In the present article, we aim at contributing to a social psychological body of research on identity processes in Europe (see Abrams & Vasiljevic, 2014), some of which have used public opinion Eurobarometer data (e.g. Bruter, 2003; Duchesne & Frognier, 1995; Kohli, 2000). In particular, we examine ways to boost Europeans’ willingness to engage in cross-border, EU-wide behaviours (e.g. living and working in other EU countries) as an individual-level proxy of European integration. We focused on identity complexity (Linville, 1982; Brewer & Pierce, 2005) and identity inclusiveness (Dovidio, Gaertner, & Saguy, 2009) as predictors of these trans-national behavioural intentions because we expected these personality dimensions to be the ‘engine room’ of a dynamic process of deprovincialization of the self (Pettigrew, 1997) within contemporary Europe. This assumption aligns with King and Ruiz-Gelices’ (2003) findings, indicating that having a European identity predisposes British students to prospective transnational behaviours.

In the following, we discuss key concepts in a mediation model at the centre of this investigation. We start with an analysis of the predictors (identity complexity and inclusiveness), and then proceed to the outcome (interest in EU-wide behaviours). Finally, we discuss proposed mediators (intergroup concerns and general life optimism).

Nurturing Complex and Inclusive Identities to Encourage Deprovincialization via EU-wide Behavioural Intentions

The way individuals represent groups to which they and others belong exerts considerable influence on the way individuals think, feel and behave (see Crisp & Hewstone, 2006). These influences should extend to the way individuals respond to immigrants and intergroup dynamics in their social reality, to how they feel about their life and future, and to their readiness to engage with their broader social context.

Examining identification processes across multiple group memberships, Roccas and Brewer (2002) first introduced the concept of social identity complexity to encompass considerations of both identity complexity and identity inclusiveness. These scholars conceptualized social identity complexity as a bipolar dimension, ranging from simple to complex. On the complex end of the spectrum, individuals would recognize that their ingroups differ in important ways, making those social representations simultaneously more complex and more inclusive. In a similar vein, bicultural identities (i.e. seeing oneself as French and European) have often been conceived as the complex and nonlinear co-existence of distinguishable identities and cultural frames (see Gaertner & Dovidio, 2014).

Within the research tradition of the dual identity and common ingroup identity models (see Dovidio, Gaertner, & Saguy, 2009 for a review), Gaertner et al. (1996) pointed out that the psychological process of identity integration does not require the representation of smaller, less inclusive identities to be dismissed. Hence, European residents may hold a dual identity structure if they develop an identity representation of being European, as well as maintain their national identity (see Dovidio, Gaertner, & Saguy, 2009).

Previous research has already shown that identity complexity and inclusiveness predict acceptance of multiculturalism and positive host communities-immigrants interactions (Brewer & Pierce, 2005; Schmid, Hewstone, & Tausch, 2014). Brewer and Pierce (2005), for instance, found that majority members’ identity complexity was associated with more positive attitudes towards multiculturalism and ethnic diversity. In a series of experiments, inclusive group categorizations were found to explain the link between intergroup cooperation and reduced intergroup bias (Gaertner, Mann, Dovidio, Murrell, & Pomare, 1990). More recently, Schmid et al. (2014) showed that social identity complexity was implicated in explaining covariations between intergroup contact and positive attitudes towards outgroups uninvolved in the contact experience. From the immigrant’s perspective, identity complexity predicted higher acceptance of national liberal practices among Turkish immigrants living in the Netherlands (Verkuyten & Martinovic, 2012).

Drawing from this evidence, we predict that identity complexity and inclusiveness should influence the way individuals perceive and evaluate themselves and others as part of a national or greater (e.g. European) whole. With this research, however, we wanted to take our investigation beyond predicting overall attitudinal responses to transnational identities (i.e. Europe or Europeans). Our main aim was to ascertain whether individual differences in identity complexity and inclusiveness are implicated in a more dynamic identity process, with potentially more far-reaching consequences for host-immigrant relations and broader European integration. Hence, we turned our attention to the process of deprovincialization of the self.

