Johanne Sondergaard

Despite the harmonizing efforts of the European Union [EU] member states, family reunification policies remain diverse across Europe. This study examines whether the changes in family migration policies stem from persistently divergent public opinion about gender roles and/or immigration/immigrants. Using data from the European Social Survey (2002-2012), the European Values Study (1990-2008) and the Migrant Integration Policy Index database (2007 and 2010), this study examines whether changes in policies in 25 European countries are influenced by these two types of public opinion previously suggested as influencing family migration policies. The study also looks at whether public opinion is influenced by changes in these policies, using a (cross-lagged) panel model. The results do not give any indication that policies influence public opinion, nor that differences in family migration policies across Europe are influenced by changes in public opinion about immigration or immigrants. In contrast, public opinion about roles in the family–namely support for shared-caring–is found to significantly influence family migration policies. More specifically, public support for more gender egalitarian roles in sharing care in the home are found to lead to more restrictive family migration policies. This finding is in line with the arguments in previous studies, that as gender egalitarianism increases, traditional gender role norms of dependency are projected on the migrant other, manifesting in the form of restrictive family migration policies, for example strict income requirements for the sponsor.

Introduction
The extension of European Union [EU] competencies into the field of family reunification for third-country nationals was first suggested in the Conclusions of the European Council in Tampere in 1999 (Kraler, 2010). The aim of the European institutions was to model the family reunification rights for third-country nationals (i.e. non-EU citizens) after the liberal rights granted to mobile EU citizens (i.e. second-country nationals), consolidated in the Free Movement Directive 2004/38 (Kraler, 2010). But throughout the negotiations of the Family Reunification Directive 2003/86 (Council of the European Union, 2003), some member states argued for the possibility of states to institute stricter conditions for third-country nationals than for mobile EU citizens. This opposition meant that when the Family Reunification Directive came into effect in 2005, it was a merely an ‘instrument of minimum harmonization’ (Boeles et al., 2009: 182), including many optional clauses, leaving much discretion to the member states about the family reunification rights granted to third-country nationals (Block & Bonjour, 2013; Boeles et al., 2009). For example, Article 4 of Directive 2003/86 states that only a sponsor’s spouse and minor children are eligible for family reunification; for other family members such as parents, adult children and unmarried partners, member states are free to set conditions. Additionally, Article 4(5) of the Directive states that member states may set an age limit for sponsors and migrant spouses up to 21 and in Article 7(1)(c) that member states may require a stable income.

The lack of a strict EU Directive means that family reunification policies across member states remain diverse, e.g. in levels of income requirement. Previous authors even suggest that family migration policies are becoming increasingly diverse across the EU (Koopmans et al., 2012; Søndergaard, 2014a). Koopmans et al. (2012) in their quantitative study of ten Western- European countries between 1980 and 2008, show that despite the harmonizing influences of the EU such as the Family Reunification Directive, marriage migration policies went from being very similar in 1980 to diverging more at every time point until 2008 (when the study ended). Additionally, family migration policies have been shown to be diverging between 2007 and 2010 (Søndergaard, 2014a).

The reasons given for this lack of harmonization of family migration policies often lie in traditional explanations of migration policymaking, namely ‘in terms of a rational balancing of economic interests, electoral pushes, and judicial constraints’ (Bonjour & De Hart, 2013: 61). But researchers have pointed out that these traditional theories often cannot explain final policymaking decisions and therefore turn to alternative explanations. One alternative explanation for family migration policymaking is the influence of public opinion. The differences across the EU in family migration policies would thus be explained by differences in public opinion across EU countries. Indeed, some authors suggest that a lack of EU harmonization of immigration policies may be rooted in divergent attitudes about immigration (Luedtke, 2005). This hypothesis is supported by previous studies showing that attitudes toward immigration are diverging in the same period as family migration policies have been found to be diverging (Søndergaard, 2014a, 2014b).

