concerning level of industrialisation in the origin country. In the Netherlands, then, there would appear to be indications of some secularising tendencies within patterns of Muslim religiosity, and conversely little evidence of revitalisation. While in the UK there is far less support for secularisation as a general process but exposure and contact seem much more pertinent to levels of attendance, though, as in the Netherlands, duration among the first generation is not associated with declining religiosity. Two particularly distinctive and contrasting features across the two countries relate to assimilation and integration arguments. In the Netherlands there was no specific second generation association that would imply a specific impact of upbringing and schooling (regardless of qualifications) on religious behaviour. By contrast, in the UK the second generation were significantly less likely to attend regularly, even after controlling for age, suggesting that upbringing – perhaps though greater contact with non-Muslims in school – matters. In the UK, British identification was positively correlated with attendance, while in the Netherlands Dutch identification was negatively correlated. These differences could tell us something about the distinctiveness of the different institutional settings and reception of Muslims in the two countries. For example, pillarization could help to account for the lack of a second generation effect in the Netherlands. However, equally important as differences in national context could be the differences in the ethnic origins of Muslims in the two countries. 24 Clearly we cannot present a counterfactual of the religious behaviour of Moroccan Muslims in the UK or Bangladeshi Muslims in the Netherlands, which might isolate ethnic origins from institutional effects, but given the differences between ethnic groups within each country we explored further the extent to which some of these key features were ethnically specific within each country, by estimating models with interactions between ethnicity and these apparently counter-intuitive associations, namely for the Netherlands, second generation, and for the UK British identity. (Tables available from authors on request.) However, these models resulted in no statistically significant differential probabilities for the interactions. 5. Conclusions There is increasing scholarly interest on the experience and practice of Muslims in European countries (Phalet and Haker 2004; Maliepaard 2010; Conner 2010). Our work contributes to these by a specific investigation – and comparison – of the factors associated with Muslim religiosity, in two rich Western countries in the period before 2001. Single country studies provide a rich understanding of the Muslim patterns of attendance and the factors that shape them, yet they implicitly suggest a continuity of experience across Muslims of different origins and in different settings (e.g. Phalet and Haker 2004; Maliepaard 2010), enhanced in a period in which the strong politicisation of discourses surround Muslims tend to assume cross-national continuities. Conversely, multiple-country studies can legitimately claim a greater degree of generalisation, but may obscure the specifics and complexities of correlates of Muslim religiosity as they are played out in particular national contexts (Ruiter and Van Tubergen 2009) or may not 25 have the scope to address minority religions in the countries considered (e.g. Kelley and De Graaf 1997). A case study approach, such as that pursued here, can retain some of the richness of analysis and interpretation of the single country studies, while highlighting the extent to which findings and theoretical frameworks of explanation can or cannot be applied consistently across more than one setting. Such an approach does not come without its difficulties in terms of harmonisation