Pettigrew (1997) introduced the notion of deprovincialization of the self within the intergroup contact literature (Dovidio, Gaertner, & Kawakami, 2003). He argued that positive interactions between members of distinct groups, like host country citizens and immigrants, have the potential to produce a broad reconfiguration of attitudes and behaviours which reflects a progressive re-positioning of the self away from narrow and partisan allegiances to the ingroup (e.g. national or majority identities and perspectives) and closer to the outgroup (e.g. immigrants) and more inclusive views and engagement with the world (e.g. European identity and perspectives). From this dynamic view of intergroup relations, psychological self-deprovincialization can be seen both as an outcome of positive intergroup relationships, as well as positive propellers of future integration efforts.

In an effort to gauge situational deprovincialization, we chose self-deprovincialization as our focal outcome variable. We measured European residents’ interest in EU-wide behaviours, like living and working abroad, to assess their context-dependent willingness to psychologically move towards more inclusive, transnational and deprovincialising behaviours. Hence, rather than regarding deprovincialization as a relatively stable attitudinal response, we treated it as a temporary and malleable end-state. To ensure we isolated situated (vs. chronic) interests and behavioural readiness, we controlled for attitudinal indicators and habitual experiences with EU-wide behaviours in our main analyses. We expected interest in EU-wide behaviours to function as a key psychological springboard to a broad range of attitudes and behaviours that, over time, support wider and inclusive EU-level group memberships.

Identity complexity and identity inclusiveness were operationalised in this work as two distinct independent variables in our models. However, we expected them to be related and exert similar influences on deprovincialising behavioural intentions. We also expected the two identity predictors to share some psychological underpinnings. In this study, we focused on intergroup concerns and general life optimism. Next, we discuss the basis for our mediational hypotheses.

Testing the Mediational Role of Intergroup Threats and General Optimism

In our investigation, we assessed whether the complexity and inclusiveness of identities boost individuals’ willingness to engage in deprovincialising behaviours by increasing individuals’ general optimism about life in Europe, and by mitigating the salience of immigration and intergroup threats.

There is evidence to suggest that positive affect facilitates more inclusive self-schemata, which in turn reduces intergroup bias (Crisp & Hewstone, 2000; see also Dovidio, Gaertner, Isen, Rust, & Guerra, 1998; Isen, Niedenthal, & Cantor, 1992). Additionally, self-complexity has been shown to coincide with reduced levels of psychological distress and frustration (Gramzow, Sedikes, Panter, & Insko, 2000; Kalthoff & Neimeyer, 1993). Based on these studies demonstrating positive relations between identity inclusiveness, self-complexity, and positive affect, we hypothesise that more inclusive and complex identities are also associated with a positive outlook in life in the European context.

Research shows that optimistic people engage in more prosocial behaviour – most of which is typically intergroup in nature (Stürmer & Snyder, 2010) – such as donating money to charity or volunteering, as a meta-analysis shows (Lyubomirsky, King, & Diener, 2005). Optimists should therefore be more prone to engage in behaviours that transcend their specific ingroup and consider broader, trans-national perspectives and opportunities as desirable for themselves and their families.

Abundant social psychological research demonstrates, on the other hand, that intergroup anxiety, intergroup concerns, and threats surrounding relationships with those we perceive as others contribute to intergroup bias, anti-immigrant attitudes, and dis-endorsement of multi-culturalism (see Paolini, Harris, & Griffin, 2016; Stephan, 2014 for reviews). Previous research indicated that anti-immigrant attitudes further predict opposition to EU-wide behaviours; research conducted found that anti-immigrant attitudes foster opposition to the EU and to European integration (Boomgaarden, Schuck, Elenbaas, & de Vreese, 2011; de Vreese & Boomgarden, 2005). From this body of evidence, we expected intergroup threats to be detrimental and compromise European respondents’ interest in EU-wide behaviours.

We considered several kinds of intergroup concerns, namely threats posed by immigration, terrorism, and crime. Our use of a composite intergroup concerns variable implies that we expect our respondents to treat these different types of intergroup concerns as psychologically equivalent. We assume that threat appraisals that are constructed around outgroups commonly discussed in the public discourse as posing a threat would negatively impact dynamics of planned deprovincialising behaviours.