Other authors suggest looking not just at opinions about immigration, but also at whether other opinions influence immigration policymaking. For family migration policies, Bonjour and De Hart (2013: 62) suggest that ‘[d]ebates about family migration policies are shaped in fundamental ways by conceptions of what the roles of men and women ought to be, what marriage ought to be, what parenting ought to be, and what family ought to be… Such gender and family norms play a crucial role in the production of collective identities, i.e. in defining who “we” are and what distinguishes “us” from “the others.”’ This argument is presented especially about family migration policies, because this type of migration poses a threat to integration and national identities (Block, 2014; Bonjour & Kraler, 2014). The relationship between gender norms and family migration policies is supported by previous findings that show a divergence of support for sharing care roles in the home (Søndergaard & Ganzeboom, 2013).

This study builds on the findings discussed above in looking at whether these different public opinions directly influence family migration policies across Europe. This study asks: Can divergent public opinion about immigration/immigrants and/or gender roles explain changes in family migration policies across European countries?

Theoretical framework
Attitudes are defined here in line with other authors, as individuals’ preferences in specific situations, e.g. whether an individual thinks that women should work (Lück, 2005). As attitudes are analyzed here at the average country-level, they are generally referred to as public opinion. Policies are defined in two ways, in line with the seminal work by Hammar (1985), on the difference between immigration and immigrant polices. Immigration policies are defined as ‘the rules and procedures governing the selection and admission of foreign citizens’ (Hammar, 1985: 52), while an immigrant policy ‘refers to the conditions provided to resident immigrants…’ (Hammar, 1985: 53). Family migration policies include both of these policy areas, as they refer to the rights of the already present immigrant (sponsors) by regulating the entry (immigration) of their family members (Bonjour & Kraler, 2014). Simply stated, immigration policies are directed at people who are not yet ‘here’, while immigrant policies are directed at people who are already ‘here’. This paper looks at whether an opinion-policy nexus and/or a policy-opinion nexus exist for two types of attitudes suggested to be related to family migration policymaking, namely gender-egalitarian attitudes and/or immigration/immigrant attitudes. The focus of the study is mainly the opinion-policy nexus and this relationship is therefore elaborated more thoroughly.

Opinion-policy nexus
In a bottom-up perspective on policymaking, social attitudes inform voting, with a majority opinion being reflected in majority voting. This majority voting in turn indirectly influences policies (Raven et al., 2011; Risse-Kappen, 1991). This is referred to by Raven et al. (2011) as the opinion-policy nexus. As expressed by Jacobs and Herman (2009: 114), ‘[o]bviously, there is by definition some link between public opinion and policymaking in democracies. Politicians and political parties cannot systematically act against public opinion and hope to get re-elected.’ Indeed, classical studies such as the work by Page and Shapiro (1983) describe how public opinion is a major influence on policy changes in the US. When opinions change, so too do policies after a 1-4 year time lag (Page & Shapiro, 1983).

Some authors have previously made the case that immigration policies are a special type of policy not influenced by public opinion because decision-making here remains within the domain of elites (Freeman, 1995). This has since been disputed, with authors claiming that immigration has become such a highly salient issue in the public and political debates and that decision-making is no longer taking place behind closed doors. According to Lahav (2004: 1158), the public sets the ‘rules of the game’ around which elites structure their discourse on immigration. In this vein, the work of Benhabib (1996) suggests a relationship between majority voting and immigration policies. According to the opinion-policy nexus, immigration policies would reflect public opinion if immigration policymakers work in polities characterized by democratic accountability. An example of this increased public scrutiny of immigration policies at EU level is the Amsterdam Treaty of 1997, moving immigration out from behind the closed doors of the intergovernmental decision-making sphere (Guiraudon, 2001; Kostakopoulou, 2000).

Gender role attitudes – family migration policy nexus
It could be expected that the opinions influencing immigration policies would be those related to immigration and immigrants, but an increasing number of authors stress the influence of opinions about cultural norms. This is likely related to the findings that attitudes toward immigrants are rooted more in concerns about differences in culture than in economic concerns. For example, in their experimental study, Sniderman, Hagendoorn and Prior (2004: 43) find that opposition to immigration stems more from immigrants not fitting in culturally rather than not integrating economically. Hainmueller and Hopkins (2014:235) in their review of studies about immigration/immigrant attitudes, state that ‘[s]ome conceptions of the national community and its boundaries can easily accommodate newcomers whereas others cannot.’ Aspects of national community and group boundaries that have been suggested recently to be key in family migration policies are gender role norms and marriage norms (Bonjour, 2011; Bonjour & De Hart, 2013; Van Walsum, 2008).

