Divorce in Turkish and Moroccan Communities in Belgium

Abstract

This paper focuses on divorce amongst Turkish and Moroccan Belgians, with a specific focus on the effect of partner-choice patterns. Divorce patterns of marriages established between 01 January 2001 and either 31 December 2003 (descriptive part), or 31 December 2005 (event-history analyses) are analysed and compared to marriages established between 01 January 1988 and 31 December 1990. We distinguish three marriage types: transnational marriages (i.e. marrying a partner from Morocco or Turkey), local intra-ethnic marriages (marrying another Moroccan of Turkish Belgian) and mixed marriages (i.e. marrying someone with a Belgian or other Western–European citizenship). To research divorce rates, we analysed population data from the Belgian national register, using piecewise constant log-rate event-history analyses with effect coding on all marriages taking place between 01 January 2001 and 31 December 2005 (N Turkish = 9631, N Moroccan = 17,786). First, the results reveal that in the past 15 years, divorce rates have doubled within Turkish and Moroccan migrant groups. Second, divorce rates are much higher amongst the Moroccan group. Third, there are clear differences between marriage types. Local intra-ethnic marriages have the lowest divorce levels, mixed marriages the highest, and transnational marriages take up a middle position.

Introduction

In the past decades, Belgium has experienced a transition towards a mass divorce society which made divorce an inextricable part of contemporary society (González and Viitanen 2009; Statistics Belgium n.d.). Divorce rates started to rise in the seventies and kept on increasing until now (Eurostat n.d.a). This divorce trend is part of the second demographic transition that started in the late sixties in many countries in Western and Northern Europe. This transition included interrelated changes, such as declining fertility rates and the weakening of the family as an institution, as evidenced by increased divorce and non-marital cohabitation rates and declining nuptiality rates (Corijn 2012; Lesthaeghe and Van de Kaa 1986). Moreover, this transition assumes ideational and cultural change (Lesthaeghe and Meekers 1987). Not only do economic factors condition individual life decisions, these decisions are steered by the emergence of self-fulfilment, personal freedom of choice, personal development and lifestyle, and emancipation as well (van de Kaa 1996).

These different transitions might have different explanations, however. For example, the rise in divorce might be explained by the emergence of the dual provider model, and therefore the increasing labour force participation of women and their growing economic independence (Kalmijn and Poortman 2006; Neels 2006; Sandström 2012). It seems that the increasing levels of female education and participation in the labour market put women in the socio-economically able position to divorce. Moreover, the increased importance of individualization eroded the importance of marriage (Cherlin 1992; Sandström 2012).

Causes of rising divorce rates have been found at the micro-level as well. Related to the growing emancipatory attitudes, problems in the realm of work and household labour have become increasingly important motives underlying the choice to divorce (De Graaf and Kalmijn 2006). For example, women often refer to the division of labour in the home and their former husbands’ working too much as motives underlying the choice for divorce (De Graaf and Kalmijn 2006). Moreover, evidence was found for economic (independence) hypotheses, which argue that divorce chances are increased if women have paid work (Kalmijn et al. 2004; Poortman and Kalmijn 2002), and cultural hypotheses, which state that the likelihood of divorce is increased if women adhere to emancipatory norms, regardless of their own work situation (Kalmijn, De Graaf and Poortman 2004). However, some studies found only weak support for the economic hypotheses, stating that women’s employment does not destabilize happy marriages but only increases the risk of disruption in unhappy marriages (Sayer and Bianchi 2000; Schoen et al. 2002).

Although in many cases people have good reasons for divorcing, the disadvantages are well documented (Amato 2000). They include financial hardship, decreased standard of living, less wealth (Zagorsky 2005), health problems (Joung et al. 1997) and mental health issues (Öngider 2011). These negative consequences might be even more pertinent to the situation of migrants. Their situation might be exacerbated by their generally more strained socio-economic circumstances relative to non-migrants, as well as reputation damage and perhaps even social isolation when marriages end in divorce (Corluy and Verbist 2010; Welslau and Deven 2003). Particularly in the case of transnational marriages (i.e. Turkish or Moroccan Belgians marrying a partner from the country of origin), migrating partners might be at greater risk of complete dependence (both economic and social) upon their partners during the marriage, given that their own families continue to live in the country of origin (Geets 2006; Timmerman 2006).

In this paper, we provide insight into divorce amongst Turkish and Moroccan Belgians1—the two largest migrant groups of Belgian residents originating from third countries (i.e. countries outside Europe) (Lievens 1998, 1999; Reniers 1999). They have been long established in Belgium, which enables us to study their marriage and divorce patterns over time. Turkish or Moroccan Belgians wishing a divorce should face no legal obstacles. If two partners married in Turkey or Morocco, they can divorce in Belgium if one of the three following requirements is satisfied (Koelet et al. 2009a): (1) they both have to live in Belgium or (2) they both have to have the Belgian nationality. If neither conditions are satisfied, it is sufficient that (3) only one partner is living in Belgium in addition to one of the following conditions: the last habitual residence of the spouses was in Belgium, the spouses file for divorce together, the spouse living in Belgium is the defendant, the spouse living in Belgium is the petitioner, and he or she is living in Belgium for at least 12 months. The ‘nationality of the marriage’, i.e. the place where the marriage occurred, is not important for the authority of a Belgian judge; the habitual residence is crucial. In practice, everyone living in Belgium for at least 12 months (even if it is illegal) can petition for divorce. However, Turkish and Moroccan migrants still have the choice to divorce in the country of origin, regardless of the place of marriage. In many cases, Moroccan and Turkish Belgians are advised to divorce in Belgium, to avoid legal problems (Koelet et al. 2009a).

Notwithstanding the weaker economic position, divorce is far from absent amongst Moroccan and Turkish Belgians. A study on divorce rates of marriages established between 1988 and 1991 of Turkish and Moroccan Belgians showed that divorce is prevalent amongst these groups as well and is related to the partner choice in the marriage (Eeckhaut et al. 2011). About 7% of all Turkish Belgians and 14% of all Moroccan Belgians were divorced by 31 December 1995 (Eeckhaut et al. 2011). For the general population, divorce rates were about 14% (Corijn 2012).

It has become increasingly evident, however, that these transitions may vary depending on the context and different sections of the population (Neels 2006; Sobotka and Toulemon 2008; Sobotka et al. 2012). Related to marital behaviour, it has been shown that higher educated persons marry later and are the least likely to divorce (Aughinbaugh et al. 2013; Isen and Stevenson 2010). As a consequence, these transitions might occur differently in migrant populations given that, relative to non-migrants, people of migrant background are more likely to live in strained, or even impoverished socio-economic circumstances, and that the activity and employment rates of migrants tend to be lower than those of non-migrants (VDAB Studiedienst 2012).

