Like parents, like children? The impact of parental endogamy and exogamy on their children’s partner choices in Sweden

ABSTRACT

This paper analyses the marriage patterns of multi-ethnic people – who have one native-born and one foreign-born parent – born in Sweden (multi-ethnic Swedes). Based on Swedish register data from the period 1997–2016 and multinomial regression analysis, this paper looks into the generational transmission of inter- and intra-marriage for multi-ethnic Swedes versus mono-ethnic individuals who have two native-born parents (mono-ethnic Swedes). It also analyses specific partner choices for multi-ethnic and mono-ethnic Swedes as well as the contribution of other factors to their marriage patterns. We find that the odds of multi-ethnic Swedes marrying individuals with a foreign background are higher than those of mono-ethnic Swedes. Living in one of the three major cities was found to be the strongest predictor among other factors affecting marital patterns. Our results also show that highly educated multi-ethnic and mono-ethnic Swedes are slightly less likely to marry individuals with a foreign background than they are to marry mono-ethnic Swedes.

KEYWORDS:

• Intermarriage
• multi-ethnic
• intergenerational transmission
• exogamy
• Sweden

Introduction

As a result of international migration and globalisation, the number of interethnic and interracial couples, as well as the academic interest in this topic, has increased. Whereas the probability of intermarriage between immigrants and natives, or between native-born people associated with different racial or ethnic groups has been widely examined (see, for example, Andersson, Obućina, and Scott 2015; Bossard 1939; Chiswick and Houseworth 2011; Dribe and Lundh 2008; Elwert 2018; Kalmijn and van Tubergen 2006; Qian and Lichter 2001), few researchers have analysed how parental patterns of inter- and intra-marriage may influence their children’s marital choices. This paper contributes to the literature by looking at the intergenerational transmission of marriage patterns. We analyse the probability of having specific partner choices for native-born individuals with one native-born parent (multi-ethnic Swedes) versus two native-born parents (mono-ethnic Swedes), as well as the contribution of other factors, such as education, to their marriage patterns.¹

The idea of intermarriage as an indicator of integration at the individual and societal levels is well rooted in the intermarriage literature and is often the primary motivation for studies on immigrants’ marriage behaviour. At the societal level, intermarriage has been associated with a higher degree of social cohesion in a geographical area (Giorgas and Jones 2002; Kennedy 1943; Price 1982). At the individual level, intermarriage with native majority groups in Western countries has been considered as a sign of successful integration (see, for example, Alba and Nee 2003; Gordon 1964; Lee and Bean 2004). However, this idea has recently been contested by scholars who question whether structural assimilation necessarily leads to intermarriage and whether the positive view of intermarriage as a promoter of integration is justified (see Rodríguez-García 2015; Safi and Rogers 2008; Song 2009). Moreover, the assimilation framework seems less suitable to the study of multi-ethnic individuals. The assumption of the assimilationist view on intermarriage is that better-integrated immigrants are more likely to deviate from the norm of endogamous marriage and to marry natives. Not only is it difficult to talk of integration with respect to native-born multi-ethnic individuals, but also the norm of marrying endogamously does not exist for them.

Despite these objections to a simplistic association between intermarriage and integration, there is no doubt that the occurrence of intermarriage requires certain societal preconditions such as some degree of acceptance of diversity and interaction among people with different origins (see, for example, Kalmijn 1991). For multi-ethnic persons, diversity of background and exposure to networks of people with different origins is often part of their primary socialisation. Therefore, as argued by Çelikaksoy (2012), it is likely that they show a higher degree of openness when forming intimate social relationships, including marriage. Since multi-ethnic Swedes – as defined in this paper – are the offspring of one immigrant and one native parent, they may be less likely than mono-ethnic Swedes to marry majority population Swedes and more likely to marry immigrants or native-born people with a foreign background.

However, what do we talk about when we speak of intermarriage when referring to multi-ethnic people? As argued by scholars like Rodríguez-García (2015) and Song (2016), one of the main challenges when studying intermarriage results from the fact that analysts from different countries employ different understandings of concepts such as ‘ethnicity’, ‘race’ and ‘immigrant’ and, therefore, what constitutes intermarriage in relation to these concepts is subject to debate within and across societies. This is especially problematic when looking at the marital patterns of multi-ethnic and mixed-raced people, who do not have an obvious fit within the dichotomies ‘immigrant versus native’, ‘white versus visible minority’ or ‘majority versus minority’. Since traditional understandings of intermarriage are not able to explain emergent unions within increasingly diverse societies (Osanami Törngren, Irastorza, and Song 2016), new ways to define and analyse complex patterns of intermarriage – including those of the children of mixed unions – need to be explored. This constitutes both a theoretical and an empirical challenge for scholars who need to make decisions about how to conceptualise and operationalise the marriage patterns of multi-ethnic individuals. Our aim of contributing to this special issue with a quantitative study on multi-ethnic persons’ marriage behaviour is particularly challenging, as neither the outcome nor the predictor are easy to define or measure. However, we believe that we should not limit quantitative research in the social sciences to simple and clearly quantifiable phenomena. Instead, we believe that the use of quantitative methods also contributes to a better understanding of more conceptual issues. Therefore, based on register data, we ask our research questions as follows:

• Are Swedish multi-ethnic individuals more likely to marry other individuals with a foreign background than mono-ethnic Swedes?

• Whom do they marry?

• What other factors influence their marital patterns?

By looking into these specific questions, we also address one of the main subjects of this special issue: the socialising and partnering patterns of multi-ethnic or multiracial individuals compared to those of individuals who are from a single ethnic or racial background. This paper extends previous studies in two ways: (i) by including all first marriages and cohabiting unions registered in Sweden between 1997 and 2016; and (ii) by disentangling the binary category of exogamy versus endogamy into four spousal groups based on country of birth and parental intermarriage. As argued by Song (2016) in relation to the UK context today, binary classifications such as endogamy/exogamy or intra-marriage/intermarriage do not allow the capturing of the diverse kinds of union experienced by multiracial people.