Summary of Key Hypotheses and the Eurobarometer Survey

Based on the aforementioned conceptualisations, we hypothesised that the integration of both a national and a European identity in European respondents’ self-descriptors to be more complex than the maintenance of either a national or a European identity alone (for a similar approach, see Gaertner & Dovidio, 2014). We further assumed that identities become more inclusive as they move from a subordinate, national framing further towards a purely superordinate representation (i.e. a European identity).

In the current article, we linked complex and inclusive identities with positive and negative affective appraisals in a novel way. We expected identity complexity and inclusiveness to encourage European respondents’ interest in trans-national, EU-wide behaviours – at least partly – through the lessening of intergroup threats and through the boosting of general life optimism. Therefore, we tested whether decreased intergroup concerns and increased life optimism acted as pathways (i.e. mediators) to the relation between more complex and inclusive identities (i.e. independent variables) and more deprovincialising behavioural intentions (i.e. dependent variable). To our knowledge, this process model in its entirety has not been tested previously.

We tested our hypothesis on cross-sectional correlational data. As such, we were unable to provide a firm basis for a temporal sequence of events. While it is likely that the relationships between variables reflect interrelated bidirectional or circular processes, we selected the order of variables based on a structural logic moving from stable (predictors) to more context-dependent (mediators) to situational (outcome) variables. Specifically, we organised identity inclusiveness and complexity as independent variables because we expected these variables to be individual differences and thus function as psychological ‘antecedents’. With this, we theorised the identity variables to influence more context-dependent processes (i.e. appraisals of threat and a positive future), and in turn shape situation-specific behavioural planning (i.e. interest in EU-wide behaviour as a behavioural intention). These research hypotheses were put to empirical test via secondary data analysis of Eurobarometer data.

The Eurobarometer is a collection of public opinion surveys regularly carried out for the European Commission; various European consortia draw representative samples through EU-wide interviews with adults residing in the European Union. These include citizens of the respective countries and immigrants holding citizenship of other countries. The number of interviews conducted per country is typically quite large (about 1,000 interviews per country), allowing for cross-country comparisons with sufficient statistical power to detect even small effects. The 2014 Eurobarometer survey data was collected between May and June 2014 (European Commission, 2014). For this investigation, we focused on the 28 EU countries (see list on Table 1).1

Table 1

Number of Interviews in the 2014 Eurobarometer Survey as a Function of EU Country.

No Country n No Country n

1 Belgium 1,012 15 Finland 1,012
2 Denmark 1,006 16 Cyprus 500
3 Germany 1,576 17 Czech Republic 1,011
4 France 1,004 18 Estonia 1,031
5 Ireland 1,003 19 Hungary 1,087
6 Italy 1,036 20 Latvia 1,012
7 Luxembourg 506 21 Lithuania 1,015
8 Netherlands 1,016 22 Malta 504
9 UK 1,373 23 Poland 1,001
10 Greece 1,017 24 Slovakia 1,024
11 Spain 1,033 25 Slovenia 1,043
12 Portugal 1,075 26 Bulgaria 1,074
13 Austria 1,000 27 Romania 1,022
14 Sweden 1,008 28 Croatia 1,003

Note. n = sample size per country. Total N = 28,004. Eurobarometer interviews took place in all 28 EU member states.

Past EU-related behaviours and attitudes towards the EU were treated as controls to ensure that our dependent variable captured situated behavioural intentions towards transnational activities, rather than chronic or habitual EU behaviours or mere EU attitudes. We further selected perceived quality of life in the EU as a control for our mediator (life optimism) to ensure it reflected a broader (non-EU specific) positive outlook to life. Thus, this approach increases the conceptual precision of our constructs of interest and, in so doing, adds to the body of extant research in this area that has used correlational data to examine relationships between variables that sit on the same (i.e. chronic) levels of analysis.