Evidence for this transition amongst Turkish and Moroccan Belgians, however, can already be found in a fertility decline and higher levels of cohabitation (Coleman 1994; Corijn 2012; Lesthaeghe 1997; Schoenmaeckers et al. 1999). Particularly amongst Moroccan men of the second generation, higher levels of cohabitation as well as a stronger tendency to remain single are prevalent (Coleman 1994; Corijn and Lodewijckx 2009; Hartung et al. 2011; Lesthaeghe 1997; Schoenmaeckers et al. 1999). Hence, this group is in theory most prone to divorce, but in practice, they might not marry at all.

In the past decade, however, it is possible that divorce trends have changed. Within Moroccan and Turkish migrant communities, divorce levels might follow those present in the country of origin. Divorce rates in Turkey are low but are increasing since the late eighties and particularly in the twenty-first century (Eurostat n.d.b; Kavas and Gündüz-Hoşgör 2010). For example, the crude divorce rate in the mid-nineties in Turkey was only 0.5, while this increased to 1.6 in 2010.2 In Morocco, divorce rates were not that low mostly due to the relative ease at which men could divorce their spouses before 2004 (Koelet et al. 2009a). After the legislation changed in 2004, divorce rates in Morocco dropped. Furthermore, socialization and integration processes amongst Turkish and Moroccan Belgians (Alba and Nee 1997) might make them more susceptible to transitions related to the second demographic transition, such as higher levels of individualization, informalization of marriage and female emancipation in Belgium. Moreover, changed household economics in migrant household might change the divorce propensity (Kalmijn et al. 2004; Poortman and Kalmijn 2002).

Divorce patterns of Turkish and Moroccan Belgians are related to their partner-choice patterns as well. In general, heterogamous marriages experience a lower stability compared to homogenous marriages (Eeckhaut et al 2011; Kalmijn et al. 2005; McPherson et al. 2001; Milewski and Kulu 2014). In many cases, these marriages lack social support and partners experience tension due to cultural differences (Eeckhaut et al. 2011). The lower divorce risks of mixed marriages established by Moroccan Belgians compared to Turkish Belgians were explained by a stronger orientation towards the host country and a less pronounced transnationalism of the Moroccan group which led to stronger cultural differences with the country of origin, and lesser social support for transnational marriages. Cultural differences can be present in transnational marriages as well, which might result in higher divorce rates compared to local intra-ethnic marriages (Timmerman et al. 2009; Van Kerckem et al. 2013).

Recent changes in the marriage pattern might impact divorce trends of Turkish and Moroccan Belgians. Between 1988 and 1991 (and before that), transnational marriages were the dominant partner choice for both Moroccan and Turkish Belgians (Eeckhaut et al. 2011). Recent studies have shown that the dominance of transnational marriages is declining in favour of local intra-ethnic marriages and, to a lesser extent, mixed marriages (Dupont et al. 2017a; Van Kerckem et al. 2013). This evolution may point to a more ‘careful’ approach to marriage partner selection, particularly with regard to transnational marriages. For these reasons, marriages might be established better considered. Furthermore, examining the partner choices of recent marriage cohorts can deepen our insight in the usefulness of explanations regarding social support and cultural difference.

Despite the relevance of examining divorce patterns of migrants, and particularly of marriage migrants, studies on divorce patterns have mainly focused on ethnic, religious or racial intermarriage versus intra-marriage (see, for example, Andersson et al. 2015; Bratter and King 2008; Kalmijn et al. 2005; Milewski and Kulu 2014; Smith et al. 2012; Zhang and Van Hook 2009). While these studies widely documented higher divorce rates of heterogamous compared to homogeneous marriages, this focus neglects cultural differences present in transnational marriages, which are ethnic homogenous marriages as well (Eeckhaut et al. 2011). Generally, research on transnational divorce is very limited (Sportel 2016). The few studies that focused on transnational divorce, were qualitative in nature and neglected other types of marriages (see Liversage 2012, 2013). Additionally, while Schultz-Nielsen and Bonke (2016) acknowledged the importance of the inclusion of transnational marriages within the homogamy–heterogamy dichotomy, they operationalized this by only differentiating between first- and second-generation couples. The only study that, to our knowledge, includes all partner-choice patterns is the one performed by Eeckhaut et al. (2011). However, given the recent evolutions in partner-choice patterns [i.e. rise in mixed marriages and declining levels of transnational marriages (Dupont et al. 2017a; Van Kerckem et al. 2013)], an inquiry in the evolutions in divorce is appropriate. While these trends might indicate a stronger orientation towards the host country, the higher levels of divorce risks of mixed marriages previously documented by Eeckhaut et al. (2011) might affect the increasing number of mixed marriages.

In this paper, we examine divorce patterns of male and female Turkish and Moroccan Belgians of the first, 1.5 and second generation3 by using population data. Our research has two dimensions. First, we describe divorce rates of marriages established between 2001 and 2003 and compare them to divorce rates of marriages established between 1988 and 1990 (Eeckhaut et al. 2011). The second dimension relates to our expectation that divorce rates are dependent upon the marriage type. As a consequence, event-history analyses are applied to analyse divorce rates of marriages established between 2001 and 2005 which were followed until 2011, to examine the difference in divorce risks of the different marriage types.

Turkish and Moroccan Belgians in a Mass Divorce Society

There is some empirical support for the claim that divorce is on the rise. Corijn and Lodewijckx (2009) show high divorce rates for Turkish and Moroccan Belgians married after 1994. In neighbouring countries fairly high divorce rates are reported for recent marital cohorts as well, particularly for mixed marriages [e.g. van Huis and Steenhof (2004) for the Netherlands and Milewski and Kulu (2014) for Germany]. There are several reasons to expect an increase in divorce rates in recent marriage cohorts in Belgium as well. First, divorce trends in the country of origin are discussed. Secondly, we describe divorce trends in Belgium. Finally, and related to divorce trends in Belgium, integration and socialization effects will be discussed.

First, according to Dutch studies the divorce propensity in the wife’s origin country relates to the divorce risk of a couple (Smith et al. 2012). In Turkey divorce rates are rather low, but on the rise—characterized by a rise of the crude divorce rate from 0.5 in the mid-nineties to 1.6 in 2010 (Eurostat n.d.b).4 For Morocco, divorce rates are hard to find but the numbers that are available point to much higher divorce rates in Morocco compared to Turkey (Koelet et al. 2009a). Particularly for men, it was quite easy to separate unilaterally from their wives. Because of these factors, it is plausible that an increase in divorce might be stimulated by the rather high (Morocco) or increasing (Turkey) divorce rate in the countries of origin.

