Abstract
Departing from the debate about the importance of gender ideology in the integration of Muslim minorities in Western Europe, this article studies the association between Islamic religiosity and gender ideology among second-generation Turkish and Moroccan men and women in Belgium. Islamic religiosity is conceptualized as consisting of religious identification, (partly gender-specific) religious practices, and belief orthodoxy. The comparability of this model across genders and national origin groups is tested with recently collected survey data. The association between Islamic religiosity and gender ideology is analyzed with regard to gender and ethnic differences. Taking a number of control variables into account, the analysis reveals no significant differences between national origin groups. More importantly, Islamic religiosity is found to be only weakly negatively associated with gender egalitarianism; this association is somewhat stronger for men than for women. These results challenge the alleged strong connection between Islamic religiosity and traditional gender ideology and highlight the importance of taking gender differences into account when studying religiosity among Muslim minorities in Europe.
The integration of Muslim minorities into Western societies is often described as one of the main challenges of contemporary societies. Gender ideology figures prominently in this debate, and differences in attitudes between Muslims and non-Muslims regarding the appropriate roles of men and women are often seen as a particularly salient field of cultural conflict (Sniderman and Hagendoorn 2007). Indeed, several studies show that Muslim immigrants on average subscribe to a more conservative gender ideology than the majority population in Western European countries (e.g., Diehl, Koenig, and Ruckdeschel 2009; Steinbach 2009). This is often attributed to the higher level of religiosity of Muslim immigrants. Some authors also relate it to Muslims’ religious affiliation as such, arguing that Islamic religious teachings or their most common interpretations encourage or even prescribe gender traditionalism (Inglehart and Norris 2003). If strong adherence to Islam is indeed closely associated with less support for gender egalitarianism, Islamic religiosity could be a hindrance to the integration of Muslim minorities into societies that consider the equality of men and women fundamental to their democratic systems.
Given comparatively lower levels of modernization of the countries of origin of Muslim immigrants to the West (Norris and Inglehart 2004), it is not surprising to find that first-generation immigrants display higher levels of religiosity and are more traditional in terms of gender ideology than Western European populations (e.g., Diehl, Koenig, and Ruckdeschel 2009; Steinbach 2009). However, less is known about the religiosity and the gender ideology of subsequent generations. Compared to their parents, the second generation grows up in very different contexts of socialization, which may affect both their level of religiosity and their gender ideology. Recent research documents generational stability regarding the level of religiosity, but also reveals changes toward more egalitarian gender ideology among non-Western immigrant groups in Europe (e.g., Diehl, Koenig, and Ruckdeschel 2009). However, several studies suggest that this intergenerational change holds only for women and less so for second-generation men (Diehl and Koenig 2009; Idema and Phalet 2007). Yet, to date, it is not clear whether the alleged strong association between Islamic religiosity and gender traditionalism applies equally to the local-born children of Muslim immigrants in Europe and whether it differs between genders.
One reason for the lack of evidence regarding the religiosity and gender ideology of second-generation Muslims in Europe is the difficulty of identifying them in many data sources. Thus, studies often include only immigrants who are foreign-born or who have a foreign nationality. The data from the TIES surveys (The Integration of the European Second generation, cf. Crul, Schneider, and Lelie 2012), which we use for our analyses, provide a unique opportunity to study multiple dimensions of Islamic religiosity in conjunction with gender ideology among large samples of the adult second generation in Belgium. 1 The comparative design of TIES, moreover, allows analysis of young adults from different national backgrounds (Turkey and Morocco) simultaneously. In both countries of origin, more than 95 percent of the population is affiliated with Islam (Pew Forum 2011, 157).
According to the 2001 Census, the Turkish and Moroccan second generation jointly represent about 4 percent of the total Belgian population (Phalet, Deboosere, and Bastiaenssen 2007). Immigrants from Turkey and Morocco arrived in the 1960s initially as (primarily single male and) low-educated “guest workers” to labor in the heavy industries and mining sector. Later, the immigrant communities grew through family reunification and formation, resulting in the emergence of a second and, now, third generation (Lesthaeghe 2000). Both groups are characterized by severe disadvantages in education and on the labor market (Phalet 2007; Phalet, Deboosere, and Bastiaenssen 2007). As with Muslim minorities in other countries, the level of religious identification and practice is high among Moroccan- and Turkish-Belgians because of efficient intergenerational transmission of religiosity (Güngör, Fleischmann, and Phalet 2011). Migration from Turkey and Morocco to Belgium is typical of post–World War II labor migration to Western Europe, and immigrants from these two countries are also found in other European destination countries (Voas and Fleischmann 2012), making these groups well suited to an empirical investigation into religiosity and gender ideology among second-generation Muslims in European societies.
