Support for collective action against refugees: The role of national, European, and global identifications, and autochthony beliefs

Abstract

To understand recent anti‐refugee protests in Europe, we examined how different levels of inclusiveness of group identities (national, European, and global) are related to intentions to protest among native Europeans. We focused on the mediating role of autochthony (a belief that the first inhabitants of a territory are more entitled) and the moderating role of threat. Survey data from 11 European countries (N = 1,909) showed that national identification was positively associated with autochthony, and therefore, with the intention to protest against refugees. In contrast, global identification was related to lower protest intentions via lower autochthony. These paths were found only among Europeans who perceived refugees as a threat. European identification was not related to the endorsement of autochthony or to collective action. These findings indicate why and when majority members are willing to participate in collective action against refugees, and underscore the importance of global identification in the acceptance of refugees.

1. INTRODUCTION

In the midst of the “refugee crisis”, over the past few years multiple protests and other forms of collective action have taken place in different European cities in opposition to the arrival of refugees.¹ Although empirical evidence suggests that people are more favorable towards involuntary than voluntary migrants (e.g., Augoustinos & Quinn, 2003; Gieling, Thijs, & Verkuyten, 2011; Verkuyten, Altabatabaei, & Nooitgedagt, 2018), it was in particular the arrival of refugees between 2014 and 2015 that incited these frequently violent protests. For instance, a recent study conducted in Germany (Benček & Strasheim, 2016) has registered 1,645 events of right‐wing violence against refugees, with a drastic increase of incidents since 2014. These protests have largely been captured by the media, affecting the public and political discourse with regard to the acceptance of refugees (Holmes & Castañeda, 2016).

Understanding the social psychological motivations of native Europeans to oppose refugees is important in order to promote inclusiveness and acceptance. In the present research, we focus on collective action oriented at preventing refugees from entering the country or at expelling them from the country. Such protest may differ from the more commonly studied collective action that limits the expressive rights of non‐Western migrants already residing in Europe (e.g., to found their schools or mosques). Collective action against the entrance of newcomers to a territory have implications not only for granting civic rights but also indirectly for saving human lives as these protests may serve to justify introducing restrictive policies oriented at closing European borders and thus diverting attention away from those who seek political asylum from violence occurring in their homeland. Compared to mobilization against other migrant groups, these protests may also be particularly driven by such justification mechanisms as autochthony beliefs because their aim is to prevent newcomers from “invading” the territory that is “ours”.

Collective action is one of the main motors of social change (Van Zomeren & Iyer, 2009) and it can challenge the policies, actions, and decisions made by authorities (Subaši, Reynolds, & Turner, 2008). What mobilizes people to participate in collective action has been a key question in social psychology, and this has been studied from different perspectives: from the minority's point of view as a means to stand up to a dominant majority (Van Zomeren, Postmes, & Spears, 2008); from the majority or dominant group's perspective as a way to express solidarity with disadvantaged groups (Subaši et al., 2008); or as a process through which majority members or high status groups exclude minorities (Postmes & Smith, 2009). Still, most research on collective action has focused on minorities, and studies assessing the willingness of advantaged groups to engage in collective action against a disadvantaged minority (e.g., refugees) are still scarce (Shepherd, Fasoli, Pereira, & Branscombe, 2017).

The aim of the present research is to study intentions to engage in collective action from the majority's perspective as a means to exclude refugees. We extend the existing literature by examining how, why, and when different levels of inclusiveness of group identities, namely, national, European, and global identification, are related to the willingness to participate in collective action opposing refugees among native Europeans. Related to the how, based on Social Identity Theory (Tajfel & Turner, 1979) and Common In‐group Identity Model (CIIM; Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993), we examine whether more inclusive categories reduce the willingness to oppose refugees. There are few studies addressing the question how European identity shapes responses towards minorities (Curtis, 2014; Kende, Hadarics, & Szabo, 2019; Klein, Licata, Azzi, & Durala, 2003; Kohli, 2000; Licata & Klein, 2002) and no research has associated European identification with collective action against a minority group by the dominant majority. Given that the recent so called “refugee crisis” was a European crisis, European identity might be especially relevant in predicting attitudes towards this particular minority group. Further, literature on the role of global identity in mobilizing collective action is also still limited and mostly focused on intergroup solidarity or positive collective action (e.g., Barth, Jugert, Wutzler, & Fritsche, 2015), but has not considered it as a tool for reducing hostile intergroup responses towards involuntary migrants.

Regarding the why, we test whether the associations between group identities and collective action against refugees can be explained via autochthony, an ideology that confers rights and entitlements on the original inhabitants of a territory. Only a few quantitative studies have been conducted regarding antecedents and consequences of autochthony, and there is an indication that this belief serves as a powerful justification for the exclusion of newcomers (see Martinovic & Verkuyten, 2013). Yet, no research has examined the mediating role of autochthony in the link between group identities and collective action in particular. In addition, existing empirical evidence suggests that autochthony is strongly related to the rejection of irregular migrants but unrelated to attitudes towards legal European migrants in Australia and the US (Martinovic, Verkuyten, Jetten, & Bobowik, 2019), and we thus argue that autochthony beliefs may be particularly relevant for understanding the processes underlying the exclusion of refugees arriving in Europe.

Related to when, we examine whether the association between autochthony and intentions to engage in collective action against refugees depends on the level of perceived intergroup threat because threat has been consistently shown to facilitate activation of hostile responses towards outgroups (Stephan, Renfro, Esses, Stephan, & Martin, 2005). We focus on threats that seem to be particularly triggered by irregular migrants and that the arrival of refugees has made salient. First, given the uncontrollable nature of this particular type of migration, Europeans might have the feeling that they are losing control over their country and borders (ownership threat, see Verkuyten & Martinovic, 2017). Second, because many refugees are Muslim, they can be perceived as a symbolic threat to Western values. Third, refugees can also pose a security threat to European citizens, because they are considered to be either perpetrators of sexual harassment or potential terrorists (Harris & Jesko, 2015). Our measure of threat thus encompasses these triggers specific to the arrival of refugees.