Second, an increase in divorce rates in the general Belgian population, which relates to the second demographic transition, has been observed over the course of merely a few decades (Eurostat n.d.a). According to Sandström (2012), there were two broad forces driving the decrease of marital stability. The first is the rise of the dual provider model which resulted in decreased economic interdependence between family members. This is strongly related to the rise of female education and participation in the labour market. It seems that only when women were socio-economically able, divorce became a real possibility for them (Kalmijn and Poortman 2006). The second factor is the increased importance of individualization (Sandström 2012). Individualization refers to the increased power of individuals to make choices in their own interests, sometimes against the will of the broader family, the community and religious institutions. Individualization might result in a decline in the importance of marriage as an institution and higher divorce and lower marriage rates (Cherlin 1992). Belgian legislation developed in line with these evolutions. The divorce legislation changed in 1994 resulting in shorter procedures and weaker conditions (Corijn and Lodewijckx 2009).

The aforementioned trends in Belgium might affect Turkish and Moroccan Belgians as well. In this section, we draw from integration and socialization theories. According to these two theories, a longer socialization period in the destination country is accompanied by a more engrained integration (Alba and Nee 1997; Lievens 1997). This might result in a stronger influence of individualization and female emancipation processes in Belgium on Turkish and Moroccan Belgians, compared to Turkish and Moroccans living in the origin country. Based on qualitative research, Koelet et al. (2009b, p. 180) observe that Moroccan and Turkish spouses stress the importance of their own autonomy and happiness in the decision to divorce. While family and partners continue to be important actors in this decision, this does suggest an increased individualization.

The rise of individualization is visible in the partner-selection process as well. Van Kerckem et al. (2013) observe for Turkish Belgians a higher acceptance of premarital relationships with intra-ethnic partners from the local community in Belgium. Moreover, the same study identified a declining parental involvement. While in the 1980s parental interference was large, nowadays, parents are in many cases merely matchmakers.

In addition, some observations signal increasing female emancipation amongst Turkish and Moroccan Belgian women. A key issue is the fact that girls from Turkish and Moroccan descent outperform boys in terms of educational achievement (Timmerman 2006). This does not necessarily guarantee favourable conditions after education, but it does empower these women to some extent. According to Timmerman (2006), the poor school and labour market achievement of Turkish boys negatively affects their status as it makes it difficult to fulfil their traditional role as economic provider. Additionally, qualitative research by Koelet et al. (2009a) suggests that there is a rising awareness and contestation amongst Turkish and Moroccan women of gender–power imbalances. This results, for example, in an increasing critical attitude against men’s lack of responsibility and initiative in running household affairs. The increase in emancipatory attitudes as well as economic independence of women might be important motives underlying the choice to divorce (De Graaf and Kalmijn 2006; Kalmijn et al. 2004; Poortman and Kalmijn 2002), particularly in unhappy marriages (Sayer and Bianchi 2000; Schoen et al. 2002). As a result, hypothesis 1 states that divorce rates of marriages established by Turkish and Moroccan Belgians rise over time (H1).

Divorce and Differences Between Turkish and Moroccan Migrant Communities

Although divorce rates may increase for Turkish and Moroccan Belgians, divorce risks might be different for both groups. This difference was already apparent in lower fertility levels amongst Moroccan women compared to Turkish women in Belgium (Schoenmaeckers et al. 1999). The distinct nature of Turkish and Moroccan migration and different levels of individualization and female emancipation in the country of origin might account for this.

Bilateral agreements between the governments of Belgium on the one hand and Turkey and Morocco on the other hand in 1964 formalized the first large inflow of Turkish and Moroccan labour migrants to Belgium (Lievens 2000; Reniers 1999; Schoonvaere 2013, 2014; Surkyn and Reniers 1997). The greater part of Turkish men who settled in Belgium was already married before migration (Schoonvaere 2013). Consequently, Turkish migration fitted within the strategy of the family as a conservative form of migration permitting the household to keep on living in its community of origin (Reniers 1999; Surkyn and Reniers 1997). After the stop on labour recruitment, however, commuting between Belgium and Turkey was no longer possible and the remaining family members moved to Belgium as well.

Within Moroccan communities, the impact of more individualistic, emancipatory and socio-culturally motives was stronger from the onset of migration, making Moroccan men less inclined to return to Morocco (Lesthaeghe 2000; Reniers 1999; Surkyn and Reniers 1997). In addition, the integration of Moroccan Belgians was facilitated due to their linguistic proficiency, as French is their second language (Lesthaeghe 2000).

In addition to the strong emphasis on the family in the case of Turkish Belgians, studies point to different active recruitment strategies in Turkey and Morocco. Labour migrants from Turkey predominantly originate from rural areas that are characterized by stronger community ties (Lievens 2000; Surkyn and Reniers 1997). These strong ties and networks resulted in transplanted communities in the country of destination (Surkyn and Reniers 1997; Reniers 1999). In Morocco, these processes were present to a lesser extent given that recruitment was more oriented on bigger cities and capitals of provinces. Consequently, the Moroccan communities are more fragmented; this heterogeneity concerns the marital status, educational level, religious–political orientation and socio-economic status at arrival as well (Lesthaeghe 2000; Lesthaeghe et al. 2000; Reniers 1999).

The different recruitment strategies combined with a weaker orientation on the family resulted in weaker social control, a more individualistic orientation and frailer ties with the country of origin within Moroccan communities (Lesthaeghe 2000; Lesthaeghe et al. 2000; Reniers 1999; Surkyn and Reniers 1997). This has been reflected in partner preferences as well. Although a transnational marriage is the dominant partner choice for both Turkish and Moroccan Belgians, transnational marriages are slightly less prevalent amongst Moroccan Belgians while mixed marriages are more prevalent (Dupont et al. 2017b). This, combined with higher divorce levels in Morocco (Koelet et al. 2009a), might increase the influence of individualization and female emancipation processes on Moroccan Belgians and might explain the higher chance of divorce for Moroccan Belgians in the past decades (Eeckhaut et al. 2011; Corijn and Lodewijckx 2009). Therefore, we hypothesize that divorce rates amongst Moroccan Belgians are higher compared to Turkish Belgians (H2).