Finally, the importance of intermarriage may lie not only in it being a signal of a cohesively integrated society but also in being a promoter of it. As such, intermarriage is said to weaken the cultural distinctiveness and salience in the future generation and to contribute to the softening of negative attitudes, prejudice and stereotypes against the out-group (Kalmijn 1998). In short, the relevance of this topic goes beyond the personal experiences of those who intermarry and relates to the different levels of integration and social cohesion of increasingly diverse societies. The intergenerational transmission of this marriage behaviour is therefore of particular importance since it shows whether or not intermarriage contributes to social change in the long run.

Previous studies on intermarriage

The literature on intermarriage alludes to three sets of factors to explain the occurrence of intermarriage: individual factors like preference, which may be influenced by, for example, levels of education; structural factors such as the availability of partners of the same origin; and factors related to third parties, including value-systems and norms such as the existence of formal and informal sanctions (Kalmijn 1998).

According to the positive assortative mating hypothesis, there is a tendency among individuals to choose partners with similar individual characteristics such as age, education, income, class, religion and phenotype (Chiswick and Houseworth 2011; Epstein and Guttman 1984; Schwartz 2013). While some of these factors may be affecting personal marital choices simultaneously, depending on the availability of potential partners of similar characteristics, it may also happen that the prevalence of some factors over others leads individuals to cross boundaries. For example, highly educated individuals from a minority group might prefer to partner with highly educated people from the majority group rather than with other minority members with lower education (Blau, Beeker, and Fitzpatrick 1984).

In fact, studies have shown that there is a positive relationship between education and intermarriage for immigrants or their descendants (see Çelikaksoy 2014, 2016; Chiswick and Houseworth 2011; Furtado 2012). These studies explain the higher propensity of university graduates to intermarry as follows: mating preference may change as a result of education, universities constitute meeting points for people from different countries and ethnicities, and highly educated people tend to leave areas with a high concentration of immigrants and ethnic minorities. However, it is not only individuals’ level of education but also the average schooling of their ethnic group that is associated with intermarriage (Kalmijn 2012). Based on the finding that highly educated people belonging to highly educated ethnic groups in Sweden are more likely to intramarry, Çelikaksoy (2016) also concludes that the relationship between education and exogamy is not always so clear.

In contrast to this rather large body of literature on the link between education and intermarriage for immigrants or ethnic minorities, very few studies look at the association between education and the exogamy of native-born majority populations in Sweden or elsewhere (see, for example, Elwert 2016; Haandrikman 2014). Interestingly, they found that education is negatively related to the probability of the native-born to intermarry. In other words, higher education seems to have the opposite impact on exogamy for foreign-born/minority populations and for native-born/majority populations.

The occurrence of intermarriage cannot be explained by exclusively appealing to individual preferences. Structural factors such as the availability of potential partners from the same country or ethnicity are necessary preconditions for marrying within the group. The more that social circles of different ethnic, educational or occupational groups intersect, the more marriage between these groups will occur even when strong in-group preferences exist. This is simply due to the fact that fewer attributes will be concentrated in one person and realising homogamy preferences in one characteristic (for example, education) will mean compromising in other characteristics (for example, ethnicity; Blau, Beeker, and Fitzpatrick 1984).

For migrants with endogamy preferences, the availability of members of their own group of the opposite sex determines whether these preferences can be realised. Migrants’ own age, ethnic or religious group size and sex ratios have been used in the literature as proxies for the availability of potential partners that affect individuals’ probability of intermarrying (Blau, Blum, and Schwartz 1982; for examples of their use, see Gevrek 2009; Meng and Gregory 2005; Meng and Meurs 2009). Understanding multi-ethnic people’s intermarriage behaviour, however, is more challenging, both theoretically and empirically. It is unclear whether individual preferences are stronger for marrying individuals from the native majority or other minorities, and opportunity structures are harder to account for. Using the group size and sex ratios of the immigrant group does not work as the native group is sufficiently large and basically gender balanced (see also Elwert 2016 on the opportunity structure for the native majority population). One strategy adopted by some scholars – probably for the sake of simplicity – treats multi-ethnic individuals as if they were mono-ethnic by ascribing multi-ethnic people their mother or father’s ethnicity (see, for example, Çelikaksoy 2016).

Finally, normative factors affecting endogamy or exogamy include formal norms or laws regulating intermarriage, informal norms based on value-systems in societies of origin and destination, and societal attitudes resulting from formal and informal norms. For example, previous studies show that black–white marriages still represent the smallest proportion of all types of marriage in the United States (US) (Rosenfeld 2008; Yancey and Lewis 2009). While anti-miscegenation laws in the US constitute a well-known example of formal norms sanctioning intermarriage, minority and majority members who intermarried in countries where intermarriage was not officially forbidden were often subject to denigration, abuse and a lowering of their social status (Merton 1941; Twine 2010). Even though Sweden has never had any official anti-miscegenation laws, formal and informal rules that have historically prevented intermarriages among different races and ethnicities may still affect intermarriage patterns long after the official norms have been revoked (Osanami Törngren, Irastorza, and Song 2016). Recent studies show the persistence of hierarchical preferences in the ethnic and racial origins of a potential partner in Sweden (e.g. Elwert 2016; Potârcă and Mills 2015) and the attitudes toward intermarriages are justified through persistent ideas of ‘us’ and ‘them’ (see Osanami Törngren 2011). While third parties may also influence the marriage behaviour of multi-ethnic individuals, it is more difficult to predict the kind of boundaries to which these individuals are subjected.