of data sources and of disentangling institutional impacts across the two countries from the differences in countries of origin of their Muslim groups, and these limitations clearly lead to caution in extrapolating from the results presented here. Yet the insights are potentially valuable in highlighting diversity and similarity among Muslims of different backgrounds and in different institutional contexts and in providing an agenda for future research. Moreover, our findings help to highlight the extent to which the questions asked in different national contexts are premised on particular assumptions processes of integration and ethnic attachment. Thus, the question on Dutch identity required people to position themselves as either Dutch or as of minority ethnicity, and an inspection of the Dutch results alone might have led to the inference that greater religiosity fosters lower Dutch identification and by extension to an assumption that religious Muslims were somewhat alienated from the host society. However, the question itself fails to recognise that own group and majority group attachments represent different dimensions of identity rather than two ends of a single pole, and operate in different ways in relation to host country interaction, as Schaafsma et al (2010) have recently shown. When we turn to the 32 Ruiter, S. and van Tubergen, F. (2009) 'Religious Attendance in Cross-National Perspective: A Multilevel Analysis of 60 Countries'. American Journal of Sociology 115 (3): 863-95. Tajfel, H. 1981. Human Groups and Social Categories: Studies in Social Psychology. Cambridge: Cambridge University Press. Thoits, P.A. (1983) 'Multiple Identities and Psychological Well-Being: A Reformulation and Test of the Social Isolation Hypothesis'. American Sociological Review, 48(2): 174-187. Tilley, J.R. (2003) 'Secularization and Aging in Britain: Does Family Formation Cause Greater Religiosity?', Journal for Scientific Study of Religion 42(2): 269-278. Turner, B.S. (2009) 'Reshaping the Sociology of Religion: Globalisation, Spirituality and the Erosion of the Social', The Sociological Review 57(1): 186-200. Van Tubergen, F. (2006) 'Religious Affiliation and Attendance Among Immigrants in Eight Western Countries: Individual and Contextual Effects'. Journal for the Scientific Study of Religion 45(1): 1-22. Van Tubergen, F. (2007) 'Religious Affiliation and Participation among Immigrants in a Secular Society: A Study of Immigrants in the Netherlands', Journal of Ethnic and Migration Studies 33(5):747-65. Vasta, E. (2007) 'From Ethnic Minorities to Ethnic Majority Policy: Multiculturalism and the Shift to Assimilationism in the Netherlands. Ethnic and Racial Studies 30 (5): 713-40. Verkuyten, M. (2005). 'Ethnic Group Identification and Group Evaluation among Minority and Majority Groups: Testing the Multicultural Hypotheses. Journal of Personality and Social Psychology 88: 121–138. Verkuyten, M. and Yildiz, A.A. (2007) 'National (Dis)identification and Ethnic and Religious Identity: A Study among Turkish-Dutch Muslims' Personality and Social Psychology Bulletin 33 (10): 1448-62. Weber, M. (1930 [1904]) The Protestant Ethic and the Spirit of Capitalism. New York: Scribner’s. Trans. by T. Parsons. 33 Ysseldyk, R., Matheson, K. and Anisman, H. (2010) 'Religiosity as Identity: Toward an Understanding of Religion from a Social Identity Perspective'. Personality and Social Psychology Review, 14(1): 60-71. Zhou, M. and Xiong, Y.S. (2005) 'The Multifaceted American Experiences of Asian Immigrants: Lessons for Segmented Assimilation', Ethnic and Racial Studies 28(6): 1119-1152. 