Marriage Type and Divorce

Divorce rates might differ between different marriage types. These marriage types include transnational marriages (i.e. marrying a partner from Morocco or Turkey), local intra-ethnic marriages (marrying another Moroccan of Turkish Belgian) and mixed marriages (i.e. marrying someone with a Belgian or other Western–European citizenship). Particularly cultural differences (heterogamy) and social support (often supplemented with social control) are crucial (Eeckhaut et al. 2011). First, we will substantiate the claim of the importance of cultural differences and social support. Second, we will apply the concepts of cultural differences and social support to the different marriage types.

Cultural Differences and Social Support

In general, people prefer a partner who is similar to them, a phenomenon called homophily (McPherson et al. 2001). Important similarities include age, socio-economic status and the cultural background. Ethnic homogenous marriages tend to be more stable compared to heterogamous marriages, resulting in lower divorce rates (Janssen 2002; McPherson et al. 2001; Kalmijn et al. 2005). However, in this paper we argue that within transnational marriages-, which are ethnically homogenous marriages, cultural differences are likely to exist as well.

The role of third parties, such as friends and family, can affect migrants’ partner choice by their social support (Huschek et al. 2011; Kalmijn 1998). This influence is particularly pronounced in migrant communities characterized by a strong internal cohesion and a high level of social control. In these communities, marriage and divorce are in many cases strongly collectivized, both families are involved, not only the spouses (Merz et al. 2009; Timmerman 2006). Evidence for higher levels of parental involvement amongst Turkish and Moroccan immigrants was found in the Netherlands (van Zantvliet et al. 2014), Germany (Straßburger 2003), France (Milewski and Hamel 2010) and Belgium (Descheemaeker et al. 2009). This role can vary from merely giving advice, to playing a matchmaker, to even completely controlling the choice of partner (Kalmijn 1998).

Cultural Differences, Social Support and Marriage Types

In many cases, parents, or the ethnic community in general, stimulate intra-ethnic marriages to preserve the internal cohesion and homogeneity of the group (Kalmijn 1998; Van Pottelberge et al. 2018). Local intra-ethnic marriages are therefore strongly supported by the family and the broader community. In general, the level of cultural similarity is the highest for these marriages, given that both partners share the country of origin and are socialized in the destination country. Due to these factors, it is not surprising that this type of marriage showed the lowest divorce rates in the nineties (Eeckhaut et al. 2011).

Cultural differences between the partner coming from the country of origin and the partner residing in Belgium, such as differences in norms and expectations concerning gender roles, might put strain on transnational marriages (Eeckhaut et al. 2011; Timmerman et al. 2009; Van Kerckem et al. 2013). These marriages are more precarious although they are strongly supported by the family and the local ethnic community (Eeckhaut et al. 2011; Timmerman et al. 2009; Van Kerckem et al. 2013). Apart from cultural differences, many more (related) problems may emerge. A first issue concerns the uncertainty about the motives of the migrated spouse because an individual can never be certain whether the partner marries for ‘true love’ or because of other, often less romantic, motives (Eeckhaut et al. 2011; Timmerman 2008). In addition, accurate knowledge about the situation in the destination country and Western–European countries in general is in many cases lacking, which might result in a distorted idealized image within the country of origin. Due to these reasons, Turkish and Moroccan Belgians might develop a careful attitude towards transnational marriages (Van Kerckem et al. 2013). This implies that individuals involved in a transnational marriage that does not fit their standard of marital life may more easily find support to divorce.

Finally, mixed marriages are characterized by both cultural differences and a lack of social support. Cultural differences between partners are likely to put barriers between them (Eeckhaut et al. 2011; van Huis and Steenhof 2004) and increase tensions (Dribe and Lundh 2012). Religion is an important source of cultural differences. Turkish and Moroccan Belgians are Muslim, while native Belgians are mostly Christian, although religious adherence is not very strong for native Belgians (Lesthaeghe and Lopez-Gay 2013). In addition to religion, different views on family life might be important sources of tension (Van Kerckem et al. 2013). In addition to cultural differences, the support from family and social networks is lower in mixed unions (Dribe and Lundh 2012; Eeckhaut et al. 2011). This negative attitude is rooted in processes of ethnic conformity pressure and feelings of not being ‘real Belgians’ (Saroglou and Mathijsen 2007) and is in many cases related to the language barrier between the parents of the spouses as well. In addition, migrants in Belgium, and Muslim individuals in particular, suffer from a negative image (Van der Bracht et al. 2013) which might make native Belgians more reluctant to establish mixed marriages as well. As a consequence, we predict that mixed marriages have higher divorce rates compared to local intra-ethnic marriages, while transnational marriages take up a middle position (H3).

Divorce and Gender

As previously stated, marriage and divorce are strongly collectivized in Turkish and Moroccan communities. Particularly women might be constrained in their partner choice, mainly because of patriarchal traditions that stress the importance of virginity and honour of women (Hooghiemstra 2003). Moreover, divorce might result in reputation damage (Welslau and Deven 2003), making women less inclined to divorce compared to men. Additionally, divorce rates of men and women might differ in their partner choice as well.

Women might prefer a man from the country of origin because of the freedom and autonomy it offers while men might prefer such a partner to establish their dominance (Beck-Gernsheim 2007; Hooghiemstra 2001; Lievens 1999). As discussed previously, there are observations signalling increasing female emancipation (Timmerman 2006) and a growing awareness and contestation of gender–power imbalances amongst Turkish and Moroccan Belgian women (Koelet et al. 2009b). Timmerman (2006) stated clearly, ‘girls see in Islam opportunities to challenge the male dominance in their communities where men invoke Islam rather to preserve the traditional male supremacy’.

However, a partner arriving in Belgium within the context of a marriage is in many cases in a dependent position due to the lack of family ties, the absence of linguistic proficiency and cultural knowledge, financial hardship, etc. (Beck-Gernsheim 2007; Surkyn and Reniers 1997). Particularly if the migrating partner is male, and the newly formed family had to move in with the in-laws, the lacking economic autonomy is considered a disgrace because it contradicts the patrilocal tradition that the newly formed family moves in with the family of the husband.

The problems within transnational marriages are less likely to result in divorce if the marriage migrant is female. For persons living in Belgium in 2004, Corijn (2009) finds that matches involving a Moroccan or Turkish female marriage migrant have lower divorce chances compared to a male marriage migrant. Based on the finding of Koelet et al. (2009b) that female marriage migrants often experience strong abuse, Corijn (2009) relates this lowered chance of divorce to the limited possibilities to escape from severe marriage problems. As a consequence, hypothesis 4a states that divorce rates of transnational marriages established by Turkish or Moroccan Belgian men with a female marriage migrant from Turkey or Morocco are lower compared to transnational marriages established by Turkish or Moroccan Belgian women with a male marriage migrant (H4a).