Several scholars have discussed the attractiveness of men and women of different racial, ethnic and religious affiliations as partners in particular contexts (see, for example, Kempadoo 2004; Lee 2015; Nemoto 2006; Rodríguez-García, Solana-Solana, and Lubbers 2016). In Sweden, several studies have shown higher preferences for dating or marrying members of the majority population compared to immigrants (Jakobsson and Lindholm 2014; Osanami Törngren 2011; Potârcă and Mills 2015; Snellman and Ekehammar 2005). Moreover, all of these studies have shown that a perceived hierarchy exists between the different immigrant groups, among which some origins are considered more attractive than others. In all the studies cited above, immigrants from Arab countries were perceived as the least attractive while immigrants from Western countries typically ranked the highest. These studies show the existence of preferences in the dating and marriage market which may be the result of racial or phenotype preference or may also be due to perceived cultural distance to Sweden. It is worth noting, however, that approximately 80 per cent of the foreign-born parents of multi-ethnic Swedes come from Nordic and other European countries and, therefore, most multi-ethnic Swedes are expected to be phenotypically and culturally very similar to mono-ethnic Swedes. Based on this and the fact that they were born in Sweden to one Swedish parent, we do not expect that multi-ethnic Swedes will be ‘ethnically less appealing’ than mono-ethnic Swedes to the majority population.

Studies on societal attitudes towards intermarriage also refer to prior contacts with outsiders to explain such attitudes. According to these studies, based on the Contact Hypothesis, individuals who have had prior interracial contact or have established friendships over racial and ethnic boundaries are more likely to be positive about intermarriages and also more likely to intermarry (Kalmijn and van Tubergen 2006; Muttarak and Heath 2010; Osanami Törngren 2011). Along similar lines, it has been found that union formation patterns of endogamy and exogamy influence those of the children. For example, in her empirical study conducted on the children of immigrants born in Sweden, Çelikaksoy (2012) reports a positive correlation between parental exogamy and that of their children. Based on a definition of intermarriage as unions in which spouses or their parents do not have any common foreign background, she explains her finding by appealing to the Contact Hypothesis. First, multi-ethnic individuals are expected to be more open-minded towards ethnic diversity, speak more languages and have more multi-ethnic individuals among their friends. Second, she argues that intermarried parents are less concerned about the ethnicity of the partners of their children and, therefore, might be less influential in their marital choices.

Our paper extends Çelikaksoy’s (2012) study as follows: first, her sample is comprised of native-born individuals with at least one foreign-born parent, whereas ours only includes native-born individuals with at least one native-born parent (unlike in her study, mono-ethnic Swedes form the comparison group in our study and not the children of immigrants). We selected these two groups – that is, mono-ethnic and multi-ethnic Swedes – based on the very large phenotypical and cultural similarities between them. Since this is not always necessarily the case with individuals with two foreign-born parents, we decided to leave this group out of the analysis. Second, her definition of multi-ethnic individuals refers to those who have no native-born parents, while our sub-sample of multi-ethnic people is restricted to Swedish multi-ethnics (that is, those who have one Swedish parent). Third, Çelikaksoy’s definition of intermarriage is based on a binary understanding of the concept. As we discussed in the introduction to this paper, we do not believe that dichotomous classifications of intra-marriage versus intermarriage or endogamy versus exogamy are able to capture the wider range of experiences lived by multi-ethnic people. Therefore, instead of using predetermined conceptualisations of intermarriage, we include several options of spousal type based on own and parental countries of origin. These classifications are further explained in the data and method section.

More recent studies on multi-racial parents and mixed families (see, for example, Song 2017) talk about a feeling of commonality among the children of interracial mixed couples. As part of an in-depth qualitative study on the experiences of 62 multi-racial parents and their families in Britain, Song (2017) asked the study participants whether they felt any connection to other multiracial people.² She reports that 28 out of the 62 participants identified with mixed people in general while six did so with people who had the same mixed ancestry as themselves. Song explains the basis for this feeling of commonality as a sense of cultural and ethnic ‘in-between-ness’. This sentiment of affiliation among some multiracial people could potentially result in social network compositions where this population is overrepresented as well as in higher intermarriage rates with other mixed individuals.

Based on the above-cited studies, we expect that the native-born children of intermarried couples will have a higher tendency to have spouses with a foreign background (including other mixed Swedes, immigrants and the children of immigrants who were born in Sweden) than will the children of Swedish endogamous couples.

Immigration to Sweden and patterns of immigrant–native intermarriage

With the exception of a few minority groups like the Sami, the Finns or the Roma, Sweden has traditionally been an ethnically and racially homogenous country (Osanami Törngren and Irastorza Forthcoming). However, as a result of international migration – and especially of non-European migration flows which started in the 1980s – the racial and ethnic landscape of the country has changed significantly. According to data provided by Statistics Sweden (2016a), 18 per cent of the population registered as residents of Sweden in 2016 were born abroad, 5 per cent of people born in Sweden had two foreign-born parents, while 7 per cent of those also born in Sweden were the children of a native- and a foreign-born parent – that is, multi-ethnic Swedes.

Migration flows to Sweden have responded to changes in migration policies and can be classified into three periods. Until the mid-1970s, immigrants were attracted by the high demand for foreign labour, a trend that was enhanced by the gradual liberalisation of immigration policies. People who migrated to Sweden during this period came primarily from neighbouring countries such as Finland, Norway, Denmark and Germany and, to a lesser extent, from Mediterranean countries. In the subsequent period, the oil crisis and the lower demand for labour prompted Sweden to shift towards a more restrictive labour migration policy. As a result, from the mid-1970s until the mid-1990s, immigration flows primarily consisted of refugees and family-reunion migrants from within and outside Europe, including former Yugoslavian countries, Chile and the Middle East. Finally, Sweden’s entry into the EU in 1995 increased migration flows from other EU countries (Lundh and Ohlsson 1999).