We address these questions using data collected in 2016 among 1,909 participants from 11 European countries, namely, Belgium, Croatia, France, Germany, Greece, Hungary, Italy, Serbia, Spain, the Netherlands, and the UK. Next to estimating the theoretical model on the overall sample, additional per country analyses were conducted to check whether the findings hold in the different national contexts. While the data are not nationally representative, the strengths are the sample size, the timing of the data collection in the midst of the refugee crisis, and the fact that the sample is regionally diverse for Europe.

1.1. Group identities and collective action

Collective action is defined in social psychology as acting on behalf of one's group in order to improve the group's conditions (Van Stekelenburg & Klandermans, 2013). It can be understood as any action with the purpose of improving a group's status, influence, or power (Tajfel & Turner, 1979). Importantly, collective action can be used as a tool for excluding those who pose a threat to the group's status, such as newcomers.

Group identification is one of the main predictors of collective action (Van Zomeren, Postmes & Spears, 2008). This has been widely evidenced for minority groups (Ellemers, 1993; Kelly & Breinlinger, 1996; Van Zomeren, Spears, & Leach, 2008; Wright & Tropp, 2002) and there are reasons to believe group identification plays a crucial role in a majority's intentions to participate in collective action as well. According to Social Identity Theory (SIT), the more people identify with a group, the more they are inclined to act on behalf of that group (Ellemers, Spears, & Doosje, 1999; Mummendey, Kessler, Klink, & Mielke, 1999; Van Stekelenburg & Klandermans, 2013), because group identification enhances shared perceptions within the in‐group in regard to the possibility and desirability of social change (Kelly, 1993).

National identification is one of the most studied forms of group identity and it has often been shown to be associated with exclusionist reactions towards newcomers (Coenders & Scheepers, 2004; Curtis, 2014; Pehrson, Brown, & Zagefka, 2009; Pehrson & Green, 2010; Verkuyten & Martinovic, 2015a), especially in situations of intergroup tensions (Brewer, 1999). The arrival of refugees in Europe in the past few years is an example of such a conflicting context, and we can argue that individuals who have high national identification want to protect their resources more and maintain their higher status by virtue of excluding refugees. Exclusionism may include a range of actions to keep lower status groups subdued, varying from overt expressions to more subtle forms, such as supporting actions and policies that have exclusionary effects (Postmes & Smith, 2009). Accordingly, we expected stronger national identification to be related to greater intentions to participate in collective action opposing refugees among members of the dominant ethnic groups in European countries.

In contrast, redefining the in‐group as a broader and more inclusive category might reduce intergroup conflict, as suggested by the CIIM (Gaertner et al., 1993). The highest level of social categorization is global identification or identification with humanity (see De Rivera & Carson, 2015; Reese, Proch, & Finn, 2015), and it is related to intergroup empathy and humanitarian relief (Buchan et al., 2011; De Rivera & Carson, 2015; Hackett, Omoto, & Matthews, 2015; McFarland, Webb, & Brown, 2012), both of which might be particularly relevant in shaping attitudes towards involuntary migrants, as some studies suggest (Nickerson & Louis, 2008). From the CIIM perspective, global identification would imply a recategorization process in which a former out‐group (i.e., refugees) becomes part of the in‐group (i.e., fellow humans), and therefore gets evaluated more positively. Thus, stronger global identification is expected to be related to weaker intentions to engage in collective action against refugees.

For citizens of European countries, European identification represents another layer of group identification that is situated between the national and global ones. Regarding European identification, theoretical arguments for its relationship to intentions to engage in collective action against refugees are not so clear‐cut. On the one hand, Europe as a social category is broader and more inclusive than the nation, so identifying with this supranational category (i.e., European identification) might promote stronger endorsement of multiculturalism, cosmopolitism, recognition of universal rights (Vertovec, 2010), and favorable attitudes towards immigrants in Europe (Curtis, 2014). This suggests that European identification might be related to weaker intentions to engage in collective action against refugees. On the other hand, although Europe is a more inclusive identity category compared to the nation, it is unlikely that refugees are perceived as part of this more inclusive in‐group. Just as national identification is sometimes related to ethnic exclusionism (Pehrson et al., 2009), strong European identification could also be associated with exclusionist reactions towards ethnic out‐groups that are not originally from Europe (Kende et al., 2019; Klein et al., 2003; Kohli, 2000; Licata & Klein, 2002). This implies that, in contrast to the earlier argument, European identification could be related to stronger intentions to engage in collective action against refugees.

1.2. Group identities and autochthony

Groups often share a sense of ownership of the territory they inhabit (Verkuyten & Martinovic, 2017) and this sense of ownership might be an important link that connects group identities to collective action against refugees. Being the first one to possess an object is usually perceived as a valid argument for claiming ownership of that object (Friedman, 2008). Similarly, the first inhabitants of a territory can claim ownership over the land by virtue of being there first (Martinovic & Verkuyten, 2013; Smeekes, Verkuyten, & Martinovic, 2015; Verkuyten & Martinovic, 2015b). Autochthony is therefore the ideology that assigns ownership and thereby related rights to the first inhabitants of a territory, and comes from the term “autochthon”, meaning “born from the soil” (Geschiere, 2009). Thus, groups that were the first inhabitants of a territory might feel particularly or exclusively entitled to decide over collective matters related to their territory and country (Martinovic & Verkuyten, 2013).

According to self‐categorization theory (SCT, Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), people who strongly identify with an in‐group are more likely to behave according to the in‐group's norms and beliefs. More specifically, whereas high identifiers will tend to adopt the dominant group ideology, low identifiers will be less likely to do so. Autochthony can be conceptualized as a hierarchy‐enhancing rather than a hierarchy‐attenuating belief. The former is an idea that legitimizes and contributes to the maintenance of or increase in group‐based inequalities, whereas the latter promotes egalitarian relations between social groups and the reduction of group‐based inequalities (Sidanius & Pratto, 1999). Martinovic and Verkuyten (2013) found that national identification is positively related to autochthony. The more people identified with their nation, the more they endorsed autochthony. Consequently, we expect stronger national identification to be positively related to autochthony.