This study asks whether and how the degree of individual Islamic religiosity is related to gender ideology. Throughout the analysis, gender differences are highlighted, both in the meanings and structure of Islamic religiosity and in its association with gender ideology. Simultaneously, differences between groups of different national origins are explored. Theoretically, the study aims to contribute to research and debates on the integration of Muslim minorities by highlighting the role of gender—both by focusing on the role of gender ideology and theorizing gender differences in religiosity and its association with gender ideology. The main empirical contribution of this study lies in its focus on the second generation, and the systematic attention to gender differences in Islamic religiosity and its association with gender ideology.
Literature Review
Research has shown that a higher degree of individual religiosity is associated with more traditional gender ideology (e.g., Thornton, Alwin, and Camburn 1983) and lower female labor market participation (e.g., Lehrer 1995). The association between religiosity and gender traditionalism is additionally documented at the macro level (Norris and Inglehart 2004). Most of these findings relate to the (often Christian) majority society and do not focus on immigrant minorities. Yet a study by Diehl and colleagues (Diehl, Koenig, and Ruckdeschel 2009) confirms the association between religiosity and traditional gender ideology also among Turkish Muslims in Germany. These findings can be explained following the perspective that religions in general link gender to symbolic norms and thus tend to legitimize inequalities and hierarchical relationships between men and women (Brinkenhoff and MacKie 1985).
In contrast to the reasoning that, irrespective of religious denomination, religiosity is associated with more traditional gender ideology (cf. Read 2003), some authors argue that one’s religious affiliation matters more than one’s level of religiosity, and that Islamic affiliation is particularly closely linked to gender traditionalism. Concretely, Inglehart and Norris (2003) refer to the religious teachings and principles of Islam to explain their finding of a negative association between Islamic affiliation and gender egalitarianism. They base their arguments on analyses of World Value surveys where the effects of religious affiliation were much stronger than the effects of religiosity. 2 Using the same data source, Alexander and Welzel (2011) also show a strong association between Islam and patriarchy at the individual and the societal level. At the same time, their findings indicate that approval of patriarchal values varies greatly among Muslims, and that similar factors as among non-Muslims explain these values, most notably the level of education and female labor market participation.
The latter results raise doubts about an alleged universal connection between Islamic religiosity and gender traditionalism, particularly with regard to the second generation. In line with the higher levels of education in the second as compared to the first generation and the widely documented positive influence of education on gender egalitarianism, the second generation on average subscribes to a more egalitarian gender ideology than their parents (e.g., Diehl, Koenig, and Ruckdeschel 2009; Leaper and Valin 1996; Phinney and Flores 2002). However, prevalent intergenerational stability is found in levels of religiosity (e.g., Diehl, Koenig, and Ruckdeschel 2009). Combining evidence on the generally negative association between (Islamic) religiosity and gender egalitarianism with the findings of intergenerational change in gender ideology and stability in religiosity, we expect that among second-generation Turkish- and Moroccan-Belgians higher levels of individual religiosity are negatively associated with gender egalitarianism, but that this association is rather weak (hypothesis 1).
Yet this weak negative association between religiosity and egalitarian gender ideology does not seem to be the whole story. Concretely, gender ideology tends to differ between genders, with men being more traditional than women (e.g., Steinbach 2009), a pattern that has also been confirmed among children of immigrants from comparatively traditional societies (Idema and Phalet 2007; Leaper and Valin 1996; Phinney and Flores 2002). At the same time, women are on average more religious than men (Sullins 2006), although this gender difference is often smaller among Muslims (at least when religiosity is assessed with religious service attendance, which is lower among Muslim women). This suggests that the negative association between religiosity and gender egalitarianism may be weaker among women than among men.