The number of multi-ethnic people in Sweden is directly related to the frequency of intermarriage – including non-marital unions – between immigrants and the native population in the parental generation which, in turn, is closely associated to the number of immigrants in the country (Lanzieri 2012a). As the number of immigrants to Sweden increased, so did the intermarriage rates; at the same time, the origins of the immigrant partners have changed according to the composition of immigrants (Cretser 1999). Intermarriage between Swedes and immigrants from other Nordic countries, which was particularly common in the decades following World War II, has decreased over time. Marriage with Finns, particularly between Swedish men and Finnish women, was common in the 1970s and, at the beginning of the 1990s, Finland was still the most common origin for partners of Swedish men and women. Since then, unions with women from Thailand, Poland, Russia and the Philippines have increased (Cretser 1999; Elwert 2018; Haandrikman 2014). The stark increase in marriages with non-European partners is not unique to Sweden but has been found in several European countries such as Switzerland and Spain (de Valk and Medrano 2014).

Between 2008 and 2010, the share of intermarried couples out of all the couples was around 9 per cent in Sweden (Lanzieri 2012b). However, the official statistics as well as the estimates only report on country of birth and thus do not distinguish between natives without a migrant background and the native-born children of immigrants. For mono-ethnic Swedes, the share of marriages with partners who also belong to the Swedish majority declined continuously in the 1990s and the 2000s, while marriages with partners with a foreign background increased – an increase which has been steeper for Swedish men than for Swedish women (Elwert 2018). Intermarriage is, however, selective with respect to immigrant origins. Immigrants from North-Western Europe and North America, for example, have higher intermarriage rates than do immigrants from outside Europe or North America (Behtoui 2010; Dribe and Lundh 2011), which explains the ethnic composition of the multi-ethnic group in our sample.

Data and method

We have already commented on the difficulties of reaching some consensus on how to classify mixed, multiracial or multi-ethnic people, which is crucial for defining the study population and understanding the intermarriage choices of mixed people. As explained by Song (2016), the British Office for National Statistics classifies all self-reported mixed people as one homogeneous ethnic group, while unions between any member of this group and other ethnic groups – regardless of whether or not they have any common ancestry – are considered as inter-ethnic marriages. On the contrary, marriages within the mixed people group do not qualify as interethnic, even when they have no common ethnic background at all. One of the reasons behind this kind of awkward classification relates to the difficulties of quantifying diversity. In fact, whereas the study of the intermarriage patterns of mixed people can be challenging from a qualitative perspective, it is even more so when conducting quantitative research.

In this paper, we define parental exogamy as a union in which one individual was born in Sweden and the other one was born abroad (that is, a binational couple) whereas parental endogamy refers to marriages between two individuals born in Sweden. We use STATIV register data covering the period 1997–2016 to analyse the marriage patterns of multi-ethnic persons in Sweden. STATIV is a longitudinal database for integration studies which contains information on all individuals registered as residents of Sweden and is updated every year. While multi-ethnic Swedes – that is, individuals with one native-born and one-foreign-born parent – represent our main group of interest, ‘mono-ethnic Swedes’ – those with two native-born parents – are included as a reference group.

Our sample is comprised of all 703,192 unions of individuals who were born in Sweden and who fulfilled the following criteria: (i) they were 20 years old or younger in 1997; (ii) they had their first marriage between 1998 and 2016; and (iii) they had at least one native-born parent. Same-sex couples are excluded from our analysis. One of the main advantages of register data is avoiding the sample being biased towards long-lasting marriages (cf. Kalmijn 1993). We restrict the sample to one observation per couple for the year in which the union was first registered. We include only first unions and exclude all individuals who were married, divorced or widowed in 1997 (which is the first year in this data extract for which this information is available). Individuals who were not present (registered) in Sweden in the year prior to the registration of the relationship were excluded from our analysis to avoid the inclusion of relationships that were formed abroad. Due to the restriction of birth cohorts, individuals are at most 39 years old at the time of marriage or union registration. The sample is therefore somewhat biased towards younger ages which, however, it has not been possible to overcome with the data extraction available for this study.

The ten major origin countries for the foreign-born parents of multi-ethnic Swedes included in this study are Finland, Denmark, Norway, Germany, former Yugoslavia, Poland, the United Kingdom and other EU countries as well as Asian and African countries (see the actual distribution for mothers and fathers in Figure 1). Considering that over 80 per cent of the foreign-born parents of multi-ethnic Swedes come from European countries, it is worth emphasising the fact that the largest share of multi-ethnic Swedes will be very similar to mono-ethnic Swedes in their physical appearance. Based on this and the fact that they were born in Sweden to one Swedish parent – and, therefore, are likely to be culturally very similar to mono-ethnic Swedes – we do not expect them to have fewer opportunities to partner with mono-ethnic Swedes; hence, we focus our analysis on the demand side of the story. In other words, if multi-ethnic Swedes are more likely than mono-ethnic Swedes to have spouses with a foreign background, we will understand this trend as being their preference rather than a result of being perceived as less desirable spouses for mono-ethnic Swedes.

Figure 1. Country of birth of the foreign-born parents of multi-ethnic Swedes (%).

Source: STATIV 1997–2016, own calculations.

A set of multinomial logistic regressions are run to predict the odds of multi-ethnic Swedes to partner with one of these four groups: (1) other multi-ethnic Swedes, (2) the native-born children of endogamous immigrant parents (from now on the ‘children of immigrants’) or (3) the foreign-born (from now on ‘immigrants’) as opposed to the reference group of (4) mono-ethnic Swedes. We test whether parental exogamy is correlated to deviations from the default option of marrying into the majority population – that is, mono-ethnic Swedes – and we explore any other factors which might also influence potential deviations.