However, people who have a strong global identification are more likely to hold an egalitarian worldview and express greater concerns for human rights (De Rivera & Carson, 2015; Hackett et al., 2015; McFarland et al., 2012). Individuals who strongly identify globally are more likely to endorse hierarchy‐attenuating beliefs, such as beliefs in global justice and the reduction of social inequalities. Accordingly, stronger global identification should be negatively related to autochthony beliefs.

When it comes to European identification, the expected association with autochthony beliefs is less straightforward. On the one hand, people who strongly identify as European might be less concerned with country‐specific borders, and therefore endorse autochthony beliefs less. On the other hand, given that the superordinate transnational category in Europe allows free borders in some areas (e.g., Schengen), the autochthony beliefs could be extrapolated to this higher level, using it as a means to exclude non‐European migrants, such as refugees, from the continent. If this were the case, European identification would be related to a stronger endorsement of autochthony beliefs.

1.3. Autochthony and collective action

Ownership also entails the right to forbid other people to use one's property (Friedman & Ross, 2011), and in a similar way, claims of autochthony can be used by the primo‐occupants of a territory to exclude newcomers from equal participation in the host society (Martinovic & Verkuyten, 2013). Some authors have argued that the endorsement of autochthony beliefs has been on the rise over the last couple of decades in Western European countries such as Belgium, the Netherlands, and Italy, and that autochthony, as a hierarchy‐enhancing ideology, has become a key argument when discussing immigration and multiculturalism (Ceuppens, 2006, 2011; Geschiere, 2009).

Higher status groups are more likely to endorse ideologies that maintain social inequality than lower status groups (Deaux, Reid, Martin, & Bikmen, 2006). These hierarchy‐enhancing beliefs may contribute to inequalities by defining who is entitled to certain rights and privileges and who can be excluded from these rights and privileges (Sidanius & Pratto, 1999). Accordingly, cross‐sectional and experimental research has found that endorsement of autochthony beliefs is related to out‐group prejudice (Martinovic & Verkuyten, 2013) and opposition to minority rights (Smeekes et al., 2015; Verkuyten & Martinovic, 2015a). Additionally, out‐group prejudice among advantaged group members has been shown to be positively associated with intentions to engage in collective action against immigrant groups (Shepherd et al., 2017).

Further, previous research has shown that certain systems of beliefs such as political ideologies (Hindriks, Verkuyten, & Coenders, 2014; Rios Morrison & Ybarra, 2009) or colorblindness (Dovidio, Gaertner, & Saguy, 2015) are justifications that facilitate expression of prejudice. Existing cross‐sectional and longitudinal studies confirm that autochthony also serves as a belief justifying the exclusion of newcomers (Martinovic et al., 2019). Given the—from the European perspective—uncontrollable nature of the arrival of refugees in the territory of the European nation‐states, we expect that the argument of autochthony will be particularly important in legitimizing collective action against refugees.

1.4. Out‐group threat as a trigger of collective action

Members of advantaged groups may be interested in maintaining their prestigious and distinct group identity in order to keep positive collective esteem (Tajfel & Turner, 1979). However, minority groups can threaten the positive distinctiveness and prestigious group identity of the advantaged groups (see Jetten, Spears, & Postmes, 2004). As a response, majority group members might engage in collective action as a way to exclude minority groups (Postmes & Smith, 2009).

This idea is in line with the group position model (Bobo & Hutchings, 1996) that posits that competition over scarce resources between social groups leads to antagonistic intergroup attitudes and possibly conflicts between groups. According to this model, in order for negative attitudes towards an out‐group to arise, feelings of proprietary claims need to go along with a fear that the minority is “getting out of place” and is therefore threatening the power and dominant position of the majority. Along these lines, Martinovic and Verkuyten (2013) have shown that autochthony was related to negative attitudes towards immigrant minorities, but this was only the case for natives who felt threatened. Moreover, identification with an advantaged group whose distinctiveness might be lost in the future increased the effects of threat on the willingness to endorse in‐group protective action (Wohl, Giguère, Branscombe, & McVicar, 2011). Therefore, it is likely for autochthony beliefs to lead to more intentions to protest against refugees, particularly for those majority members who feel more threatened by refugees.

In sum, we expected stronger national identification to be related to more intentions to participate in collective action against refugees via stronger endorsement of autochthony beliefs (H1). Conversely, people who strongly identify globally were expected to be less willing to participate in collective action against refugees due to their weaker endorsement of autochthony beliefs (H2). European identification was expected to be associated either with more (H3a) or less willingness to participate in collective action (H3b) via autochthony beliefs. Finally, we hypothesized that autochthony beliefs would be related to more protest intentions for people who felt more threatened (H4). Bringing these hypotheses together, the mediation processes were expected to be stronger for people who perceived higher levels of threat (H5). Our hypotheses were not preregistered.

2. DATA AND METHODS

2.1. Data and participants

Data were collected in 2016 among 2,112 participants residing in 11 European countries: Belgium, Croatia, France, Germany, Greece, Hungary, Italy, Serbia, Spain (specifically, the Basque Country), the Netherlands, and the UK (an additional 617 participants opened the link to the survey but did not answer any of the items used in this study, so they were excluded). We worked with convenience samples and we aimed for at least 100 participants per country.² Participants who self‐identified as natives were selected, decreasing the sample to 1,909 people (see Table 1 for per country sample sizes). In the Basque Country participants who self‐identified as Basque or Spanish were included, and in Belgium those who self‐identified as Belgian, Walloon, or Flemish. The age of the participants ranged from 16 to 79, and was measured in years (M = 29.73, SD = 12.46).³ Females represented 53% of the participants. Our participants showed a somewhat more liberal left‐wing political orientation, with 26.1% placing themselves at the political center, 21.4% at the political right (center right, right and strong right), and 48.6% at the political left (center left, left, and strong left). Further, regarding occupational status, although 53.0% of our sample were undergraduate students, almost a half of our sample represent employed or self‐employed (32.3%), unemployed (6%), retired (3.5%), and housewives/housemen (1.7%). Student populations ranged from 41.6% (the Netherlands) to 81.7% (Italy and Germany), except for Croatia where 18% of the sample were students, and except for France where only students were recruited. As regards participants who were studying, in the majority of the countries the survey was administered to psychology students (except for Spain in which they were criminology students). Even though not representative, our sample was reasonably diverse in terms of its socio‐demographic composition.