A gender-specific association can be argued to result from the different consequences of gender ideology for men and women. Traditional gender ideology confines women to the home and limits their opportunities regarding economic activity and range of occupations. These restrictions, however, do not hold for men, though traditional gender ideology may have other negative consequences, for example, increasing the pressure to achieve in the labor market. These differential consequences for men and women not only explain the more egalitarian gender ideology of women, they also provide an explanation for a weaker association between Islamic religiosity and gender ideology among women. Several studies show that Muslim identity and Islamic practices are a central part of the identity of second-generation men and women (e.g., Duderija 2007; Fleischmann and Phalet 2012). However, a strong association between Islamic religiosity and traditional gender ideology would lead to conflict for young Muslim women who want to realize their chances of education and employment in Western societies. For men, in contrast, being religious and subscribing to a traditional gender ideology does not interfere with educational and occupational opportunities.
More evidence for a gender-specific association between Islamic religiosity and gender ideology comes from ethnographic research among Muslim women of immigrant origin in Western societies, some of them very religious and fully veiled. These studies have repeatedly shown how Muslim women actively contest the association of Islamic religiosity with gender traditionalism and how they negotiate gender and religion in their new societies (Jacobsen 2004). In these discourses, patriarchal norms and practices are often attributed to “un-Islamic” cultural practices of the (parental) countries of origin (Warner, Martel, and Dugan 2011; Williams and Vashi 2007). Along these lines, feminist reinterpretations of the Qur’an strive to deconstruct the links between Islam and patriarchy (Wadud 1999, 2006), arguing that the religious texts actually prescribe gender equality (e.g., Abu-Ali and Reisen 1999). This suggests that at least for some second-generation Muslim women, there is no negative association between Islamic religiosity and gender egalitarianism, while a negative association may persist among second-generation men. Because of the ethnographic research design, however, it is hard to conclude whether this finding applies solely to the specific subgroups studied. We will examine this question among large random samples in order to find out whether the absence of the association between Islamic religiosity and gender traditionalism among women can be generalized beyond existing ethnographic research. We will test the hypothesis that the association between Islamic religiosity and gender ideology differs between second-generation men and women. Concretely, we expect that the negative association between religiosity and traditional gender ideology will be weaker for women than for men (hypothesis 2). In the following section, we describe the data source and the methods used to test these hypotheses.
Methods
Data Source
We use the Belgian TIES survey that was conducted in 2007 and 2008 (see Swyngedouw et al. 2008 for a technical report). In the cities of Antwerp and Brussels, 18- to 35-year-old Belgian-born children of immigrants from Turkey and Morocco were randomly sampled from municipal population registers (Antwerp) or commercial address lists (Brussels) and interviewed face to face. We restrict our analyses to participants who self-identified as Muslims. 3 The total sample size available for analysis is N = 984, which we divide into analytical groups on the basis of gender and national origin, resulting in four groups: Turkish-Belgian men (n = 269), Moroccan-Belgian men (n = 205), Turkish-Belgian women (n = 233), and Moroccan-Belgian women (n = 277). 4
Measures
Religiosity
In line with sociologists of religion, we define religiosity as a multidimensional construct consisting of identity, belief, and practices (Voas 2007). Religious identity expresses the strength of identification with one’s religion. Participants answered four questions regarding their religious identification on 5-point rating scales, with higher values reflecting stronger religious identification (see Table 1 for the wording of the items).
Descriptive Statistics of Religiosity and Gender Ideology by Gender and Ethnic Background (Means [M] and Standard Deviations [SD] per Group and Gender)
NOTE: The sum of categorical items can deviate from 100% because of missing values. Means with different subscripts a–c differ significantly from each other according to univariate F tests, p < .05.
For separated or divorced men: “Did your partner wear a headscarf when outside the house?”; for men without partner: “Would you want your partner to wear a headscarf outside the house?”
Our data contain two indicators of belief orthodoxy that are drawn from a scale of religious cognitive styles (Duriez, Soenens, and Hutsebaut 2005), and that assess the degree to which individuals support literal interpretations of religious scripts and their openness for multiple answers to religious questions. These questions were answered on 5-point rating scales, with higher values reflecting greater belief orthodoxy.