According to the ‘conceptual model of mixedness’ proposed in the Introduction to this special issue, the experiences of mixed individuals may relate to a greater or lesser extent to the experiences of members of the majority or minority groups depending on a set of individual and contextual factors (see Osanami Törngren, Irastorza, and Rodríguez-García 2021, in this volume). In our case, those experiences refer to the marital choices of multi-ethnic Swedes, which we compare to those of the majority group, that is, mono-ethnic Swedes. Some of the specific factors cited in the Introduction to this volume are gender, age, education, place or residence and time period, which we include as control variables.

To show the influence of these and other control variables on the main coefficient of interest (multi-ethnic Swede), we follow a stepwise approach. Model 1 only includes information on parental exogamy (being multi-ethnic versus being mono-ethnic), Model 2 adds individual demographic and human capital controls (gender, age and education) as well as controls for the partner’s age and education. Education is measured as a dummy variable, indicating whether the individual graduated from university or not. Structural factors (residence in one of Sweden’s three largest urban areas – Stockholm, Gothenburg or Malmö – versus any other municipality) are added in Model 3 to account for the opportunity structure for meeting partners of different backgrounds. Time period, represented as six categories of the year of union registration, is also included in this model. For non-marital cohabitation, period and age at marriage are based on the first year of union registration which, in the Swedish registers, is the year of birth of a common child. The last model, Model 4, includes the same control variables as Model 3 but restricts the sample to individuals and their partners whose (parental) background in terms of country of origin is identifiable in the data. The exclusion of all individuals whose background is only measured in regions of origin leads to a sample size of 664,451 unions. This restriction allows us to explore whether multi-ethnic individuals tend to have partners of the same or a different foreign background as their foreign-born parent (by looking at percentages of foreign background matches/mismatches for multi-ethnic individuals by major source countries).

Based on the large similarities in phenotype and socialisation between multi-ethnic and mono-ethnic Swedes, as mentioned above, and in the absence of the potential influence of the foreign-born parent and parental intermarriage, we would not expect to see differences in partner choices between these two groups. If we did find differences, and depending on the specific spousal types, different factors or a combination of them could be affecting each of these choices. For example, if multi-ethnic Swedes are more likely than mono-ethnic Swedes to marry immigrants or the native-born children of immigrants, this might indicate that they are more familiar with or open to diversity than their counterparts, as argued by Çelikaksoy (2012). If multi-ethnic Swedes are more likely than mono-ethnic Swedes to partner with other multi-ethnic Swedes, this might show not only openness to diversity but also an affinity with individuals based on their shared experiences of being mixed (see, for example, Song 2017) and having grown up in Sweden with one Swedish parent. Finally, in line with Çelikaksoy (2016), if the foreign-born parent of a multi-ethnic Swede was born in the same country as his or her foreign-born spouse or the foreign-born parents of their native-born spouses, this could, in addition, be understood as endogamy.

Table 1 provides descriptive statistics for multi-ethnic and mono-ethnic Swedes. The few differences between the two groups are as follows: the percentage of the population living in Stockholm is higher among multi-ethnic people than among their counterparts (14 versus 10 per cent), whereas there are more university graduates among mono-ethnic Swedes than among the multi-ethnic population (37 and 33 per cent, respectively).

Table 1. Descriptive statistics for multi-ethnic and mono-ethnic Swedes (%).

	Multi-ethnic	Mono-ethnic
Male	44.0	44.8
Mean age at marriage	27.5	27.6
Partner’s mean age at marriage	28.8	28.8
Education (university)	33.0	36.9
Partner’s education (university)	33.5	35.8
Municipality
Stockholm	13.9	9.7
Gothenburg	7.0	6.1
Malmö	4.1	3.1
Other	75.0	81.1
Period
1998–2000	1.7	1.6
2001–2003	5.2	5.1
2004–2006	10.7	11.3
2007–2009	19.1	19.7
2010–2012	25.5	25.3
2013–2016	37.9	37.1
N	68,385	634,807

Source: STATIV 1997–2016, own calculations.

Findings

The cross-tabulation of parental exogamy (that is, if the individual is a multi- or a mono-ethnic Swede) with marital patterns represented in Table 2 shows that marriages with mono-ethnic Swedes are the most common among both mono- and multi-ethnic Swedes, whereas those with the children of immigrants are the least common. This is not surprising considering the differences in the number of potential partners among the four spousal groups included in the study. The table also shows that, as expected, unions with spouses with a foreign background are more common among multi-ethnic than among mono-ethnic Swedes. Interestingly, the smallest gap between the two subsamples is found among those who marry multi-ethnic Swedes.

Table 2. Marital patterns of multi-ethnic and mono-ethnic Swedes, 1998–2016 (%).

Spousal background	Multi-ethnic	Mono-ethnic
Mono-ethnic Swedes	75.4	83.5
Multi-ethnic Swedes	10.4	8.2
Children of immigrants	4.0	2.1
Immigrants	10.2	6.2
Total	100.0	100.0

Source: STATIV 1997–2016, own calculations.

To further explore these patterns, we ran a set of multinomial regression analyses on the probability of multi-ethnic Swedes partnering with other multi-ethnic Swedes, the children of immigrants or immigrants, as opposed to mono-ethnic Swedes. The results of these analyses are reported in Table 3, where the coefficients are shown in their exponential form, as odds ratios.

Table 3. Odds ratios of intermarriage for multi-ethnic versus mono-ethnic Swedes. Standard errors in parentheses.