Studies indicate that migrant men are more likely to establish mixed marriages compared to women (Dupont et al. 2017b; Hooghiemstra 2001; van Tubergen and Maas 2006). A possible explanation refers to the socio-economic resources of men and women. As men’s resources remain more important and prestigious, marrying a native woman might be a form of status exchange (Kalmijn 1998). A second possible explanation is related to patriarchal traditions that constrain women in their partner choice (Hooghiemstra 2001). Resilience exists against Muslim women marrying non-Muslim men because the religion of the father passes on to the children. These children are for this reason considered as being lost for the family and Islam (Kulzycki and Lobo 2002).

Given the opposition to female mixed marriages, these marriages might be established more better considered. Furthermore, according to the ‘gender difference hypothesis’ (Dribe and Lundh 2012) the effect of heterogamy on divorce might be strongest for immigrant men. In their study in Sweden, Dribe and Lundh (2012) claim that ‘since many immigrant groups differ from Swedish norms when it comes to female employment, division of housework and gender roles, mixed marriages can be expected to have different implications for men and women’. The traditional gender roles of migrant men and their lacking socio-economic resources (Kalmijn and Poortman 2006; Koelet et al. 2009b; Pels 2000) are difficult to reconcile with the native woman’s attitudes and hence create tension and disagreement (Dribe and Lundh 2012). The value dissimilarity between a native man and an immigrant woman is less strong, however, and for this reason less destabilizing. Previous research for the Netherlands (Smith et al. 2012) and Germany (Milewski and Kulu 2014) has shown that Turkish and Moroccan men who marry native partners have a much higher chance of divorce compared to Turkish and Moroccan women who marry natives. Therefore, we hypothesize that divorce rates of mixed marriages are lower for migrant women compared to men (H4b).

Data and Method

Data

Our data comprise an extraction of the Belgian national register, containing all partnerships (i.e. marriages and legally registered cohabitations) of migrants originating from third countries5 formed between 01 January 2001 and 31 December 2008. Two conditions were required for inclusion: (1) at least one partner was a resident in Belgium before the partnership and (2) with a third-country nationality at birth. Because we do not solely focus on current divorce rates, but compare them to divorce rates of marriages established between 01 January 1988 and 31 December 1990 as well, we try to resemble the data from the analyses in Eeckhaut et al. (2011) as accurately as possible. For both the descriptive and event-history analyses, Eeckhaut et al. (2011) selected a 3-year marriage cohort (first marriages between 01 January 1988 and 31 December 1990 of Turkish and Moroccan Belgians, which was followed until 31 December 1995 or until the marriage dissolved. The maximum duration of marriages established in 1988 was 8 years, while for those established in 1990, and the maximum duration was only 6 years. For our descriptive analyses, we select a 3-year marriage cohort (first married between 01 January 2001 and 31 December 2003, excluding legal cohabitations) as well of Turkish and Moroccan Belgians, which is followed until 31 December 2008. The maximum duration of marriages established in 2001 is 8 years, and for marriages established in 2003 only 6 years.

For our event-history analyses, we select all marriages of Turkish and Moroccan Belgians taking place between 01 January 2001 and 31 December 2005, which are followed until 31 December 2011 or until the marriage dissolved. In the event-history analyses, we will only compare the relative divorce rates of the marriage types, rather than focusing on the comparison through time.

We only take opposite sex marriages into account due to the low prevalence of same sex marriages (2% and 1% for Moroccan and Turkish Belgians, respectively). Legal cohabitations were excluded as well, given their occurrence in only 1% of all partnerships in the dataset.

Variables

To maximize the comparability to data used in Eeckhaut et al. (2011), we operationalize our variables as similarly as possible. The only exception is the level of education, which is missing from our dataset.

Divorce status varies by time and is a dichotomous variable (0 = Not divorced, 1 = Divorced) indicating whether a marriage dissolved in a given year or not. If a marriage dissolved in a given year, the divorce status will be ‘1’ only for that year. Previous years are designated as ‘Not divorced’.

Marriage duration reflects the length of time from the marriage until divorce or censoring: marriage still intact in 2011. The minimum duration of the marriage is zero years (married and divorced in the same year), and the maximum duration is 10 years (married in 2001 and divorced in 2011).

Marriage type is categorical variable consisting of five categories that combines gender (man or woman) with the three possible marriage types. Divorce rates are situated at the couple level rather than the individual level. As a consequence, local intra-ethnic marriages are included only once to avoid double-entry of the same marriage, given that both partners are Turkish or Moroccan residents in Belgium, while transnational and mixed marriages are included separately for men and women. The following five categories are distinguished: a Moroccan or Turkish Belgian marrying a partner from the country of origin (‘Transnational marriage’), a local intra-ethnic partner (‘Local’) or marrying someone with a Belgian or other Western–European citizenship (‘Mixed’). We used effect coding to compare the effect of each category to the mean for the specific group (Turkish or Moroccan).

Marriage year is a metric variable, based on the year of marriage. With this variable, we control changes through time in partner preferences.

Generation is a categorical variable consisting of three categories (1 = First generation, 2 = 1.5 Generation, 3 = Second generation) based on the country of birth and the age of migration. The first generation consists of migrants who were older than 15 years of age at migration. Respondents who migrated between the ages of 6 and 15 years are designated as 1.5-generation migrants. Finally, respondents who migrated before the age of 6 years or who were born in Belgium are classified as second-generation migrants. This variable is based on the generation of both the ‘resident’ migrant and his/her partner. In the case of transnational marriages and mixed marriages, the generation of the resident migrant is taken. For local intra-ethnic marriages, we include the highest migrant generation.

The age at marriage is a categorical variable consisting of three categories (1 = Young age, 2 = Normal age, 3 = Older age). The cut-off points for each category are based on the quartiles of each subpopulation (based on nationality at birth and gender) (< 0.25, between 0.25 and 0.75 and > 0.75, respectively) (Table 1). We chose a categorical age variable because we are not interested in the absolute age at marriage. The definition of marrying at a young, normal or older age is dependent upon the subpopulation based on the nationality at birth and gender (Lievens 1999). For this reason, the age categories have the same meaning in each of the subpopulations. Similarly to the migrant generation variable, we include the age at marriage of the resident in the case of transnational marriages and mixed marriages. For local intra-ethnic marriages, the youngest age category was used.

Table 1.

Categories ‘age at marriage’

	Moroccan		Turkish
	Men	Women	Men	Women
Young age	< 24	< 20	< 22	< 20
Normal age	24–31	20–26	22–27	20–24
Older age	> 31	> 26	> 27	> 24

N Turkish = 5819; N Moroccan = 10,434

Method

Our descriptive analyses are based on the original person file that contains couple data on marriages. This file consists of 17,786 cases for Moroccan Belgians and 9631 cases for Turkish Belgians. We converted the person file into a person-period file for our event-history analyses (Yamaguchi 1991). The person-period file generates 143,665 and 84,996 person-years, respectively.