Regarding religious practices, TIES includes indicators of gender-neutral as well as gender-specific practices. Concretely, the frequency of praying, fasting during Ramadan, and consuming proper (halal) food can be considered gender-neutral as they are practiced by both men and women, and often privately. The frequency of religious service attendance or mosque visits, however, is known to be much higher among men, because attendance during Friday prayer is required only for men, but not for women. Although it can be shown that female attendance of religious services and other activities in the mosque increase in the context of migration (Predelli 2004, 2008), a gender difference in the frequency of religious service attendance in the second generation is still likely. Moreover, TIES collected information about the wearing of a headscarf. Clearly, this practice applies only to women, who were asked whether or not they cover their hair when outside their home. Male participants also received questions about the headscarf referring to their female partner. 5 The question about the headscarf is a dummy variable that takes the value 1 for women who wear a headscarf and for men whose partner wears a headscarf (or whose past partner used to wear a headscarf, or who would wish a future partner to wear a headscarf). The other religious practices were rated on 5-point scales, with higher values indicating higher frequencies.
Gender ideology
We conceptualize gender ideology as attitudes toward the appropriate roles of men and women, which can vary on a continuum from more traditional to more egalitarian (cf. Idema and Phalet 2007). Traditional gender ideology denotes attitudes supporting differential roles for men and women where men take primary responsibility for earning an income in the labor market and are in charge of most important household decisions, whereas women take care of the household and children (Thornton, Alwin, and Camburn 1983). Egalitarian gender ideology, on the other hand, relates to equality of men and women with regard to participation in the labor market, distribution of household chores and childcare, and decision-making in the household. Three measures of gender ideology assess participants’ agreement with female labor market participation when there are young children in the household, the acceptance of women supervising men at work, and the relative importance of higher education for boys and girls (see Table 1 for the wording of the items). Agreement with the statements was assessed on 5-point Likert scales which were recoded such that higher values represent a more egalitarian gender ideology.
Control variables
We control for age (in years), education, economic status, religious socialization, and partnership status. Regarding education, three dummy variables were created based on the highest degree attained or currently pursued in case participants were still following full-time education. We distinguish between lower secondary or less, upper secondary (the reference category), and tertiary degrees. Economic status is indicated by three dummy variables, separating employed and self-employed participants from full-time students and inactive (including unemployed) participants. Religious socialization was assessed by two indicators. A dummy variable indicates whether participants attended Qur’an lessons as a child; further, the frequency of parental religious service attendance during primary school age was rated on a 5-point scale, with higher values indicating more frequent attendance. With respect to partnership status, a dummy variable separated participants who were married or in a common-law union from participants who were single, divorced, or widowed.
Analyses
Conceptualizing religion as a multidimensional construct, we treat religiosity as a second-order latent variable constituted by the latent variables religious identification, belief orthodoxy and religious practices. Concretely, this implies that we assume that religiosity cannot be directly observed by the measurements of the TIES-survey, but that this unobserved characteristic drives respondents’ answers to questions about their religious identification, practices, and beliefs. Structural equation modeling is used to handle the latent nature of religiosity. In comparison to the more commonly used ordinary least squares regression, structural equation modeling takes into account measurement error and the potential lack of comparability of scales across groups. The latter is particularly important in our case as previous research strongly suggests that at least some aspects of Islamic religiosity are not equivalent for men and women.
First, we tested the measurement equivalence of the model of religiosity across genders and national origin groups. Generally speaking, confirming the equivalence of latent variables across groups is a prerequisite for making meaningful comparisons across groups (Schmitt and Kuljanin 2008; Steinmetz 2011). In testing measurement equivalence, several increasingly strict criteria can be defined and were assessed in this study. First, configural equivalence implies that the structure of latent variables is the same. In our case, this would mean that religious identification, religious practices, and beliefs are significant indicators of religiosity across all analytical groups (instead of, for instance, mosque visits being significantly associated with religiosity in one group, but not in another). Metric equivalence additionally requires factor loadings to be equal across groups, meaning that associations between measured and latent variables are equivalent across groups. This is usually interpreted as indicating equivalence of meaning of the latent variable across groups. Finally, scalar equivalence requires the means of latent variables and intercepts of measured variables to be equal across groups, meaning in this case that the frequency of religious practices, the level of identification, and orthodox beliefs do not differ across groups. Methodologists debate which level of equivalence is minimally required to make meaningful comparisons across groups (Steenkamp and Baumgartner 1998; Steinmetz 2011). Here, we subscribe to the view that metric equivalence is sufficient, because it implies that the internal structure of the latent variables is the same across groups. This can be interpreted as the latent variables having the same meaning for the different groups, while allowing for the possibility that their mean values differ across groups.