	Model 1	Model 2	Model 3	Model 4
Spousal group 1: Multi-ethnic Swedes
Multi-ethnic Swede (Ref: Mono-ethnic)	1.40*** (0.02)	1.39*** (0.02)	1.35*** (0.02)	1.32*** (0.02)
Female		0.93*** (0.01)	0.93*** (0.01)	0.92*** (0.01)
Age at marriage		0.99*** (0.00)	0.99*** (0.00)	0.98*** (0.00)
University education		0.95*** (0.01)	0.91*** (0.01)	0.90*** (0.01)
Partner’s age at marriage		1.01*** (0.00)	1.01*** (0.00)	1.01*** (0.00)
Partner’s university education		0.88*** (0.01)	0.84*** (0.01)	0.82*** (0.01)
Municipality (Ref: Other)
Stockholm			1.62*** (0.02)	1.40*** (0.02)
Gothenburg			1.30*** (0.02)	1.17*** (0.02)
Malmö			1.44*** (0.03)	1.30*** (0.04)
Period (Ref: 1998–2000) (2001–2003			0.94 (0.04)	0.93 (0.04)
2004–2006			0.93* (0.03)	0.92* (0.04)
2007–2009			0.96 (0.03)	0.95 (0.04)
2010–2012			0.99 (0.03)	0.97 (0.04)
2013–2016			1.01 (0.04)	0.99 (0.04)
Constant	0.10*** (0.00)	0.10*** (0.00)	0.12*** (0.01)	0.11*** (0.01)
Spousal group 2: Children of immigrants
Multi-ethnic Swede (Ref: Mono-ethnic)	2.12*** (0.02)	2.08*** (0.04)	2.00*** (0.04)	1.85*** (0.05)
Female		1.02 (0.02)	1.01 (0.02)	0.98 (0.02)
Age at marriage		0.97*** (0.00)	0.97*** (0.00)	0.97*** (0.00)
University education		0.78*** (0.01)	0.73*** (0.02)	0.72*** (0.02)
Partner’s age at marriage		1.03*** (0.00)	1.03*** (0.00)	1.03*** (0.00)
Partner’s university education		0.80*** (0.02)	0.74*** (0.02)	0.72*** (0.02)
Municipality (Ref: Other)
Stockholm			1.82*** (0.05)	1.70*** (0.05)
Gothenburg			1.68*** (0.05)	1.51*** (0.05)
Malmö			1.91*** (0.08)	1.72*** (0.08)
Period (Ref: 1998–2000)
2001–2003			0.92 (0.06)	0.90 (0.06)
2004–2006			0.87* (0.05)	0.85** (0.05)
2007–2009			0.80*** (0.05)	0.77*** (0.05)
2010–2012			0.77*** (0.04)	0.74*** (0.04)
2013–2016			0.76*** (0.04)	0.72*** (0.04)
Constant	0.03*** (0.00)	0.03*** (0.00)	0.03*** (0.00)	0.03*** (0.00)
Spousal group 3: Immigrants
Multi-ethnic Swede (Ref: Mono-ethnic)	1.83*** (0.03)	1.84*** (0.03)	1.76*** (0.02)	1.68*** (0.03)
Female		1.02 (0.01)	1.01 (0.01)	0.89*** (0.01)
Age at marriage		0.99*** (0.00)	0.98*** (0.00)	0.97*** (0.00)
University education		0.88*** (0.01)	0.81*** (0.01)	0.82*** (0.01)
Partner’s age at marriage		1.05*** (0.00)	1.04*** (0.00)	1.05*** (0.00)
Partner’s university education		1.46*** (0.02)	1.34*** (0.02)	1.25*** (0.02)
Municipality (Ref: Other)
Stockholm			1.89*** (0.03)	1.69*** (0.03)
Gothenburg			1.71*** (0.03)	1.62*** (0.04)
Malmö			2.29*** (0.05)	2.18*** (0.06)
Period (Ref: 1998–2000)
2001–2003			0.98 (0.05)	0.97 (0.05)
2004–2006			0.90* (0.04)	0.87** (0.04)
2007–2009			0.91* (0.04)	0.87** (0.04)
2010–2012			0.95 (0.04)	0.90* (0.04)
2013–2016			0.94 (0.04)	0.87** (0.04)
Constant	0.07*** (0.00)	0.03*** (0.00)	0.03*** (0.00)	0.03*** (0.00)
N	703,192	703,192	703,192	664,451

Source: STATIV 1997–2016, own calculations.

Our findings confirm the marital patterns described in Table 2. Model 1 shows that, as expected, multi-ethnic Swedes are more likely to have a partner with a foreign background (other multi-ethnic Swedes, children of immigrants or immigrants) than are mono-ethnic Swedes. The largest difference in the odds of multi-ethnic versus mono-ethnic Swedes for marrying within these three groups, as reported in Model 1, is found for those partnering with the children of immigrants (2.1 versus 1.4 and 1.8, respectively, for those who marry multi-ethnic Swedes and immigrants). In other words, multi-ethnic Swedes are about twice as likely to marry first- and second-generation immigrants as are mono-ethnic Swedes. As we showed in Table 2, this gap is slightly smaller between multi-ethnic and mono-ethnic Swedes marrying other multi-ethnic Swedes.

The contributions of additional socio-demographic and human capital variables are presented in Model 2. The coefficients for the variable ‘multi-ethnic Swede’ remain basically unchanged. The literature highlights higher education as one of the most significant factors affecting immigrants’ and minority populations’ intermarriage with natives (see, for example, Çelikaksoy 2014, 2016; Chiswick and Houseworth 2011; Furtado 2012). On the contrary, the very few studies linking education to exogamy for majority populations conclude that it has the opposite impact on their probability of intermarriage (Elwert 2016; Haandrikman 2014). The educational level of both spouses is included in our model as a binary for university education or lower.