Method

Respondents

Twenty-eight thousand European adults (N = 28,004; females 54.3%; age M = 50.05, SD = 18.34) residing in one of the 28 EU member states completed the 2014 Eurobarometer survey (Standard EB 81, 2014) and contributed to our secondary data analyses. Respondents were predominately EU citizens (n = 26,831, 96% of total N) with a smaller number of immigrant residents not holding citizenship of their host country (n = 695, 2.5% of total N)

Multi-nationals holding citizenship of their residing country and other citizenships were also included (n = 478, 1.7% of total N).

Measures

Due to the nature of the Eurobarometer questionnaire items, all our indices were created using responses to single interview questions.

Identity complexity and inclusiveness as predictors. The independent variables’ indices were both derived from the same item asking participants about their national and European identity (“Do you see yourself as…?”) and requiring respondents to choose among five options (0 = none, 1 = national identity, 2 = national first then EU, 3 = EU first then national, 4 = EU identity); the actual response options provided to respondents replaced ‘national’ with the specific country in which the interview was conducted. We computed an index of identity complexity counting the number of identities in the chosen response option; this index could thus vary between 0 and 2 (0 = none, 1 = national or European, 2 = national first, then European or European first, then national). Further, an index of identity inclusiveness was computed, with higher scores indicating purer superordinate or inclusive (EU) identities (0 = none, 1 = national only, 2 = national first then EU, 3 = EU first then national, 4 = European only). Because the two independent variables were highly correlated (r = .77, see Table 3), we controlled for the effect of the other IV when testing for mediation effects. By doing so, we were able to isolate the unique effects of each independent variable on the dependent variable and mediators, over and above the IVs’ shared variance.

Intergroup concerns and general optimism as mediators. To identify whether participants felt threatened by intergroup-relevant issues, we inspected their responses to the item: “What do you think are the two most important issues facing our country at the moment?” Response options were crime, economic situation, rising prices/inflation, taxation, unemployment, terrorism, housing, financial situations of your household, immigration, health and social security, education system, environment, climate and energy issues, and pensions. Crime, terrorism, and immigration were identified as intergroup in nature due to their centrality in discussions about intergroup dynamics by experts and the media (e.g. Cohrs & Stelzl, 2010; Dancygier, 2010; Das, Bushman, Bezemer, Kerkhof, & Vermeulen 2009; Hewstone, Rubin, & Willis, 2002). The intergroup concern index was calculated by counting the number of intergroup issues selected as the “two most important issues” and therefore could vary between 0 and a maximum of 2.

Respondents also indicated their general optimism: “What are your expectations for the next twelve months when it comes to your life in general?” Respondents indicated their prospects by selecting one of three response options (1 = worse, 2 = same, 3 = better).

Interest in EU-wide behaviours as outcome. Deprovincialising orientations were indexed with a measure of interest in EU-wide behaviours. The interview question: “About which rights as a citizen of the EU would you like to know more?” allowed respondents to express their interest (if any) in a maximum of two EU citizens’ rights, such as buying goods or receiving medical assistance in another EU country. The eight options available were weighted based on the degree of active engagement with trans-national entities required by the behaviour; by doing so, we created an index by which greater values signify greater active engagement with people from other EU countries. For example, buying services received a low weight of 1, as it refers to a relatively formalized process that does not require high levels of effort or engagement, whereas living abroad received the highest weight of 8, as it requires intensive and lasting personal involvement with EU citizens and their customs. The eight options were scored as follows, 0 = none chosen, 1 = buying services in another EU country, 2 = buying goods in another EU country, 3 = benefitting from a consular protection of any EU country, 4 = receiving medical assistance in another EU country; other, 5 = voting in elections in another EU country, 6 = studying in another EU country, 7 = working in another EU country, 8 = living in another EU country. The individual score was created averaging the weight of the two chosen options; as a result, this index could vary between 0 and 7.5.

Past EU behaviours, EU attitudes, and quality of life as covariates. We considered several other EU-relevant variables that could influence the deprovincialising EU-broad behavioural intentions.