Table 1.

Means, standard deviations, and reliability statistics of the six constructs in the 11 countries and total sample (N = 1,909)

	N	Collective action intentions			Autochthony			Threat			National identification			European identification			Global identification
		M	SD	ρ	M	SD	ρ	M	SD	ρ	M	SD	r s	M	SD	r s	M	SD	ρ
Belgium	106	3.43	2.30	0.950	4.09	1.85	0.939	3.94	2.22	0.982	5.76	1.25	0.893	4.74	1.68	0.926	3.87	1.32	0.777
Croatia	468	3.28	1.44	0.692	4.35	1.70	0.942	3.84	1.53	0.942	4.62	1.72	0.915	4.22	1.60	0.930	4.53	1.35	0.801
France	60	2.42	1.17	0.577	3.18	1.57	0.932	2.52	1.22	0.923	5.27	1.33	0.808	4.41	1.59	0.924	3.83	1.24	0.731
Germany	356	1.97	1.19	0.782	2.88	1.50	0.932	2.22	1.25	0.945	4.18	1.50	0.837	4.57	1.51	0.875	3.99	1.16	0.714
Greece	146	2.61	1.05	0.538	3.02	1.81	0.946	2.48	1.29	0.933	5.31	1.53	0.895	4.59	1.67	0.883	5.30	1.08	0.762
Hungary	155	2.76	1.17	0.711	3.90	1.62	0.935	3.59	1.48	0.924	5.58	1.42	0.874	5.47	1.42	0.863	4.11	1.13	0.647
Italy	100	2.30	1.13	0.791	3.50	1.47	0.932	2.93	1.15	0.925	5.32	1.31	0.888	4.67	1.33	0.917	4.78	1.10	0.762
Netherlands	118	2.29	1.60	0.881	3.57	1.61	0.923	3.00	1.60	0.948	5.29	1.41	0.729	4.51	1.63	0.863	4.34	1.47	0.831
Serbia	157	2.50	1.43	0.740	3.96	1.66	0.924	2.61	1.32	0.914	4.72	1.80	0.918	3.97	1.73	0.904	4.17	1.32	0.756
Spain	117	2.00	1.12	0.611	3.07	1.74	0.946	2.17	1.26	0.947	4.19	1.86	0.913	4.66	1.60	0.894	4.71	1.23	0.778
UK	126	2.41	1.25	0.768	3.65	1.42	0.941	3.19	1.38	0.945	4.64	1.36	0.855	4.27	1.68	0.943	4.12	1.48	0.860
Total sample	1,909	2.62	1.47	0.753	3.63	1.72	0.941	3.05	1.58	0.951	4.79	1.64	0.892	4.52	1.62	0.905	4.35	1.32	0.768

For all the variables the range was 1–7. ρ = Raykov's composite reliability scale, r _s  = Spearman‐Brown correlation. These values were estimated based on the 6‐factor CFA model. National and European identification were measured by two items, global identification and collective action by three items, autochthony by five items, and threat by nine items.

Undergraduate students participated voluntarily in exchange for course credits, and in some cases students’ mailing lists were used asking for their participation. To diversify the sample, additional participants were recruited via social media (an online survey was posted on the researchers’ Facebook and Twitter accounts asking native people to participate) and they were asked to spread the link to the online survey among their native friends and relatives, turning it into a snowball sampling.⁴ The survey was offered in the main language of the respective countries.⁵ The original survey was created in English and later translated to the respective languages by the research group in every country. No back translation method was used. Prior to filling in the survey, participants were informed about the aim of the research, the voluntary nature of their participation, and the right to withdraw their participation at any moment, and they were assured their answers would be treated anonymously.

2.2. Measures

All items were measured using a Likert‐scale ranging from (1) completely disagree to (7) completely agree. The full scales are presented in Appendix A. We provide Raykov's composite reliability score (Rho) because it accounts for measurement error and it does not produce an underestimation bias as Cronbach's alpha (Raykov, 1997).

2.2.1. National identification

It measured the extent to which people identified with their native country and comprised two items: “I strongly feel [nationality]” and “My [national] identity is an important part of me”. These items were taken from previous studies (e.g., Verkuyten, 2009) and they were strongly correlated (r _s = 0.89 in the total sample; range between r _s = 0.73 and r _s = 0.92 across countries).

2.2.2. European identification

It assessed the extent to which people identified with Europe and it consisted of two items: “I strongly feel European” and “My European identity is an important part of me”. The items were strongly correlated (r _s = 0.91; range across countries from 0.86 to 0.93).

2.2.3. Global identification

It measured identification on the highest level of social abstraction (Reese et al., 2015), and was captured by three items from the global‐human identity scale by Der‐Karabetian and Ruiz (1997): “I feel like my fate and future are bound with all of humankind”, “I feel I am related to everyone in the world as if they were my family”, and “I think of myself as a citizen of the world”. The items formed a reliable scale (Raykov's composite reliability ρ = 0.77; range from ρ = 0.65 to ρ = 0.86 across countries).

2.2.4. Autochthony

It comprised five items that assessed a belief that the first inhabitants of a country are more entitled to certain rights than newcomers. These items have been used in previous research on autochthony beliefs (Martinovic & Verkuyten, 2013). Examples are “The earliest inhabitants of a country are more entitled than newcomers to decide about important national matters”, “Every country belongs primarily to its first inhabitants”, and “The earliest inhabitants of a country should have the most right to define the rules of the game”. The items formed a reliable scale (ρ = 0.94 in the total sample; range from ρ = 0.92 to ρ = 0.95 across countries).

2.2.5. Threat

It was covered by nine items that assessed the extent to which natives perceived refugees as a menace to their culture (e.g., “Refugees pose a threat to the [country's] culture”); their economic wellbeing (e.g., “Refugees are increasing the tax burden on [country's] citizens”), or their right to control and decide over the country (e.g., “I am worried that the refugees will pretty soon start telling the [natives] how to run this country”). Reliability tests showed the nine items formed a consistent scale (total sample ρ = 0.95; range from 0.91 to 0.98 across countries).