All the coefficients of the educational variables are statistically significant. It is, however, worth noticing that the size of the coefficient is rather small for most groups, especially for the first spousal category of multi-ethnic Swedes, with the exception of the immigrant spouses. Mono- and multi-ethnic Swedes with a university degree are less likely to marry any of the spousal groups compared to the reference group of mono-ethnic Swedes. Considering that 90 per cent of our sample is comprised of mono-ethnic Swedes and the other 10 per cent are multi-ethnic Swedes (and, therefore, there are no first- or second-generation immigrants in the sample of reference persons), this finding supports previous studies on the negative association between education and intermarriage for natives.³

As for the education of the spouse, having a partner with a university degree decreases the probability of marrying a multi-ethnic Swede or a child of immigrants while the opposite is true for those marrying immigrants. This association is likely to be related to compositional differences between the groups and selection into intermarriage. The finding that having a partner with a university degree is positively associated with intermarriage confirms previous studies on the positive association between immigrants’ education and intermarriage with natives (see, for instance, Kalmijn and van Tubergen 2006; Lichter, Qian, and Tumin 2015; on Sweden, see Dribe and Lundh 2008). This finding also seems to be in line with previous research, which has found patterns of educational heterogamy in exogamous couples where members of minority groups often have higher education than their majority group partner (Guetto and Azzolini 2015; Gullickson 2006; Gullickson and Torche 2014; Kalmijn and van Tubergen 2006).

In Model 3, we include the last set of control variables (structural and time-related). Living in one of the three main urban areas of Sweden turned out to be a stronger predictor for multi-ethnic and mono-ethnic Swedes’ probability of coupling with individuals with a foreign background than education: people who live in any of the three major cities in Sweden are more likely to have spouses with a foreign background. While this is the case for all three spousal groups tested in the model, there are some differences between them: living in Stockholm slightly increases the odds of marrying a multi-ethnic Swede by 1.6 as compared to the reference group ‘other municipalities’ but people who live in Malmö have the highest probability of partnering with immigrants and their children (2.3 and 1.9 times more than the reference group). The differences are smaller, though, among those who marry the children of immigrants. Despite Stockholm having the largest population of immigrants, Malmö has the largest per capita immigrant population (Statistics Sweden 2016b). Moreover, because Malmö is smaller in area and population than Stockholm or Gothenburg, the segregated areas are closer to the more central parts of the city, which could encourage mobility and interaction. As argued by Song (2009), not only does living in larger, cosmopolitan cities as opposed to smaller ones matter for increasing or decreasing opportunities to interact for different groups but the level of segregation within cities also does.

The rest of the control variables included in the analysis were statistically not significant or the size of the coefficients was too small to have an impact on multi- and mono-ethnic individuals’ probability of having a spouse with a foreign background. It is worth noting that the differences in marital patterns between multi-ethnic and mono-ethnic Swedes, as reported in Table 3, Model 1, decrease only slightly after controlling for human capital, socio-demographic and structural factors in Models 2 and 3. In other words, the control variables included in our models do not explain such differences.

Our last model restricts the sample to individuals and their partners whose ancestry is identifiable in the data as country of parental origin rather than only in regions. It was included in order to further explore the extent to which multi-ethnic individuals have partners of the same background as their foreign-born parent. The coefficients of the regression analysis do not differ substantially from those presented in Table 3, Model 3 and, therefore, do not need further explanation.

Overall, our findings confirm the marital patterns described in Table 2. Multi-ethnic Swedes are more likely to have a partner with a foreign background (other multi-ethnic Swedes, children of immigrants or immigrants) than are mono-ethnic Swedes. Individual and structural controls change this association only slightly. Moreover, being multi-ethnic versus mono-ethnic is the strongest predictor for being in a union with a partner with a foreign background – along with the municipality of residence.

The results of the exploratory analysis to see whether there are endogamy preference patterns are depicted in Figure 2, which includes eight groups of multi-ethnic Swedes by major parental source countries (four European and four non-European). It also shows whether their partners have a shared foreign background, a different one or a Swedish mono-ethnic background. The spouses can be first-generation immigrants, the children of immigrants or other multi-ethnic Swedes.

Figure 2. Foreign background match between multi-ethnic Swedes and their partners by European and non-European major source countries.

Source: STATIV 1997–2016, own calculations.

Figure 2 shows that unions with mono-ethnic Swedes are less common among multi-ethnic Swedes of non-European ancestry than among multi-ethnic Swedes whose parents came from Europe. Multi-ethnic Swedes with Turkish and Chilean backgrounds have the lowest marriage rates with mono-ethnic Swedes while those with a German or a Finnish parent have the highest rates. Among all multi-ethnic Swedes who have spouses with a foreign background, only one-third of them share some ancestry with their spouses. Interestingly, it is the mixed Finnish-Swedes who show the highest proportion of intra-group unions among those who have partners of foreign background (47 per cent). The fact that the Finnish constitute the largest foreign ethnic group in Sweden and, therefore, that the pool of potential spouses is larger among them, probably explains this trend. The Chilean-Swedes and Turkish-Swedes follow them with approximately 30 per cent of intra-group unions among those who do not marry mono-ethnic Swedes. Finally, the German-Swedes (6 per cent) and the Iranian-Swedes (15 per cent) have the lowest rates of intra-group unions among mixed individuals who do not marry mono-ethnic Swedes.

In sum, while intra-group unions vary among multi-ethnic Swedes of different origins, overall only one-third of those who marry other individuals of foreign background form intra-group unions. Hence, our exploratory analysis does not provide any evidence to support the endogamy preference patterns for multi-ethnic Swedes. Furthermore, the results of our regressions show that multi-ethnic Swedes are more likely than mono-ethnic Swedes to marry, firstly, the native-born children of immigrants, secondly, immigrants and, thirdly, other multi-ethnic Swedes than they are to partner with mono-ethnic Swedes. These findings support the hypothesis that they might do so because they are more open to diversity than their counterparts. On the contrary, as multi-ethnic Swedes are not more likely than mono-ethnic Swedes to partner with other multi-ethnic Swedes to a greater extent than they are to marry first- or second-generation immigrants, our findings do not confirm previous studies suggesting the existence of a sense of commonality with multi-ethnic individuals based on their shared experiences of being mixed.