Past EU behaviour was measured asking participants how much they had engaged in EU-broad behaviours (“For each of the following achievements of the EU, could you tell me whether you have benefited from it or not?”; e.g. no/less border controls when travelling abroad, living in another EU country; benefitted/not benefitted). Like for the interest in EU-wide behaviours index, we weighted the nine response options for degree of personal engagement and averaged participants’ responses across the nine response options, resulting in values ranging between one and nine (minor/major engagement in past EU behaviours). Attitudes towards the EU were measured by an item asking respondents to indicate whether they had a positive or a negative image of the EU (“In general, does the EU conjure up for you a positive image?”; 1 = very negative, 5 = very positive). As one of the proposed mediators gauged appraisals of a positive future in the EU, we considered also participants’ satisfaction with their EU residency (“How would you judge the current situation in each of the following? Quality of life in EU?”; 0 = very bad, 4 = very good).

Results

Missing Data

As responses to the survey items were not mandatory, there was a substantial amount of missing data, especially on the responses to the outcome variable (>15,500) and the control variable past EU behaviours (>14,500). The other variables had far fewer missing values (between 500 & 5,300). A Little’s (1988) Missing Completely at Random test yielded a significant result (χ2 = 5033.102, df = 146, p = .000), indicating that the data was not missing completely at random. To account for the missing data, we employed the multiple imputation technique: Multiple imputation produces several imputed data sets (we chose 5), which account for the uncertainty of the missing data (for a review on missing data handling, see Kang, 2013). This technique can further handle large data sets, large amounts of missing data, and is robust against violations of the normality assumption (Kang, 2013).

Descriptive Statistics

Descriptive statistics for the main indices are shown in Table 2, including possible range. Means for identity complexity were high at 1.55 (SD = .52; 0 vs. 2), indicating that, on average, respondents reported complex identities with between one or two identities selected (national/European). The mean value for identity inclusiveness (M = 1.65, SD = .67, 0 vs. 4) indicated that respondents typically regarded their nation as their primary identity before considering their EU identity.

Table 2

Descriptive Statistics of Eurobarometer Indices.

Index M (SD) Range of values Eurobarometer Assessment

Identity complexity 1.55 (.52) 0–2 0 = No national or EU identity selected, 1 = one (national or EU), 2 = two (national + EU) identities selected
Identity inclusiveness 1.65 (.67) 0–4 0 = No national or EU identity selected, 1 = national only, 2 = national followed by European, 3 = European followed by national, 4 = European identity only
Intergroup concerns .27 (.49) 0–2 Immigration, terrorism, or crime chosen as two most important issues facing the EU
General optimism 2.15 (.60) 1–3 General life expectations for the next 12 months (worse; same; better)
Interest in EU-wide behaviours 4.28 (2.05) 0–7.5 Interests weighted from most to least accessible (e.g. buying goods in another EU country vs. living in another EU country)
Past EU behaviours 4.33 (1.36) 1–9 Benefits experienced from EU membership weighted from small to major (e.g. cheaper calls vs. living in another EU country)
EU attitudes 3.11 (.94) 1–5 Image of EU ranging from very negative to very positive
Quality of life in the EU 2.33 (1.13) 0–4 Quality of Life in EU ranging from very bad to very good

Note. M = Mean; SD = Standard Deviation.

Intergroup concerns were rarely chosen as one or both of the two most pressing national issues by the respondents (M = .27, SD = .49, 0 vs. 2). Optimism, as measured by outlook in life, was moderate (M = 2.15, SD = .60, 1 vs. 3), indicating that respondents generally expected their future to be comparable or slightly better from their current state.

The mean for interest in EU-wide behaviours (M = 4.28, SD = 2.05, 0 vs. 7.5) suggested that there was generally a moderately high level of interest in EU-wide behaviours. Participants reported past EU benefits (see Past EU behaviours; M = 4.33, SD = 1.36, 1 vs. 9), relatively positive attitudes towards the EU (M = 3.11, SD = .94, 1 vs. 5) and moderate quality of life in the EU (M = 2.33, SD = 1.13, 0 vs. 4).