Gender role attitudes refer to attitudes about what roles men and women should adopt within the family. These roles refer to how the earning of the family income should be arranged (i.e. single, shared-earning or 1.5 model) and how childcare should be arranged (i.e. one parent or shared between partners, with other family members, and/or with state/market institutions). These views have changed greatly since the 1970s, when women entered the workforce en masse, creating a vacuum of childcare in European homes (Pfau-Effinger & Rostgaard, 2011). Van Walsum (2008) is one author who examined these changes in family norms in the Netherlands, but she made a novel comparison, namely linking changes in family norms to changes in family migration policies.

Van Walsum (2008) traced Dutch family migration policies from 1945 to 2000 and suggested that family norms were used to distinguish the ‘national’ from the ‘foreign’. Interestingly, she observed that while family norms became more egalitarian, these egalitarian family norms were not transferred to family migration policies. In fact, an increasingly non-gender-egalitarian view of the family was projected on migrants in the Netherlands. She discussed, for example, the appearance of the gendered notion of ‘dependency’ in family migration policies, referring to the income and housing requirements for sponsors. These requirements necessitate the sponsor to provide for the incoming family member, which is very much in line with the traditional view of the dependency of one (female) spouse on the other (male). Van Walsum (2008:239) points out that the aim of these policies was to prevent the welfare state supporting entire immigrant families, but that another way to prevent migrants from relying on welfare would be to allow for the earnings of the incoming family member to count towards the income requirement. This would mean that neither partner would be expected to provide for the other, but that both can contribute to the family earnings. Such an alternative policy approach would portray very different family norms, namely shared-earning (shared between partners) rather than the breadwinner norm implied by a single income requirement.

Bonjour and De Hart (2013) argue that family norms have played an important role in policymaking on fraudulent and forced marriages since the 1970s in the Netherlands. They trace how the view of a ‘proper’ Dutch family was a way of ‘othering’ migrant families and marriages. Also, Bonjour (2011) in her analysis of pre-departure language tests in the Netherlands (Civic Integration Examination Abroad or: het basisexamen inburgering in het buitenland) adopted in 2005 argues that migrant women are portrayed as ‘weak’ and ‘vulnerable’ dependents. She shows that this policy was framed specifically as aiding the emancipation of migrant women by enabling migrant women to speak Dutch upon arrival, thus enabling them to free themselves from dependency on their supposedly oppressive Muslim husbands. Similarly, Roggeband (2007) argues that a link is made in the Netherlands between women’s emancipation and family migration. Elsewhere, Eggebø (2010) shows how gender norms are specifically referred to in the Norwegian political debates on family migration, and Borevi (2014) shows how the legacy of Sweden’s gender norms in welfare policies are prevalent in the Swedish family migration debate.

The reasoning that the authors give for this relationship between family norms and family migration law is that the ‘family’ is an important way for the native population to distinguish themselves from the migrant ‘other’. As mentioned above, family migration is especially construed as ‘a problem of culture, identity, and belonging’ (Bonjour & Kraler, 2014: 4), with the national identity being ‘construed in opposition to the perceived culture and identity of migrants, epitomized by the “migrant”—especially “Muslim”— family. Whereas the “Western” family is imagined as modern, emancipated, and egalitarian, the “migrant” family is associated with tradition, patriarchy, oppression, and even violence’ (Bonjour & Kraler, 2014: 4). Family migration is thus part of defining belongingness to a polity (Block, 2014). As egalitarian gender role attitudes develop within a country, they are used as a marker between insiders and outsiders. In line with this argument, gender norms are used specifically because gender role norms are part of the foundation of culture (Bonjour & De Hart, 2013), with women at the center of ethnic and national reproduction (Anthias & Yuval-Davis, 1992). As more egalitarian norms develop, they are used to distinguish between ‘us’ and ‘them’ with one manifestation being within family migration policies.