2.2.6. Collective action intentions

Collective action intentions were assessed with three items that measured the extent to which participants were willing to engage in activities in order to achieve a common goal. The items were adapted from a scale used by Van Zomeren, Spears, and Leach (2008). Participants were asked to what extent they would be willing to “sign a petition to restrict the number of refugees arriving in [country]”, “participate in a demonstration against refugees” and “support a policy aiming at expelling refugees from Europe”. The items formed a reliable scale (ρ = 0.75 in the total sample and range between ρ = 0.54 and ρ = 0.95 across countries; see Table 1). Participants responded on a scale ranging from 1 very much unwilling to 7 very much willing.

2.2.7. Control variables

We controlled for two standard demographic characteristics: gender (0 = female, 1 = male) and age measured in years. Additionally, we controlled for political orientation, which has been associated with higher levels of autochthony (e.g., Verkuyten, Martinovic, Smeekes, & Kros, 2016) and higher out‐group derogation (Van Prooijen, Krouwel, Noiten, & Eendebak, 2015). We used the political self‐placement scale (Jost, 2006). Participants were asked to indicate their political stance on a scale from 1 strongly left to 7 strongly right.

2.3. Method of analysis

Data were prepared using the Statistical Package for Social Sciences (SPSS), version 24 and later exported to MPlus in order to run the analyses.⁶ Firstly, confirmatory factor analyses were conducted in order to test whether the latent constructs were empirically distinct. Secondly, the theoretical model was tested using structural equation modeling (SEM), with collective action, autochthony, three different types of identification, and perceived threat introduced as latent factors, using the overall sample (i.e., without distinguishing between countries). Due to the skewed distribution of the dependent variable “collective action” (50% of the responses fell into answer categories 1 and 2, towards “strongly disagree”), the estimator used was MLR, i.e., maximum likelihood estimation with robust standard errors and with adjusted chi‐square tests that are robust to non‐normally distributed populations (Muthén, 2002). Finally, in order to test the robustness of the results we explored whether the countries were comparable and if the findings could be replicated across countries.

Missing values, which were assumed to be missing at random, were dealt with in MPlus (version 7.3, Muthén & Muthén, 1998–2017) using Full Information Maximum Likelihood Estimation. The exogenous variables (the three types of identification and threat) were endogenized. In the total sample, there were 39 missing values for collective action (2.0%); 60 for autochthony (3.1%); 5 for threat (0.3%); 44 for national identification (2.3%); 47 for European identification (2.5%) and 45 for global identification (2.4%).

3. RESULTS

3.1. Measurement model

A confirmatory factor analysis with six correlated latent constructs (national, European, and global identification, autochthony, threat, and collective action intentions) was fitted using the total sample. The model fitted the data reasonably well: comparative fit index (CFI) = 0.947, Tucker‐Lewis coefficient (TLI) = 0.939, root mean square error of approximation RMSEA = 0.056, 95% CI [0.054; 0.059], standardized root mean square residual (SRMR = 0.036). Although the MLR (i.e., calculated with Yuan‐Bentler scaling correction factor) chi‐square test was significant, χ² (239) = 1693.052, p < 0.001, this is often the case with relatively large samples (Kline, 2005). Each of the items loaded high on the designated factor, with standardized loadings ranging from 0.69 to 0.93. To identify the model, the residual variance of the items “I strongly feel [nationality]” and “I strongly feel European” were set to zero. We compared this model to a model in which national identification and European identification were forced to load on one factor. The new model had a worse model fit (ΔAIC = 2,020.01). Similarly, we tested a model in which European identification and global identification were forced to load on one factor, which also had a worse model fit (ΔAIC = 1,559.53). We can conclude that the three types of identification are empirically distinct. Additionally, we compared the aforementioned model to a model in which threat and collective action were forced to load on one factor. The new model had a significantly worse model fit (TRd⁷  = 398.12(5), p < 0.001). We can conclude that threat and collective action are two separate and empirically distinct constructs. Moreover, it is important to note that we combined the nine items of threat into an overarching measure because even though the three threat factor model fitted the data as well as a one threat factor model (three factor: RMSEA = 0.051, CFI 0.960, SRMR = 0.034 v/s one factor: RMSEA = 0.056, CFI. 947, SRMR = 0.036, df = 276), the three types of threat were highly correlated with each other (symbolic and realistic threat: r = 0.951, p < 0.001; symbolic and ownership threat: r = 0.921, p < 0.001; realistic and ownership threat: r = 0.910, p < 0.001) and thus had to be considered as one construct in the subsequent analyses.

3.2. Descriptive results

Means and standard deviations of the total sample and per country are presented in Table 1. In the total sample, intentions to engage in collective action were significantly lower than the midpoint of the scale (Wald (1) = 1,363.881, p < 0.001), indicating that participants on average were not willing to engage in collective action opposing refugees. Similarly, participants on average held relatively low autochthony beliefs compared to the midpoint of the scale, (Wald (1) = 83.07, p < 0.001), and perceived low levels of threat from refugees (Wald (1) = 649.80, p < 0.001). Further, participants showed relatively high levels of national (Wald (1) = 391.38, p < 0.001), European (Wald (1) = 174.70, p < 0.001), and global identification (Wald (1) = 99.11, p < 0.001), suggesting that participants’ bonds to their respective countries, Europe, and the world were important aspects of their identities.

Looking at the correlations in the total sample, both national identification and European identification correlated positively with intentions to engage in collective action and the endorsement of autochthony beliefs (see Table 2). In contrast, global identification was negatively correlated with both intentions to engage in collective action against refugees and with the endorsement of autochthony beliefs. Additionally, perceived threat was strongly and positively correlated with the endorsement of autochthony beliefs and intentions to participate in collective action. The particularly high correlation between threat and collective action intentions suggests both constructs share strong commonalities. However, as stated previously, these constructs are separate factors both empirically and theoretically. European identification correlated positively with both national and global identification, whereas the latter two did not correlate with each other.

Table 2.