Considering that some of the religious practices we measure are gender-specific, we expect the model of religiosity to be equivalent for men and women with regard to religious identification, belief orthodoxy, and gender-neutral religious practices; however, we expect gender differences regarding mosque attendance and headscarf use. More specifically, we expect that men and women will not differ in the structure of the model of religiosity—that religious identification, religious practices, and beliefs will be significant components of religiosity for both men and women—thus confirming configural equivalence. Yet, both the mean level of gender-specific religious practices and their associations with other components of religiosity may differ between men and women. For instance, visiting the mosque may be more strongly associated with other (gender-neutral) aspects of religiosity among men than women; in this case, the assumption of metric equivalence across genders would be violated. Moreover, mosque visits are likely to be more frequent among men than women, implying different intercepts of the item mosque visits, thus violating the assumption of scalar equivalence. As noted above, the rejection of metric (but not necessarily scalar) equivalence of the latent variable religiosity across genders would jeopardize the comparability of religiosity and its association with gender ideology across genders and national origin groups.
Confirmatory factor analysis in structural equation modeling was used to test the equivalence of the latent variable religiosity and its dimensions religious identification, belief orthodoxy, and religious practices across the four analytical groups (men and women of Turkish and Moroccan origin). Scalar invariance was tested by restricting the means of the latent variables and the intercepts of the measured variables, in addition to the factor loadings of the measures on the latent variables, so as to be equal across analytical groups. We compared model fit indicators of models with and without constraints and tested significant differences with chi-square tests. If the fit of the constrained and the unconstrained model do not differ significantly, this implies that imposing the constraint (i.e., assuming that means, intercepts, and factor loadings are equal) is a good representation of the data and thus equivalence is supported.
After testing equivalence, the covariance between religiosity and gender ideology was computed in structural equation models, while including all significant paths from control variables to religiosity and gender ideology. Because our interest is in the association between religiosity and gender ideology, and because we do not assume a particular causal ordering of the two phenomena (in the sense that one’s religiosity determines one’s gender ideology, or vice versa), estimating the covariance between religiosity and gender ideology is more appropriate than approaching these concepts in terms of independent and dependent variables.
Findings
Descriptives
Table 1 displays means and standard deviations of all variables in the analyses by gender and national origin. Mean differences were tested with univariate F tests, and significant group differences are indicated with different subscripts. Participants in all groups displayed high levels on all subscales of religiosity, although belief orthodoxy received less support. With the exception of mosque attendance, participants of Moroccan origin showed higher levels of religious identification and practices than participants of Turkish origin. In line with earlier findings about their gender specificity, the mean levels of mosque attendance and headscarf use differed significantly between genders, with men showing higher levels on both practices. Regarding gender ideology, the means lie above the neutral midpoint of the scale, indicating prevalent support for egalitarian rather than traditional gender ideology among the second generation. In line with previous research, women are significantly more egalitarian than their male counterparts.
Confirmatory Factor Analyses of Islamic Religiosity
Given the mean differences in gender-specific religious practices documented in Table 1, it is important to test whether these indicators are also associated with other indicators of religiosity in a different way for men and women. This would imply estimating different factor loadings of these two religious practices for men and women, and make it difficult to compare levels of religiosity and their association with gender ideology across genders.
The second-order model of religiosity with equal factor loadings and equal intercepts across all groups had a significantly worse fit in comparison to the unconstrained model; thus, full scalar equivalence is not supported. 6 The model fit improved significantly by allowing the intercepts of mosque visits and headscarf use to vary across genders. 7 This restates the previous finding from Table 1 that these two religious practices are followed with significantly different frequencies by men and women. Specifically, the intercepts of both indicators were higher for men than for women. 8 Allowing the intercepts of these items to vary between national origin groups further improves model fit in comparison to the previous model. 9 The partially equivalent measurement model of religiosity, with freely estimated intercepts of the items mosque visit and headscarf use but otherwise equal intercepts and factor loadings across all four groups, has an acceptable fit. 10 Factor loadings and intercepts of this measurement model can be found in Table 2.
Factor Loadings and Means of Religiosity and Gender Ideology in Measurement Models
NOTE: All parameter estimates are significant at p <.001. The column labeled B (SE) contains the factor loadings of the manifest items on the latent variables and, between parentheses, their standard errors. Columns labeled M (SE) display the mean of the respective variable in each group (or for all groups if it does not differ significantly) and the standard errors, in parentheses.