In our event-history analyses, we use piecewise constant log-rate models with effect coding to interpret the effect of the variable ‘Marriage type’, which compares the effect of each category to the mean for the specific group (Turkish or Moroccan). The model is built incrementally. First, we specify the baseline of the hazard function. The fourth-order polynomial of the time function fits the data most accurately (figures not shown) and is not correlated with the marriage types (results not shown). This means that the risk of divorce as a function of marital duration is similar across marriage types. After that, we add the core variables (marriage types) and control variables (generation, age and marriage year). The analyses are based on data from the Belgian national register, which constitute population data. In this case, significance testing (i.e. testing the probability of drawing a sample by chance from a population that meets the assumptions of the null hypothesis) is not needed, as the population data are not random samples.

Results

Divorce Rates

We hypothesized that divorce rates of marriages of Turkish and Moroccan Belgians rise over time (H1). Table 2 shows that within Turkish and Moroccan groups 7.2% and 14.6%, respectively, of marriages established between 01 January 1988 and 31 December 1990 ended in divorce by 31 December 1995 (Eeckhaut et al. 2011). These percentages have almost doubled; 13.3% and 23.9% of all marriages established between 01 January 2001 and 31 December 2003 were dissolved by 31 December 2008. We predict higher divorce risks within the Moroccan group as well (H2). The higher divorce risk of the Moroccan group observed by the end of 1995 is still visible in 2008 and is even almost two times higher compared to Turkish Belgians.

Table 2.

Percentage of divorces by 31 December 2008 for marriages established between 01 January 2001 and 31 December 2003

	Turkish	Moroccan
Not divorced (%)	86.7 (92.3)a	76.1 (84.8)
Divorced (%)	13.3 (7.2)	23.9 (14.6)

N Turkish = 5819; N Moroccan = 10,434

^aItalic: Percentage of divorces by 31 December 1995 for marriages established between 01 January 1988 and 31 December 1990, derived from Eeckhaut et al. (2011)

Table 3 displays the prevalence of marriage types (transnational, local intra-ethnic and mixed) established between 01 January 2001 and 31 December 2003 of Moroccan and Turkish men and women. Transnational marriages are the most common marriages amongst both nationalities, followed by local intra-ethnic and mixed marriages. Comparing these numbers to the prevalence of marriage types established between 01 January 1988 and 31 December 1990 (Eeckhaut et al. 2011) reveals that the percentage of transnational marriages has declined from about 43 to 30% for Turkish Belgians and from about 48% to 22% for Moroccan women. For Moroccan men, the already low prevalence of transnational marriages (around 27%) remained stable. While the prevalence of mixed marriages rose (from about 3.5% for both Turkish men and Moroccan women to 4.4%; and from about 8% for Moroccan men to 11%), local intra-ethnic marriages became much more popular. About 32% of all marriages of Turkish Belgians are local intra-ethnic marriages, compared to only 10% in the late nineties. For Moroccan Belgians, these numbers are similar: the prevalence of local intra-ethnic marriages rose from about 14% to 36%.

Table 3.

Percentage of marriage types according to nationality and gender ^a (established between 01 January 2001 and 31 December 2003)

	Turkish	Moroccan
Transnational—men	29.1 (43.3)b	26.9 (26.6)
Transnational—women	31.6 (43.6)	22.1 (48.2)
Mixed—men	4.6 (3.1)	10.6 (7.9)
Mixed—women	2.4 (–)	4.4 (3.7)
Local intra-ethnic	32.3 (9.9)	36.1 (13.6)

N Turkish = 5819; N Moroccan = 10,434

^aThe gender refers to the gender of the Turkish or Moroccan Belgian resident

^bItalic: Percentages of partnership types from 01 January 988 until 31 December 1990, derived from Eeckhaut et al. (2011)

Furthermore, divorce rates have risen and in many cases even doubled for almost every marriage type (Table 4). The only exceptions are the stable divorce rate of mixed marriages established by Turkish men. This suggests that there is a general force underlying the rise of divorce that cannot be reduced to particular aspects of the partner-choice pattern.

Table 4.

Percentage divorces according to nationality, gender^a and marriage type for marriages established between 01 January 2001 and 31 December 2003, followed until 31 December 2008

	Turkish	Moroccan
Transnational—men	15.2 (7.0)b	22.9 (11.9)
Transnational—women	14.2 (7.5)	24.8 (17.2)
Mixed—men	19.2 (19.8)	40.3 (11.1)
Mixed—women	13.4 (–)	25.8 (16.0)
Local intra-ethnic	9.9 (3.6)	19.0 (9.1)

N Turkish = 5819; N Moroccan = 10,434

^aThe gender refers to the gender of the Turkish or Moroccan Belgian resident

^bItalic: Percentage of divorces in 1995 for marriages established between 01 January 1988 and 31 December 1990, derived from Eeckhaut et al. (2011)

The results on the differences between divorce rates of each marriage type will only be briefly discussed in this section, but will be explored more in depth in our event-history models to confirm or reject hypotheses 3 and 4. Our prediction stating that mixed marriages would display the highest divorce rates, followed by transnational and local intra-ethnic marriages (H3), can be partially confirmed. Generally, the lowest divorce rates arise from marriages to local intra-ethnic partners (Turkish Belgians: 9.9%, Moroccan Belgians: 19.0%) while transnational marriages display higher divorce rates (Turkish men: 15.2%, Turkish women: 14.2%, Moroccan men: 22.9%, Moroccan women: 24.8%).

The situation is less clear-cut for mixed marriages. Turkish and Moroccan men within a mixed marriage are more likely to divorce compared to the other groups (19.2% and 40.3%, respectively). However, the hypothesis does not hold for Turkish and Moroccan women within mixed marriages (13.4% and 25.8%, respectively).

Hypothesis 4 predicts lower divorce rates of transnational marriages established by men compared to women (H4a) and lower divorce rates of mixed marriages established by women compared to men (H4b). For both groups, there is only a small gender difference in divorce rates of transnational marriages, which does not confirm hypothesis 4a. The gender difference is stronger for divorce rates of mixed marriages. Mixed marriages established by men display the highest divorce rates. Divorce rates are particularly high for Moroccan men; about 40% of all mixed marriages result in divorce, which confirms our hypothesis.