Correlations

Table 3 reports zero-order Pearson correlations between all variables. In light of the large sample size, we focused on the magnitude of the relations rather than their significance. There was a robust (albeit not perfect) correlation between identity complexity and identity inclusiveness (r = .77), which partly reflects common method variance. Identity complexity and inclusiveness had small positive associations with interest in EU-wide behaviours (rs = .08), indicating that our individual differences predictors were coinciding to a small degree with our situated, behavioural intention outcome.

Table 3

Pearson Correlations between Variables.

Variables 1 2 3 4 5 6 7

1. Identity complexity
2. Identity inclusiveness .77 **
3. Intergroup concerns –.07 ** –.06 **
4. General optimism .12 ** .11 ** .02
5. Interest in EU-wide behaviours .08 ** .08 ** –.05 ** .08 **
6. Past EU behaviours –.04 ** –.01 –.02 –.06 ** –.00
7. EU attitudes .30 ** .28 ** –.07 ** .18 ** .08 ** –.04 **
8. Quality of life in the EU .20 ** .18 ** –.05 ** .12 ** .11 ** –.01 .25 **

Note. Two-tailed correlations. * p < .05, ** p < .01, imputed N = 28,004; Correlations equal to or greater than .10 are highlighted in bold.

Identity complexity and identity inclusiveness further showed weak positive correlations with general optimism (rs = .12 & .11) and weak to moderate positive correlations with the EU attitudes (rs = .30 & .28) and quality of life in the EU (rs = .20 & .18). In contrast, identity complexity and identity inclusiveness were both negatively correlated with intergroup concerns, albeit weakly (rs = –.07 & –.06). Hence, more complex and inclusive identities were associated with more positive views of the EU, with life in general and in the EU, while reducing the likelihood of feeling threatened by intergroup issues.

Mediation Analyses

We performed Hayes’ (2016) bootstrapped mediation tests (Model 4, with mediators operating in parallel) using the SPSS PROCESS macro (v2.16.3). We tested the effects associated with each of the two focal identity indices in turn, while controlling for the other IV, so to isolate the unique effects of each identity index. We managed differences in metrics (see Table 2) by z-transforming all variables.

Explaining the identity complexity-interest in EU behaviours link. First, we examined whether general optimism and intergroup concerns mediated the relation between identity complexity and interest in EU-wide behaviours. A graphical representation of this model and key results is in Figure 1. The total effect of identity complexity on interest in EU-wide behaviours was significant, β = .073, SE = .009, t = 7.74, p < .001, 95% CI [.054, .091]. The direct effect controlling for identity inclusiveness was also significant, β = .062, SE = .009, t = 6.60, p < .001, 95% CI [.044, .080]. In addition, we found significant indirect (mediational) effects through general optimism (β = .008, SE = .001, 95% CI [.009, .013]) and intergroup concerns (β = .002, SE = .001, 95% CI [.002, .004]). These results indicate that identity complexity predicted interest in EU-wide behaviours through increased general optimism and reduced intergroup concerns.

Figure 1 

Mediation Model of Identity Complexity on Interest in EU-wide Behaviours via General Optimism and Intergroup Concerns, Controlling for Identity Inclusiveness.

Explaining the identity inclusiveness-interest in EU behaviours link. For the independent variable of identity inclusiveness, only the mediation effect of intergroup concerns yielded statistical significance (see Figure 2). In this mediation analysis, the total effect of identity inclusiveness on interest in EU-wide behaviours was not significant, β = .013, SE = .009, t = 1.41, p = .159, 95% CI [–.005, .032]. In addition, the direct effect controlling for identity complexity was also not significant, β = .011, SE = .009, t = 1.17, p = .243, 95% CI [–.008, .029]. Yet, there was a significant indirect (mediational) effect of intergroup concerns (β = .001, SE = .000, 95% CI [.001, .002]); the indirect effect of general optimism was instead not significant (β = .001, SE = .001, 95% CI [–.000, .003]). Hence, in contrast to identity complexity, identity inclusiveness influenced interest in EU-wide behaviours only through the perceived salience of intergroup concerns and these effects were not a mere product of positive attitudes towards the EU or past EU-wide behaviours.

Figure 2 

Mediation Model of Identity Inclusiveness on Interest in EU-wide Behaviours, via General Optimism and Intergroup Concerns, Controlling for Identity Complexity.