This analysis shows that when gender differences in the frequency of gender-specific practices are taken into account, which is achieved by letting the intercepts vary across groups, the factor loadings are equal across groups. This implies that metric equivalence is confirmed (i.e., the associations of the gender-specific practices with other indicators of religiosity do not differ significantly across genders or national origin groups). Hence, it is feasible and meaningful to use these gender-specific indicators of religious practices in comparisons of Islamic religiosity and its association with gender ideology, as long as the gender differences in the frequency of visiting the mosque and wearing the headscarf are taken into account.
The Association between Islamic Religiosity and Gender Ideology
To test our hypotheses on the (gender-specific) association between Islamic religiosity and gender ideology, we estimated a structural equation model specifying a covariance between the second-order factor religiosity and the first-order factor gender ideology. The intercepts and factor loadings of the latent variable gender ideology were constrained to be equal across all groups. Moreover, significant paths from all control variables to these latent factors are included; the pertaining regression coefficients can be found in Table 3.
Regression Coefficients of Control Variables in the Final Structural Model
Reference category: upper secondary education.
Reference category: (self-)employed.
p < .05, **p < .01, ***p < .001.
Taking into account all significant effects of control variables, the model with gender-specific estimates of the covariance between religiosity and gender ideology fits the data equally well as a model with equal covariance across all groups. 11 Estimating the covariance between religiosity and gender ideology independently for men and women reveals a higher value for men than for women. 12 Thus, in line with hypothesis 1, the association between Islamic religiosity and egalitarian gender ideology is negative but weak for both men and women, implying that more religious members of the second generation tend to subscribe to a more traditional gender ideology. Our second hypothesis regarding gender differences in this association, however, is neither clearly supported nor rejected, as a model with gender-specific covariance fits the data equally well as a model with the same covariance for men and women. Though the association between religiosity and gender ideology tends to be somewhat stronger among men, this association is weak for both men and women in light of correlations of r < −0.3 (Cohen 1988), in line with hypothesis 1.
Additional Analyses
By way of sensitivity tests, we reestimated the models described above with a number of alternative specifications to test whether our results are robust to the inclusion or exclusion of specific variables and constraints. First, we modified the structural equation model to find out to what extent the (gender-specific) covariance between religiosity and gender ideology depends on the chosen model specifications. We first reestimated the model without the headscarf item, because it may lead to problems in the estimation of parameters because of its dichotomous nature, and because its meaning differs between men and women. In addition, we also removed the mosque item in order to reestimate the covariance using only gender-neutral indicators of religious practices. Moreover, we reestimated our final model after removing items with little variance in order to test to what extent the finding of a weak association between religiosity and gender ideology is merely an artifact of a lack of variance in the latent variables. The results of these additional analyses generally support the findings described above. The correlation between religiosity and gender egalitarianism remains negative and weak; depending on model specifications, the difference between men and women reaches significance or not.
Our analyses suggest that Islamic religiosity is only weakly negatively associated with egalitarian gender ideology among the Turkish and Moroccan second generation in Belgium. There are some indications that this negative association is more outspoken among men than among women, but this gender difference could not be consistently confirmed throughout all analyses.
Discussion
The assumption that a higher level of Islamic religiosity inevitably goes together with more traditional and less egalitarian gender ideology figures prominently in societal and political debates about the integration of Muslim minorities into Western societies. Indeed, if a universally strong and negative association between Islamic religiosity and gender egalitarianism were found among all Muslims, this would jeopardize the integration of this religious minority into societies that consider the equality of men and women fundamental to their liberal democracies. Studies in majority Muslim societies and among first-generation Muslim immigrants in Western societies document a negative association between the level of religiosity and gender egalitarianism (Diehl, Koenig, and Ruckdeschel 2009; Inglehart and Norris 2003; Steinbach 2009). The results of our study of Belgian-born children of immigrants from Turkey and Morocco, however, show that this association is rather weak in the second generation. Moreover, the negative association was consistently found among men, while the association was less strong among women, and, depending on model specifications, failed to reach significance. This suggests that gender traditionalism has differential functions for men and women in conjunction with Islamic religiosity. For young men, subscribing to traditional gender ideology and being highly religious presents no contradiction and does not limit their opportunities regarding educational and labor market careers. For young women, on the contrary, traditional gender ideology sets severe limitations to their opportunities, particularly with regard to higher education and labor market participation.