Results Event-History Models

In this section, we relate the likelihood of divorce to each marriage type. Table 5 displays the relative divorce rates presented as odds ratios. These rates compare the effect of each marriage type to the mean for the specific group (Turkish or Moroccan). For example, mixed marriages established by Turkish men have a 1.74 times higher divorce risk than the average divorce risk of Turkish Belgians. However, we want to compare divorce risks of each marriage type. For this reason, we calculated pairwise comparisons between each marriage type based on the relative divorce rates of all marriages.

Table 5.

Relative divorce rate (Exp. B) according to marriage type, nationality and gender^a for marriages established between 01 January 2001 and 31 December 2005, followed until 31 December 2011^b

	Turkish	Moroccan
Transnational—men	0.89 (0.92)c	0.96 (1.03)
Transnational—women	0.90 (0.92)	0.92 (1.42)
Mixed—men	1.74 (2.5)	1.57 (0.81)
Mixed—women	0.94 (–)	1.01 (1.19)
Local intra-ethnic	0.76 (0.47)	0.72 (0.71)

N Turkish = 84,996; N Moroccan = 143,665

^aThe gender refers to the gender of the Turkish or Moroccan Belgian resident

^bWe controlled for marriage year, migrant generation and age at marriage

^cItalic: Relative divorce rates for marriages established between 01 January 1988 and 31 December 1990 followed until 31 December 1995, derived from Eeckhaut et al. (2011)

In our third hypothesis, we predict that mixed marriages display the highest divorce rates, followed by transnational and local intra-ethnic marriages. Within the Turkish group, local intra-ethnic marriages have the lowest divorce rates. The difference between divorce rates of transnational marriages and local intra-ethnic marriages is rather small. Transnational marriages of both men and women only have a 1.2 times higher likelihood of divorce compared to local intra-ethnic marriages. Mixed marriages of Turkish men on the other hand have a 2 times higher likelihood of divorce compared to transnational marriages and a 2.3 times higher likelihood compared to local intra-ethnic marriages. This confirms the third hypothesis for men. Mixed marriages of Turkish women only have a 1.2 times higher likelihood of divorce compared to local intra-ethnic marriages and a 1.04 times higher likelihood compared to transnational marriages. Therefore, the third hypothesis can be confirmed for Turkish women as well, although the difference in divorce risks between the marriage types is much smaller compared to Turkish men.

Within the Moroccan group, local intra-ethnic marriages have the lowest likelihood of divorce as well. Transnational marriages established by Moroccan men and women have approximately a 1.3 higher likelihood of divorce compared to local marriages. Mixed marriages of Moroccan men have a 2.2 higher likelihood compared to local marriages and a 1.6 higher likelihood compared to transnational marriages. This confirms our third hypothesis for Moroccan men as well. Mixed marriages have a 1.4 times higher likelihood of divorce compared to local intra-ethnic marriages and only a 1.1 higher likelihood compared to transnational marriages which confirms the third hypothesis for Moroccan women as well.

Furthermore, we predict that divorce rates of transnational marriages (H4a) and mixed marriages established by women (H4b) are lower compared to divorce rates of transnational marriage and mixed marriages established by men. Divorce rates of transnational marriages established by Turkish men only have a slightly elevated divorce risks compared to those established by women (1.01), which does not confirm the gender hypothesis for transnational marriages (H4a) amongst Turkish Belgians. Furthermore, mixed marriages of Turkish women have much lower divorce rates compared to men; those of men are 1.9 times higher, which is a confirmation of our gender difference hypothesis for mixed marriages (H4b) amongst Turkish Belgians.

The divorce rates of transnational marriages established by Moroccan men are only 1.04 times higher compared to women. Therefore, the gender hypothesis cannot be confirmed for transnational marriages (H4a). However, the divorce rates of mixed marriages are 1.6 times higher for Moroccan men compared to Moroccan women, which clearly confirms the gender hypothesis for mixed marriages (H4b).

Essentially, the same conclusions can be drawn for the Moroccan group as for the Turkish group: mixed marriages display the highest divorce rates, followed by transnational and local intra-ethnic marriages (H3), and the gender difference hypothesis is confirmed for mixed marriages (H4b) while the difference is negligible for transnational marriages (H4a).

Conclusion and Discussion

In this paper, we investigate the prevalence of divorce amongst Turkish and Moroccan Belgians, with a specific focus on the effect of partner-choice patterns. A few conclusions can be drawn.

First, in the past 15 years, divorce rates have doubled within Turkish and Moroccan migrant groups for all marriage types. In our interpretation, this might have several causes. First, these changes may be related to the second demographic transition, such as the increased importance of individualization that eroded the importance of marriage as an institution, and a rising female emancipation (Cherlin 1992; Kalmijn and Poortman 2006; Sandström 2012). Second, changed household economics might increase the likelihood to divorce as well (Kalmijn, De Graaf and Poortman 2004; Poortman and Kalmijn 2002). Finally, higher divorce rates in the country of origin might stimulate higher divorce rates of migrants (Kavas and Gündüz-Hoşgör 2010; Koelet et al. 2009a). However, given the limitations related to our dataset, we are unable to identify the exact causes of the rise in divorce rates.

Second, divorce rates are much higher amongst the Moroccan group, which might have several causes. High divorce rates in Morocco (Koelet et al. 2009a), the weaker orientation on the family and the lower levels of social control and support within the Moroccan group (Lesthaeghe 2000; Reniers 1999; Surkyn and Reniers 1997) might stimulate divorce. In addition to this, the Moroccan group might be influenced more heavily by changes related to the second demographic transition, or the increasing labour force participation of women might be more pronounced amongst this group.

Third, there are clear differences between marriage types. Local intra-ethnic marriages have the lowest divorce levels, mixed marriages the highest, and transnational marriages take up a middle position. In line with Eeckhaut et al. (2011) and van Huis and Steenhof (2004), this might be explained by cultural dissimilarity within the couple (for mixed and transnational marriages) and lacking social support (for mixed marriages). Local intra-ethnic marriages are the most stable, because they often enjoy parental support (Eeckhaut et al. 2011) and because of the cultural similarities between partners (McPherson et al. 2001).

Fourth, the previous finding needs some qualification. For mixed marriages, there is a clear effect of gender. Turkish and Moroccan men in mixed marriages have much higher divorce rates compared to Turkish and Moroccan women. This is a confirmation of the ‘gender difference hypothesis’ which states that the effect of heterogamy on divorce might be stronger for immigrant men due to value dissimilarity (Dribe and Lundh 2012). However, the gender difference in divorce risks of transnational marriages (particularly for the Moroccan group) that was present in the study of Eeckhaut et al. (2011) has not been replicated in this study. Apparently, the reputation damage following divorce that might influence women in particular (Welslau and Deven 2003) is not strong enough to account for a gender difference in the divorce risks of transnational marriages.