Both mediational patterns reported persisted when controlling for the other IV, past EU-related behaviours, attitudes towards the EU, perceived quality of life in the EU, and the sociodemographic variables of age and gender.

Respondents residing in countries that have a longer tradition of being part of the European Union might have more “tried-and-tested” self-concepts and thus may respond differently compared to residents of relatively newer EU member states. To test whether the mediation results hold across earlier and later established EU country groups, we created a dummy variable that distinguished between core EU9 member states (Germany, France, UK, Italy, Denmark, Ireland and the former Benelux states, see countries 1–9 in Table 1, N = 9,532), and the remaining member states (countries 10–28; N = 18,472). We then entered the dummy variable as a control into both mediation models. The mediational patterns remained unchanged, indicating that the basic models hold across older and newer EU country groups.

Discussion

In these secondary analyses of the 2014 Eurobarometer data, we found that identity complexity and identity inclusiveness, while highly correlated, each had unique relations to EU residents’ interest in transnational, EU-wide behaviours. In line with our hypotheses, these relations were significantly mediated by reduced intergroup concerns (including concerns regarding immigration). When identity complexity was entered as the independent variable in our path model, increased general optimism was also a significant mediator. These mediational patterns held when accounting for past EU behaviour, general attitudes towards the EU, and quality of life in the EU, suggesting that they captured mechanisms of situated behavioural intentions, rather than mere behavioural habits or attitudes towards the EU. As such, they may be better placed to predict respondents’ actual behaviours in EU-relevant areas.

Identity Complexity and Inclusiveness as Residents’ Perspectives

Our results indicated that complexity and inclusiveness of the social self were each beneficial in terms of their associations with interest in trans-national behaviours even when controlling for each other’s influence. This means that the two identity variables are similar but not interchangeable. With our focus on the European context, we defined identity complexity as an identity structure potentially moving towards a dual identity representation (i.e. upholding both a national and an EU identity) and we defined identity inclusiveness as an identity structure moving towards a pure superordinate EU identity. While the theoretical and operational overlap between the two is clear, the key difference is that in our way of measuring complex identity (i.e. a dual identity), the national identity is also preserved. Our findings of unique effects by identity complexity and identity inclusiveness therefore support the notion that Roccas and Brewer’s (2002) construct of identity complexity and Gaertner et al.’s (1994, 1996) construct of identity inclusiveness are not synonymous, albeit their conceptualisations show considerable overlap.

It needs to be pointed out that most residents interviewed in the Eurobarometer sample were citizens of the country they were living in (95.8%). Therefore, our findings mainly captured the perspectives of citizens in their home country, rather than the perspective of immigrants to EU countries. When we tested the mediational pathways in a relatively small sample of immigrants (n = 695) taken from the same data collection, they did not yield statistical significance (CIs including 0). The effects were small and the operationalisation of who constituted an immigrant imprecise. It is therefore possible that these tests are simply underpowered and reflect a poorly defined immigration status variable, preventing meaningful mechanisms to present. Keeping these limitations in mind, the indirect effects found in the current majority-driven citizen sample may indeed reflect unique host perspectives.

Overall, the present results suggest that greater identity complexity and inclusiveness function as positive propellers for engaging in cross-border behaviours, including trading with or even living among fellow Europeans who do not share the same nationality as respondents. By being willing to engage in these cross-border behaviours, respondents expressed a readiness to relinquish, at least temporarily, a majoritarian focus that they hold within their respective nation state. Following previous research on the prejudice-alleviating effect of inclusive identities (Brewer & Pierce, 2005; Gaertner, et al., 1990; Schmid et al., 2014), anticipating beneficial exchanges with non-state members may evoke more positive views of other states’ nationals at large and thus affect also how social minorities (e.g. immigrants, refugees) are being perceived by majority members in general (for meta-analytical data, Pettigrew & Tropp, 2006).