The results of our study thus confirm previous findings documenting the gender-specific role of gender ideology also in the context of migration (Idema and Phalet 2007; Leaper and Valin 1996; Phinney and Flores 2002), and they extend these findings by pointing to a partly gender-specific association with Islamic religiosity. Moreover, our results provide a first quantitative confirmation of a recurrent finding from ethnographic studies documenting how very religious, including fully veiled, Muslim women in Western societies actively dispute an interpretation of Islam as prescribing patriarchal values and practices (Warner, Martel, and Dugan 2011; Williams and Vashi 2007). By questioning the alleged strong association between Islamic religiosity and traditional gender ideology, these women create the space to express their high level of religiosity in their daily lives without jeopardizing their opportunities in education and the labor market.
Our detailed investigation and careful testing of equivalence of the multidimensional model of Islamic religiosity for men and women revealed important gender differences in the structure and content of Islamic religiosity. Specifically, we showed that religious identification, belief orthodoxy, and gender-neutral religious practices (praying, fasting, and following the rules of consumption) were related to each other in the same way and were equally high for men and women and thus represent Islamic religiosity equally well for both genders. This was, however, not the case for religious service attendance and the use of a headscarf. These gender-specific religious practices were found to be more frequently practiced among men (or their female partners) than women, yet the associations with other indicators of religiosity were the same for men and women. As a consequence, these indicators could be meaningfully used in a comparison of Islamic religiosity and its association with gender ideology across men and women under the condition that the gender difference in frequency of these practices is taken into account. The implication of this finding for future research is that one needs to pay careful attention to gender differences in empirical analyses of Islamic religiosity, either by estimating separate models for men and women as we did in this study or by taking interactions of gender and gender-specific religious practices into account.
A number of limitations need to be kept in mind when drawing conclusions from this study. One of them is the limitation to a single national context, which occurred for practical reasons. Theoretically, it would be interesting to replicate the analyses in other Western European countries that may differ with respect to levels of religiosity or prevalent gender ideology. In addition, including Muslim participants from different national origins than Turkey and Morocco could extend the scope of this research and reveal to what extent our findings generalize to the second generation of Muslims in Europe. Moreover, a wider range of measures to assess gender ideology and related attitudes could further strengthen the conclusions, particularly if additional measures would also cover gender ideology related behavior, such as task distributions within households. With the limited set of measures available in the TIES-surveys, our results allow only conclusions about the role of attitudes toward women’s qualified participation in the labor market. These limitations, which relate to the use of existing survey data, should be appreciated in light of the comparative design, including randomly selected participants from two different national origin groups and careful testing of equivalence of latent variables as strong points of our analyses. As we could establish partial scalar equivalence across groups, the findings of our structural equation models should be very robust against measurement error, and they provide the basis for a meaningful comparison of Islamic religiosity and gender ideology across different national origin groups and genders. Though our sample sizes may be considered small, particularly in relation to the rather complex analytical models, to our knowledge no alternative data set currently exists that provides larger sample sizes and allows studying our research question with a focus on the second generation of Muslim minorities in Europe.
To conclude, our study replicated the prominently discussed negative association between Islamic religiosity and egalitarian gender ideology among the second generation of Turkish and Moroccan origin in Belgium. However, we found this association to be rather weak and for women even nonsignificant when taking a range of control variables into account. Thus, our findings pose a challenge to the widely shared assumption of a universal and strong association between Islamic religiosity and more traditional gender ideology, and the related assertion that Islamic religiosity hinders the integration of Muslim minorities into Western societies.
Footnotes
Notes
Jana A. Scheible has recently graduated from the master’s program in sociology and social research at Utrecht University with a thesis on intergenerational transmission of gender ideology of immigrants in Europe. During her studies she joined the Social Science Research Center Berlin for an internship from which this article resulted.
Fenella Fleischmann is Assistant Professor at the European Research Centre on Migration and Ethnic Relations at Utrecht University. Her research focuses on immigrant integration, relating to minorities’ attainment in education and on the labor market, as well as intergroup relations, identity, discrimination, and religion among Muslim minorities in Europe.