Correlations between the core constructs in the total sample (N = 1,909)

		1.	2.	3.	4.	5.
1.	Collective action intentions
2.	National identification	0.304***
3.	European identification	0.073*	0.397***
4.	Global identification	−0.406***	−0.030	0.249***
5.	Autochthony	0.627***	0.323***	0.071**	−0.264***
6.	Threat	0.878***	0.317***	0.067*	−0.387***	0.637***

*p < 0.05.

**p < 0.01.

***p < 0.001.

3.3. Mediation by autochthony in the link between identifications and collective action against refugees

In order to test our hypotheses, we first estimated a mediation model in which collective action was regressed on autochthony, three different types of identification, perceived threat, and the control variables: age, gender, and political orientation. Furthermore, autochthony was regressed on the identification constructs and the controls. The model fitted the data well, CFI = 0.934, TLI = 0.922, RMSEA = 0.059, 95% CI [0.057; 0.061], SRMR = 0.090.

Results are presented in Table 3. We found that national identification was related to more and global identification to less support for autochthony, whereas European identification was not related to autochthony (see direct effects). Autochthony was furthermore related to more willingness to engage in collective action against refugees. Combining these core findings, national identification was indirectly associated with more willingness to engage in collective action, via a higher endorsement of autochthony beliefs, as shown by a significant indirect effect. This finding is in line with H1. Conversely, and following H2, global identification was negatively related to collective action via a lower belief in autochthony. Regarding European identification, neither of the contrasting mediation hypotheses (H3a and H3b) was supported, as shown by a non‐significant indirect effect.⁸ In addition, threat was related to more willingness to engage in collective action against refugees.⁹

Table 3.

Direct, indirect, and total effects from a mediation model of collective action intentions (N = 1,909)

	B	SE	p‐Value
Direct effects
NI on Autochthony	0.232	0.032	<0.001
EI on Autochthony	0.031	0.031	0.324
GI on Autochthony	−0.357	0.051	<0.001
NI on CA	−0.005	0.023	0.816
EI on CA	0.034	0.023	0.128
GI on CA	−0.100	0.033	0.003
Autochthony on CA	0.136	0.027	<0.001
Threat on CA	0.764	0.030	<0.001
Indirect effects
NI on CA via autochthony	0.032	0.008	<0.001
EI on CA via autochthony	0.004	0.004	0.331
GI on CA via autochthony	−0.049	0.011	<0.001
Total effects
NI on CA	0.026	0.024	0.277
EI on CA	0.039	0.023	0.096
GI on CA	−0.149	0.034	<0.001

Unstandardized coefficients presented. Significant coefficients are emphasized in bold.

Abbreviations: CA, Collective Action; EI, European Identification; GI, Global Identification; NI, National Identification.

3.4. Moderated mediation

Next, to test our hypotheses about moderated mediation, we extended the model above by adding a latent interaction term between autochthony and perceived threat in relation to collective action against refugees. Results indicated that there was a significant and positive interaction effect, thereby supporting H4 (see Figure 1). In order to further interpret the interaction, we conducted simple slope analyses in which the results were estimated for participants who perceived high levels of threat (i.e., one standard deviation above the mean) and low levels of threat (i.e., one standard deviation below the mean). Results revealed that autochthony and collective action were positively and significantly related only at high levels of threat, whereas at low levels of threat the two were unrelated (see Table 4). Furthermore, the indirect effects of national and global identification on collective action, via autochthony, were only found among participants who perceived high levels of threat (in line with H5). The analyses did not yield significant moderated mediation effects for European identification.

Figure 1.

Moderated mediation model of collective action against refugees (N = 1,909). Note: Unstandardized coefficients, with standard errors in parentheses. Total effects in square brackets. **p < 0.01, ***p < 0.001. Control variables age, gender, political orientation were included in the model but are not shown in the figure

Table 4.

Simple slope analysis for a moderated mediation model of collective action intentions (N = 1,909)

	High threat			Low threat
	B	SE	p‐Value	B	SE	p‐Value
Total effects
NI on CA	0.053	0.025	0.034	−0.001	0.023	0.969
EI on CA	0.045	0.024	0.054	0.038	0.022	0.081
GI on CA	−0.188	0.035	<0.001	−0.105	0.033	0.001
Direct effects
Autochthony on CA	0.260	0.026	<0.001	0.031	0.026	0.233
Indirect effects
NI on CA via autochthony	0.061	0.010	<0.001	0.007	0.006	0.242
EI on CA via autochthony	0.008	0.008	0.313	0.001	0.001	0.440
GI on CA via autochthony	−0.094	0.016	<0.001	−0.011	0.009	0.234

Unstandardized coefficients presented. Significant coefficients are emphasized in bold.

Abbreviations: CA, Collective Action; EI, European Identification; GI, Global Identification; NI, National Identification.

3.5. Country comparisons

As a robustness check, we examined whether the pattern of associations found in the main analysis was similar across the eleven countries. To do a meaningful comparison, we first examined whether the items measured the same constructs by testing for measurement invariance. Due to the small sample size of some countries, we tested for measurement invariance between three groups: the two countries with the most participants, namely Croatia (N = 468) and Germany (N = 356), and the remaining nine countries pooled together (N = 1,085). We pooled the remaining nine countries because none of them reached a necessary minimum sample size required for a model estimated with 27 items. The rule of thumb is to have 10 participants for each item and thus we would need 270 participants minimum for each group. Unfortunately, only two countries fulfilled this requirement, and thus the remaining countries were pooled together. Metric invariance was accepted across the three groups (see Appendix B), which suggests the regression coefficients may be compared between groups, and presumably across the 11 countries.

Then we estimated the moderated mediation by fitting a multi‐group structural equation model while splitting the sample per country. However, in this case, we used observed variables (i.e., mean scores of every variable) because (as indicated earlier) the relatively small sample size per country did not allow fitting a model with six latent factors, and we omitted the control variables, whose exclusion from the model tested on the pooled sample also did not affect the main conclusions, to increase power. In order to compute the interaction term between autochthony and perceived threat both variables were centered on the country's mean. It is important to note that by using observed variables, measurement error is not explicitly represented (Kline, 2005), and therefore the path coefficients might be biased, usually towards zero. Yet, it was the best available option to test if the relations could be replicated across countries given the small sample sizes, and due to their low statistical power the focus of this robustness check was on the direction of the associations between variables rather than their significance.