Conclusion

This paper has compared the marital patterns of multi-ethnic and mono-ethnic Swedes by looking at the association between parental endogamy and exogamy and specific partner choices (namely, other multi-ethnic Swedes, immigrants and the children of immigrants) for the two comparison groups. We used register data on first union formation between 1998 and 2016 to answer the following research questions: Are multi-ethnic Swedes more likely to marry other individuals with a foreign background than mono-ethnic Swedes? Whom do they marry? What other factors influence their marital patterns?

We found that the odds of multi-ethnic Swedes marrying individuals with a foreign background are indeed greater than those of mono-ethnic Swedes. This finding confirms our hypothesis and is consistent with previous studies on the intergenerational transmission of intermarriage (see, for example, Çelikaksoy 2012). Furthermore, we have explained this pattern by appealing to the literature on societal attitudes towards intermarriage, according to which individuals who have had prior interracial or interethnic contact – including friendships – are more likely to be positive about intermarriages and to intermarry (Kalmijn and van Tubergen 2006; Muttarak and Heath 2010; Osanami Törngren 2011). Our regression analysis has also shown that the largest difference in the odds of marrying immigrants or their descendants between these two groups is observed for people partnering with the children of immigrants in the first place, secondly, with immigrants and finally with other multi-ethnic Swedes. The pool of potential spouses within each group probably explains the minor differences behind these trends. Concerning the third question, living in one of the three major cities was found to be the strongest predictor among other factors affecting multi-ethnic and mono-ethnic marital patterns: while living in Stockholm increases the odds of partnering with multi-ethnic Swedes, residing in Malmö has the same effect for unions with immigrants. We argued that perhaps the fact that Malmö has a higher per capita number of immigrants and is a smaller and less segregated city might explain this finding.

While highly educated mono- and multi-ethnic Swedes are less likely to have spouses of foreign background, the evidence is less conclusive when it comes to the impact of the education of their spouse on their probability of being in one of the four spousal types that describe our dependent variable. Having a partner with a university degree decreases the probability of marrying multi-ethnic Swedes or the children of immigrants while the opposite is true for those marrying immigrants. Previous studies have shown that education has a different impact on patterns of endogamy and exogamy for majority versus minority populations (as discussed in the previous research section). Marriages between mono- and multi-ethnic Swedes – the majority of whom belong to the White majority population – marrying other multi-ethnic Swedes or the native-born children of (mostly European) immigrants might be classified as exogamy or endogamy depending on how one categorises multi-ethnic people. Because of this conceptual and methodological issue, we are not able to conclude whether or not the association between spousal education and the likelihood of marrying multi-ethnic Swedes or the children of immigrants supports the literature presented above.

This reflection brings us back to the theoretical question we discussed in the introduction to this paper: What constitutes intermarriage for multi-ethnic individuals? Whereas attempting to answer this question is beyond the scope of this empirical paper, our findings – which inform about who multi-ethnic people couple with and what factors affect their spousal choices – contribute to a greater understanding of this understudied population. In sum, we have shown that multi-ethnic Swedes are more likely to marry individuals with a foreign background. We conclude that our results can be interpreted as an indication of a positive association between early exposure to diversity within people’s own families – and probably also within family networks – and openness to otherness in intimate relationships such as intermarriage, as argued by other scholars (see, for example, Çelikaksoy 2012). Furthermore, our additional exploratory analysis of the ethnic choices of multi-ethnic Swedes who couple with individuals with a foreign background showed that most of them do so outside their group and, therefore, the idea of the endogamy preferences of multi-ethnic individuals is not supported by our study.

While this study is an important contribution to the scarce quantitative literature on multi-ethnic individuals’ marriage behaviour, there are several limitations. Most importantly, the relatively crude measures of local residence presumably do not fully capture opportunity structures for meeting and dating. Small units of residence within cities and other areas would allow for a more elaborate analysis of the importance of opportunity structures for the marriage patterns of multi-ethnic people. Furthermore, as described above, due to data limitations our sample is somewhat biased towards younger ages. This may have an impact on our results, since marriage patterns may be different among older individuals. We have also restricted the sample to first unions. Future research should explore whether patterns of union formation and partner choices differ for multi-ethnic individuals in higher-order unions. Finally, we acknowledge that the group of multi-ethnic individuals may be more diverse than treated in our analyses, with respect to the background and gender of the foreign-born parent among other things. Due to the small numbers, we have summarised all multi-ethnic individuals with one Swedish parent into one category. Future research should account more for the diversity within this group and its potential impact on partner choices.

Despite these limitations, the findings of our study show that reducing heterogeneity by treating multi-ethnic individuals as part of the majority group or as the children of immigrants and consequently defining intermarriage as marriage with mono-ethnic natives, as in previous studies, does not adequately reflect the distinctiveness of the individuals’ backgrounds and their partner choices. Therefore, in line with some of the findings from other contributions to this special issue (see, for example Alba and Reitz 2021), we conclude that multi-ethnic individuals should be regarded as a separate group with partner preferences that neither follow the norms of the mono-ethnic majority group nor those of mono-ethnic minority groups.

Acknowledgements

The authors would like to thank an anonymous reviewer, the editors of this special issue, the participants of a MIM seminar and the participants of an IMISCOE panel for their valuable comments on previous versions of this article.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 In our study, the term marriage refers to both formal marriages and non-marital cohabitations with common children. Non-marital cohabitation without common children cannot be identified in the data.

2 Like in our paper, one of the parents of the participants in this study was a member of the majority population (most of them were White British while some were White European).

3 To assess the association between education and intermarriage separately for mono-ethnic and multi-ethnic individuals, we tested a model including an interaction term between education and being mono-/multi-ethnic (available from the authors upon request). The model did not show any substantial difference between the two groups.