The works of Van Walsum, Bonjour and De Hart all look at changes in family norms within one country over time. Another way to look at the influence of gender norms on family migration policies would be to see whether changes in prevalent norms in different countries over time are reflected in different countries’ family migration policies across time. Such an analysis would see whether countries with more egalitarian gender norms such as Denmark have restrictive family migration policies, whereas those with less egalitarian norms such as Italy would have more permissive family migration policies. It could also look at whether as norms become more egalitarian, policies become more restrictive. Such a country comparison can be done using large cross-national surveys and quantitative policy measures. The link between family norms and family migration may indeed be a plausible hypothesis considering the previous findings that family migration policies are diverging (Søndergaard, 2014a) and so too are ideas about sharing care in the home across Europe (Søndergaard & Ganzeboom, 2013). But the possible influence of migration attitudes should also be considered, as these opinions have also been suggested to be diverging (Søndergaard, 2014b).

Immigration/immigrant opinion – immigration/immigrant policy nexus
The above literature on the link between gender norms and family migration policies stands apart from the literature exploring the relationship between immigration/ immigrant policies and another type of public opinion, namely opinions about immigration/immigrants. Beutin et al. (2007: 390) provide the following explanation of this proposed relationship: ‘suppose that the public perceives migration predominately as a phenomenon associated with dead bodies in the Mediterranean, human trafficking, and unemployment. Calls for tighter border controls are often the consequence.’

In her review of eighteen studies looking at the relationship between integration policies and public opinion, Callens (2015: 16) states that a ‘consistent and positive relationship emerged in several studies between countries with more inclusive integration policies (i.e. higher MIPEX overall scores) and lower levels of perceived threat and, to some extent, lower levels of negative attitudes towards immigrants’. But authors such as Simon and Lynch (1999) do not find a direct relationship between the attitudes toward immigration and immigrants and countries’ immigration policies. Similar to others, they claim that there is no influence of general public opinion, but rather claim that lobbying or pressure groups influence immigration policies (Facchini & Mayda, 2008; Freeman, 1995). Several measurements of immigration/immigrant attitudes are included in this study to further explore the disputed link between these attitudes and policies.

Policy-opinion nexus
The opinion-policy nexus discussed above, where opinions influence policies has been suggested previously only to exist for newer social policies (Raven et al., 2011). Only for policies that are not yet institutionally well-established, are politicians open to public opinion. Well-established social policies are not suggested to be open to change from public opinion because policies are locked in ‘path-dependency’ (Pierson, 2001). For well-established welfare policies, previous studies rather have found a policy-opinion nexus, in other words that policies influence opinions (Raven et al., 2011).

That policy influences opinion perspective is also illustrated by normative theories of law. As expressed by Schlueter et al. (2013: 672), ‘majority group members adapt their pre-existing attitudes in response to legislative measures, presumably because they recognize that deviations from a social norm produce negative sanctions’. According to normative theories of law, law can influence conduct and beliefs not just through sanctions, but also through conveying a consensus about a topic (Albiston et al., 2011). If a legal system is legitimate, then a law will be perceived as expressing a consensus. This consensus may be an actual consensus of public opinion or it can be driven by a small elite. In line with social psychological hypotheses, a majority opinion conveyed by laws will influence individual beliefs because people’s attitudes change toward the perceived consensus to avoid cognitive dissonance (Albiston et al., 2011; Schmidt, 2008).

The perceived consensus conveyed by open immigration/immigrant policies can lead to two different types of reactions in public opinion, according to theories about group conflict and intergroup norms (cf. Callens, 2015; cf. Schlueter et al., 2013). Research on group conflict suggests that permissive integration policies promote group conflict, as majority group members will perceive an extension of rights to a minority group as a threat to the majority’s resources. These resources can be economic, but they can also be cultural, with the majority group seeing a threat to their cultural dominance. The literature on intergroup norms suggests an opposite relationship between public opinion and policies. This literature suggests that policies promote social norms for adequate intergroup relationships, meaning that as integration policies become more open, intergroup relationships will also become more open. In their comprehensive study, Schlueter, Meuleman and Davidov (2013) find a negative association between the permissiveness of a country’s integration policies and citizens’ perceived group threat, supporting this