In line with our previous results, perceived threat was positively related to collective action intentions, and national identification was positively related to autochthony, in all 11 countries. Furthermore, for global identification we replicated the negative association with autochthony beliefs in 10 out of 11 countries (except for Hungary), and autochthony beliefs were positively associated with collective action in nine countries, Italy and Spain (the Basque Country) being the exceptions (see Appendix C). Thus, we can conclude that these paths were rather robust, even though they did not always reach significance.

The interaction between perceived threat and autochthony in predicting collective action was positive in eight out of 11 countries (reaching significance in four), however, it was negative in France and even negative and significant in Greece and Italy, which calls for some caution when evaluating hypothesis 4. Although the positive interaction between threat and autochthony was significant only in four countries, these effects were not statistically different from the positive interaction coefficient obtained in several other countries, indicating a trend towards a positive interaction but one that did not always reach significance, Wald(1) = 3.243, p = 0.072.

Simple slope analyses revealed that in four countries with a positive and significant interaction term (Belgium, Germany, the Netherlands, Spain), participants who perceived higher levels of threat were significantly more willing to participate in collective action, whereas for those who perceived lower levels of threat the simple slope was not significant in three countries (Belgium, Germany, the Netherlands), and in Spain it was negative, in line with our predictions. Further, in Croatia, Hungary, and the UK—the countries with a positive but not significant interaction term—the simple slope was positive both for those who perceived high and low threat (though significant only in Croatia) (see Appendix D). Next, we assessed simple slopes for Greece and Italy, the countries with the unexpected negative interaction term. For those who perceived high threat the simple slope was not significant in Greece and it was negative and significant in Italy. In both countries the simple slope was positive and significant for those who perceived low threat. This means that in these two countries autochthony is related to less collective action among those who feel threated, which is the reverse of what we expected.

Finally, the association between European identification and autochthony was rather inconsistent. It was positive in six countries (most notably in Italy and the Netherlands) and negative in five (with the largest effect in Greece, followed by Hungary and Belgium), which might explain why this association was not significant in the main analysis with the pooled sample.¹⁰

4. DISCUSSION

The arrival of refugees is one of the biggest challenges contemporary Europe is facing, and has not necessarily been successfully dealt by the European Union (EU). For instance, the EU imposed more restrictive border and security policies instead of responding to this tragedy by saving lives and protecting human rights of those arriving in Europe. These restrictive policies are frequently reflected in the predominant discourse in the media, which frame the arrival of refugees as a “crisis” or an “avalanche”. They also legitimize responses of those European communities which perceive newcomers as a threat to their welfare and safety. Our research provides some responses regarding how to stop the “tsunami” of protests and other forms of collective action aimed at expelling refugees from Europe, frequently accompanied by hate speech and hate crime incidents. Thus, our research can serve to address the phenomenon of increasing overt racism in practice, which may be an equally (or even more) urgent matter than promoting prosocial responses such as providing basic humanitarian aid.

Further, why and when members of disadvantaged minorities participate in collective action has often been a key question in social and behavioral sciences, yet why and when majority group members get involved in collective action against a less advantaged group, such as refugees, has been studied less frequently. In this article we contributed to the scientific literature by examining how different levels of inclusiveness of group identities were related to intentions to participate in collective action against refugees among native Europeans. We examined if the endorsement of autochthony beliefs (i.e., more entitlements for first comers) could explain the link between group identities and collective action. To the best of our knowledge, this is the first article to examine the association between autochthony beliefs and collective action intentions. Additionally, we studied whether this association was conditional upon perceived threat.

Using a sample of 1,909 native Europeans collected in 11 countries, we found that national identification was related to stronger endorsement of autochthony beliefs, which confirms previous findings from Martinovic and Verkuyten (2013), and these beliefs were in turn related to stronger intentions to participate in collective action opposing refugees. These results were also quite consistent across countries. More precisely, national identification was associated positively with autochthony in all countries (and significantly in eight of them). In turn, autochthony was also related to willingness to protest against refugees in most (nine) countries, although this path did not always reach statistical significance. Thus, our findings provide reasonable support for our hypothesis regarding the mediating role of autochthony in the link between national identification and collective action against refugees. This resonates with the existing literature where stronger national identification has been shown to go hand in hand with a stronger sense of psychological ownership of the country (Brylka, Mähönen, & Jasinskaja‐Lahti, 2015), enabling high national identifiers to believe they are more entitled to make decisions over the country than newcomers. In turn, a sense of ownership often implies the right to prohibit other people from using one's property (Friedman & Ross, 2011), which is frequently used as an argument to exclude foreigners (Geschiere, 2009).

In contrast, global identification was related to less support for autochthony, and therefore weaker intentions to participate in collective action against refugees. It has been shown that people who identify globally tend to endorse hierarchy‐attenuating, egalitarian views about society rather than hierarchy‐enhancing beliefs (autochthony being an example of the latter), and therefore are more willing to engage in prosocial behaviors that could reduce global inequality (Reese, Proch, & Cohrs, 2014). This is consistent with the CIIM (Gaertner et al., 1993), which suggests that if people strongly identify with broader social categories, in which former members of the out‐group are recategorized as members of the in‐group, intergroup relations are likely to improve. Along these lines, our research showed not only with pooled data but also in separate countries that those participants who identified globally were less likely to support autochthony, and were therefore less willing to participate in collective action against refugees (the link between global identification and autochthony was negative in 10 out of 11 countries, and this relationship was significant in eight of them), thus confirming that our findings together provide convincing evidence for our hypothesis regarding the mediating role of global identification.