Pathways to Deprovincialization

The present article suggests two pathways to greater interest in deprovincialising behaviours. These being through increased general optimism and reduction of intergroup concerns. Reduction of intergroup concerns explained why identity inclusiveness and complexity predicted interest in deprovincialised behaviours. Increased optimism also explained why identity complexity predicted deprovincialised behaviours. Therefore, reductions of intergroup concerns were a more general mediational pathway, while optimism seems more specific to complex identities. These findings resemble Velasco González, Verkuyten, Weesie, and Poppe’s (2008) showing that perceived Muslim threat mediate the relation between multi-ethnic sentiments and anti-Muslim attitudes in the Netherlands. Our findings were also in line with other research suggesting that decreased intergroup threats alleviate nationalistic tendencies (Paolini, Hewstone, Cairns, & Voci, 2004; Pettigrew & Tropp, 2008; Pettigrew, Tropp, Wagner, & Christ, 2011). We expected optimism to be a greater contributor, although it did manifest as a significant mediator for identity complexity.

Limitations and Future Research

The current article presented data from a survey poll that was not specifically designed to answer our research questions. This lead to a suboptimal match between the concepts we were interested in and the resulting operationalisations. For example, it is possible that our weighting of EU-related behavioural intentions did not match the underlying concept optimally. We ranked EU-related behaviours in order from requiring lowest to highest active involvement with another EU country but recognize that our respondents might see things differently (e.g. “Is living in another EU country indeed more involved than studying in another EU country?”).

Similarly, our decisions as to which issues facing the nation constitute intergroup issues is debatable. Our operationalisations focused on commonly discussed intergroup topics in the media and in the literature. However, it is possible that some respondents perceived other kinds of threat, such as economic issues or pensions, as mostly intergroup in nature. Further, while our operationalisation of identity complexity is in line with Dovidio and Gaertner’s conceptualisation of dual identities (e.g. Dovidio et al., 2009), it falls short of assessing more refined degrees of overlaps in identity contents, (e.g. as described in Van Dommelen, Schmid, Hewstone, Gonsalkorale, & Brewer, 2015). Finally, we expect the interest in EU-wide behaviours variable to be indicative of behavioural intention. However, this is not certain. While this large-scale data set was valuable in identifying potential avenues for social cohesion within the EU, the present findings need to be tested in future examinations that carefully operationalise and psychometrically validate the variables to be tested. Future research should also consider designs that afford confidence on the direction of causality. Longitudinal, multi-wave designs would be in the position to establish the temporal and causal order of constructs.

Due to the sample composition of the Eurobarometer survey, the present research mainly captured host citizens’ identities and perspectives. However, in previous research, complex dual identity structures have also been related to immigrants’ psychological and social adaptation in terms of wellbeing and harmonious host-immigrant relations (Berry, Phinney, Sam & Vedder, 2006). Therefore, it will be important to include minority and immigrants’ perspectives to understand consequences and mechanisms of minority and immigrants’ identity structures, cognitions, and behaviours which ultimately foster productive intergroup relations (see also Zagefka et al., 2014; Zagefka, González, & Brown, 2011).

Closing Remarks

In the current article, we looked at trans-national dynamics and touched on immigration processes through a multilevel lens (Tajfel & Turner, 1979). We investigated intergroup variables like intergroup concerns and intrapersonal variables like general optimism. It is fair to argue that our discipline is gaining from an increasing recognition of the interplay of processes that sit at different levels of analysis (e.g. Stürmer, Benbow, Siem, Barth, Bodansky, & Lotz-Schmitt, 2013; Turner & Feddes, 2011). This work makes a little contribution to these emerging trends.

Constrained by the archival nature of Eurobarometer data, we focused on majority members’ perspectives. The European Council (2004), however, recognises that integration is a two-way process involving immigrants and national citizens alike: While immigrants are asked to adapt to their host society, receiving societies are asked to provide the opportunity for full immigrant participation (Armbruster, 2011). Hence, host perspectives play an integral part of successful immigration policy which need to be investigated and understood. The present article contributed to this understanding by distilling social psychological processes implicated in how majority members develop interests in engaging in behaviours that transcend their national borders and that they (and we) regard as desirable. In light of the small effect sizes observed, it is imperative that these relations are tested and replicated in future research.