34 Table I: Binary Logistic Regressions of Weekly Attendance: Netherlands 1998 Basic Secularisati on Security Assimilation Integration Ethnic origins Full Women -1.10*** -1.13*** -1.24*** -1.06*** -1.07*** -1.03*** -1.11*** (0.13) (0.13) (0.13) (0.13) (0.14) (0.13) (0.15) Partner 0.65*** 0.66*** 0.69*** 0.62*** 0.63*** 0.59*** 0.60*** (0.11) (0.12) (0.11) (0.11) (0.12) (0.11) (0.13) Age 0.05*** 0.04*** 0.04*** 0.05*** 0.05*** 0.06*** 0.04*** (0.00) (0.00) (0.00) (0.01) (0.00) (0.00) (0.01) No qualification Ref. Ref. Basic -0.18 -0.19 (0.12) (0.13) Secondary -0.49*** -0.52** (0.15) (0.17) A level -0.50* -0.49* (0.20) (0.23) University -0.94*** -0.78** (0.24) (0.25) Fluent in Dutch -0.43*** -0.30* (0.11) (0.14) Employed -0.47*** -0.30* (0.10) (0.12) 2nd generation -0.16 -0.01 (0.22) (0.27) Duration of stay 0.01 0.01 (0.01) (0.01) Dutch identity -0.94*** -0.74* (0.28) (0.31) No association Ref. Ref. Ethnic association 0.41** 0.56*** (0.15) (0.17) Majority association -0.14 0.19 (0.22) (0.23) Own group area 0.00 -0.00 (0.00) (0.00) Moroccan Ref. Ref. Turkish 0.21* 0.19 (0.10) (0.13) Surinamese -1.21*** -0.85** (0.23) (0.27) Constant -2.40*** -1.61*** -1.89*** -2.30*** -2.47*** -2.58*** -1.50*** (0.18) (0.24) (0.21) (0.20) (0.23) (0.19) (0.29) pseudo R2 0.13 0.15 0.14 0.13 0.14 0.15 0.17 Log likelihood -1411 -1327 -1398 -1372 -1384 -1389 -1259 N 2354 2263 2352 2288 2322 2354 2181 Standard errors in parentheses + p<.10, * p<.05, ** p<.01, *** p<.001 Standard errors for area effects take account of correlation of errors across areas. 35 Table II: Binary Logistic Regressions of Weekly Attendance: UK 1993 Basic Secularisation Security Assimilation Integration Ethnic origins Full Woman -0.92*** -1.13*** -0.97*** -1.00*** -0.92** -0.89*** -1.03*** (0.18) (0.20) (0.20) (0.20) (0.31) (0.19) (0.30) Partner 0.33 0.18 0.35 -0.09 0.25 0.31 -0.12 (0.21) (0.23) (0.22) (0.25) (0.23) (0.22) (0.25) Age 0.01* 0.01 0.01+ -0.00 0.02 0.02* 0.00 (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) No qualification Ref. Ref. Basic -0.15 -0.15 (0.29) (0.43) Secondary 0.27 0.25 (0.35) (0.26) A level -0.50 -0.50 (0.33) (0.33) University -0.08 0.16 (0.39) (0.53) Fluent in English -0.68** -0.32+ (0.25) (0.18) Employed -0.18 0.05 (0.23) (0.28) 2nd generation -0.80* -0.40 (0.32) (0.29) Duration of stay 0.01 0.01 (0.02) (0.01) British identity 0.27** 0.25*** (0.09) (0.07) Own ethnic identity 0.26* 0.23* (0.12) (0.10) No association Ref. Ref. Ethnic association 0.35 0.65* (0.39) (0.31) Majority association -0.30 -0.22 (0.29) (0.33) Own group area 0.02+ 0.02*** (0.01) (0.01) Pakistani Ref. Ref. Bangladeshi 0.41+ 0.54 (0.22) (0.34) Indian / African Asian -0.29 -0.14 (0.27) (0.40) Constant 0.45 1.25** 0.53+ 1.47*** -1.92** 0.35 -1.14 (0.28) (0.39) (0.30) (0.41) (0.70) (0.29) (0.71) pseudo R2 0.05 0.08 0.05 0.07 0.09 0.06 0.13 Log likelihood -375 -345 -375 -359 -350 -373 -307 N 944 877 944 917 895 944 808 Standard errors in parentheses + p<.10, * p<.05, ** p<.01, *** p<.001 All models are estimated applying design weights. Standard errors for area effects take account of correlations across area categories. 36 Appendix: Variable Descriptives Netherlands UK* N Mean / % N Mean / % Attendance 2399 47 944 67 Women 2411 22 1028 49 Partner 2406 71 1028 70 Age (years) 2380 39 1028 36 Educational qualifications 2358 1016 No qualifications 37 50 Basic 31 16 Secondary 21 14 A levels 7 9 University 5 11 Fluent in Dutch / English 2371 33 968 45 Employed 2414 50 1028 29 2nd generation 2414 8 994 32 Duration in country (years): for 1st generation 2125 17 752 19 Dutch identity 2387 2 Strength of British identity (score out of 5) 984 3.6 Strength of own ethnic identity (score out of 5) 994 4.3 Associational activity 2405 1022 No association 87 79 Ethnic group association 8 8 Majority association 5 12 Group proportion in postcode (%) 2414 34 1028 14 Ethnic group 2414 1028 Turkish 45 Moroccan 49 Surinamese / Antillean 6 Pakistani 64 Bangladeshi 21 Indian / African Asian 15 Note: all are proportions (percentages) except age, duration and British and own ethnic identity. *UK statistics are weighted to adjust for design effects, but the Ns are unweighted.