From a more general perspective, the findings of this paper contribute to the idea that there might be a (slow) transition amongst Turkish and Moroccan Belgians. Although we are unable to pinpoint the exact causes of this transition, in this paper we demonstrate that changes in marital behaviour are present. In recent decades, there was a decrease in transnational marriages and an increase in mixed marriages (Dupont et al. 2017a; Van Kerckem et al. 2013). These changes in the marriage pattern are accompanied by a decline of spatial segregation and some upward social mobility (Verhaeghe et al. 2012) which suggest that the barriers between Turkish and Moroccan Belgians on the one hand and native Belgians on the other hand are weakening. These findings show that amongst migrant groups characterized by strong (i.e. collectivistic) family systems processes of individualization and emancipation might be potentially strong as well. This does not mean that these transitions will be identically replicated amongst Turkish and Moroccan Belgians and it does of course not exclude the presence of different trajectories. Particularly amongst Moroccan men of the second generation, a pronounced rise in mixed marriages as well as higher levels of unmarried cohabitation and a stronger tendency to remain single is present (Hartung et al. 2011; Corijn and Lodewijckx 2009). Moroccan men from the second generation might therefore be pioneers of this trend. Yet, as divorce figures for mixed marriages amongst Moroccan men are (very) high, it remains an open question whether other groups will follow their example given the instability of these marriages.

The finding that differences in divorce rates between almost all marriage types have been reduced or remained stable, with the exception of mixed marriages established by Moroccan men, adds more insight to this discussion. The reduced gap might be explained by several factors. For the convergence in divorce rates between intra-local and transnational marriages, a first factor may be that transnational marriages are established more carefully, considering the documented difficulties within these marriages (Van Kerckem et al. 2013). This might result in a slower and more deliberate or cautious marriage formation process. Yet, given that divorce risks of all marriage types have increased, it is hard to explain this evolution only in terms of a more careful approach towards marriage. If anything, one can conclude that transnational marriages do not longer differ that much with intra-ethnic local marriages in this regard. Secondly, higher divorce rates of local intra-ethnic marriages compared to 15 years ago might account for the convergence of divorce rates.

Concerning the convergence of divorce rates of mixed marriages and the other types, a possible explanation may be that mixed marriages receive increasing social support within the involved populations, as reflected by the rise of these marriages (Dupont et al. 2017b). Nevertheless, it is clear that mixed marriages continue to have higher divorce rates. Although social support for these marriages might be increasing, there is still a relative lack of social support compared to the other marriage types. Apparently, the lacking social support and cultural differences continue to play an important role in the prediction of the stability and vulnerability of marriages of Turkish and Moroccan Belgians. Moreover, this convergence is not visible for every group. For Moroccan men, the gap between the divorce risks of mixed marriages and the other marriage types has increased, because the divorce risks of those mixed marriages rose significantly in 15 years. Eeckhaut et al. (2011) attributed the low divorce risks of mixed marriages established by Moroccan men very optimistically to a stronger orientation towards the host country and a less pronounced transnationalism within this group, which led to stronger cultural differences to the country of origin. Although mixed marriages established by Moroccan men rose over time, signalling an integration and individualization process, it accompanied a striking rise in divorce risks of these marriages as well, which might impede integration.

Finally, this paper is subject to several limitations. The first one is related to the automatic naturalization of ethnic minority individuals in Belgium beginning in the early 1990s. As a result, second-generation migrants might be under-represented in our dataset. Since 1984, Belgian nationality has automatically been granted at birth to children born in Belgium to at least one parent with Belgian nationality. Although these individuals are not included in the second-generation group, they could be regarded as belonging to it. In our estimation, however, this is not a substantial problem, as the large-scale naturalization programs occurred only in the early 1990s. Given that the minimum legal age for marriage is 18 years, automatic naturalization eliminates only a small number of individuals from this group.

Second, while our dataset includes the complete population of Turkish and Moroccan Belgians who married and divorced between 01 January 2001 and 31 December 2008, which allows us to bypass the problems of representativeness, attrition, etc., some censoring is inevitable. Concerning the marriages included in the event-history analyses (established between 01 January 2001 and 31 December 2005, followed until 31 December 2011), we are unable to evaluate what happened to 71.2% of all marriages established by Moroccan Belgians, and 83.7% of all marriages established by Turkish Belgians after 31 December 2011. We have to assume that these censored cases are not a selective subgroup and that the censoring occurred independently of the risk of divorce. Within life course research, ‘Because the end of the observation window (…) is normally determined independently from the substantive process under study, this type of right censoring is unproblematic’ (Blossfeld 2001, p. 41).

Third, our dataset lacks some crucial information. There is no information on the ethnic background of migrants. Given that it only includes the nationality at birth, country of birth and current nationality, ethnic differences within nationalities are obscured. Furthermore, there is no information on socio-economic variables as well as the presence of children. The educational level of individual migrants and their work situation are for this reason unknown. This is a major limitation of this study given that women’s employment might destabilize marriages. Moreover, ideational variables are lacking. For this reason, the causes of changes in divorce rates are unknown and might be related to the second demographic transition such as an increasing individualization and emancipation, to changes related to household economics, or to changes in the country of origin. Therefore, further research might compare changes between the migrant population in the country of origin and the country of residence to disentangle the net effect of migration and might incorporate clear measures to measure changes related to the second demographic transition.

In addition, the divorce process remains largely out of sight. Although we can examine divorce rates within different groups in absolute terms and sketch the evolution of those rates in function of the marriage duration in function of time, we cannot know from this research whether the motives underlying divorce have changed over time. We have no knowledge on who initiated the divorce as well. If Moroccan and Turkish Belgians are truly undergoing an individualization combined with an emancipation process, we should note a rising number of divorces initiated by women. Furthermore, while the original data provided information on marriages and legally registered cohabitations, we only focused on marriages. Unregistered cohabitations and single migrants were excluded from the original data. If we assume that Turkish and Moroccan Belgians are influenced by a process of individualization, we should observe a rise in legally registered cohabitations, unregistered cohabitations and single households. Within the timeframe of this paper, legally registered cohabitations were established only sporadically, but were rising, nonetheless.

Funding

This study is part of the project ‘Transnational endogamy in migrant communities in Belgium (2001–2008) which is funded by BOF.

Compliance with Ethical Standards

Conflict of interest

The authors declare that they have no conflict of interest.

Research Involving Human Participants and/or Animals

Not applicable.

Informed Consent

Not applicable.