European identification, however, was not associated with the endorsement of autochthony beliefs, and therefore it was not related to intentions to participate in collective action. A further inspection showed that in some countries the association between European identification and autochthony was negative (most notably in Greece but also showing a negative trend in Hungary and Belgium), whereas it was positive in other countries (particularly strongly in Italy and the Netherlands). This suggests that Europe can sometimes work as an inclusive category, promoting multiculturalism, pluralism, and the acceptance of newcomers, whereas in other contexts it can work as a more exclusive category, in which the continent boundaries are used as a way to exclude newcomers from non‐European countries. In line with our findings, Licata and Klein (2002) found that Belgian participants associated the EU with humanistic values and cultural diversity (assessing these values positively) and that, paradoxically, those who showed higher European identification, and thus acknowledged and adhered to these values, expressed less tolerance towards immigrants. A study from Hungary also showed a negative link between European identification (glorification component) and anti‐immigrant and anti‐Muslim attitudes (Kende et al., 2019). Klein et al. (2003) found a similar negative link among Greek participants. One possible explanation for these negative effects in countries such as Greece and Hungary is that in both societies the meaning of belonging to the European community might have been affected by either economic crisis or a shift towards a populist right‐wing government, respectively. That is, it may actually be Euro‐skepticism (low European identification) that, together with strong national identification, mobilizes people to protest against accepting refugees through autochthony beliefs.

Accordingly, Europe as a supranational entity might be related to conflicting attachments. Although it links nation‐states politically and economically, it might not link them culturally, forming an overarching identity that might entail “contradictory meanings or logics of action” (Kohli, 2000, p. 131). This suggests that there is no clear meaning or content of European identity, and therefore there is no clear association of European identification with autochthony beliefs and collective action intentions across countries. In may be that this link differs across countries because in contexts where European identity is based on more ethnic (instead of civic) concepts of citizenship autochthony may be used as a legitimizing myth. Existing research indeed suggests that the endorsement of ethnic citizenship is related to more exclusionary reactions towards newcomers and this relationship is due to a weaker normative sense of common national belonging and higher adherence to autochthony beliefs (Verkuyten & Martinovic , 2015a). These findings suggest that European identity may also become an exclusive category used to justify hostile reactions towards refugees. Future research could test the role of representations of citizenship in this link.

Further, examining European data, Curtis (2014) found that the positive link between European identification and favorable views towards immigrants was stronger under conditions of cross‐cutting cleavages (for instance, a simultaneous membership in multiple social groups, see Curtis, 2014 for more detailed definition) and in countries with longer duration of membership in the EU. Therefore, future research (including more European countries) should take into account multiple possible country‐level moderating factors that would explain differential effects of European identification on protest against refugees across countries.

Regarding the role of threat, although our data confirmed that threat is the main and the strongest driver of collective action against refugees, we still found that autochthony beliefs add a significant contribution to the explanation of collective action over the effects of threat. Using the pooled data, we also provided empirical support for our expectation that the association between autochthony and collective action was conditional upon the level of perceived threat. In accordance with the group position model (Bobo & Hutchings, 1996), our results indicated among the overall sample that the positive association between autochthony and collective action was only present for those participants who perceived that refugees were posing a threat to the majority's culture, security, or sense of ownership and control. For participants who did not perceive refugees as a threat, autochthony was not related to intentions to participate in collective action. This suggests that endorsement of autochthony may be a driver for people to want to participate in collective action against refugees; however, as long as refugees do not pose a threat to the native's dominant or advantaged position, autochthony will not translate into collective action against refugees. Consequently, higher perceived threat strengthened both indirect paths from national and global identification to collective action. National identification was associated with more autochthony and therefore more willingness to participate in collective action particularly for those participants who perceived more threat. This finding is similar to what Martinovic and Verkuyten (2013) found with regard to out‐group prejudice. Higher national identification was associated with more autochthony beliefs and via these beliefs to stronger out‐group prejudice among those majority members who feared the minority group was getting out of hand. Conversely, global identification was in the present study related to less willingness to protest via lower autochthony, and this “pacifying” effect of global identification was particularly beneficial for participants who felt threatened. This hints at the importance of promoting broader and more inclusive group identities particularly among people who are prone to feeling threatened, which might reduce their exclusionary reactions towards refugees.

Yet, the results about the positive moderating effect of threat were not always consistent when comparing the countries. The association between autochthony and collective action turned out to be less positive for participants in Greece and Italy (and to some extent France) who perceived high levels of threat, as compared to those participants who perceived low levels of threat. In fact, in Italy and Greece, the more participants who do not feel threatened by the arrival of (numerous in the case of these countries) refugees endorse autochthony, the more they are willing to engage in collective action against these newcomers. In Italy, it is even the case that participants who feel threatened use autochthony claims to disengage from protest against refugees. One explanation is that in these countries, with a significant presence of refugee camps, the humanitarian drama is more salient and thus may mobilize solidarity among the population that is in favor of protecting human rights of refugees entering Europe, even alongside existing autochthony beliefs. However, the same findings were not obtained in Hungary, Croatia, or Serbia, which also received a substantial number of refugees. Yet, compared to Italy or Greece, the governmental control of the situation (e.g., closing borders, the presence of the military in the zones of the refugee camps) might have been stricter in these countries and at the same time the local community might have been less engaged in contact and empathized less with the refugee population. This pattern is worth exploring in future research in order to examine whether the negative interaction can be replicated, and whether the opposite pattern of results is related to the particular context of these Mediterranean countries (Italy and Greece). Nevertheless, the conditional effect of autochthony on willingness to protest against refugees among those who feel threatened was positive (though not always significant) in the remaining seven countries and it is consistent with the theoretical expectations and previous findings, indicating a clear trend.

4.1. LIMITATIONS AND FUTURE RESEARCH

Our participants were not too willing to engage in collective action against refugees, which was reflected in the skewed distribution of this measure. According to rational choice theory, individuals as well as groups perform a cost‐benefit analysis before deciding whether to take action in a particular matter (Ostrom, 1998). Moreover, people who highly identify with an in‐group will engage in collective action as long as they perceive the action as beneficial to the entire group (Louis, Taylor, & Neil, 2004). Engaging in collective action against refugees might be considered a high cost activity for many native Europeans when compared to the expected benefits of participating. However, low willingness in our sample could also have to do with the fact that the data were not nationally representative and the participants were above average young and mostly liberal.

Another limitation of the present study is that it was cross‐sectional, so we cannot make any causal claim about the mediating paths. We have strong theoretical reasons to argue that national identification increases support for autochthony and, via stronger autochthony beliefs, motivates people to participate in collective action against refugees, but it could be that people who are more willing to exclude refugees justify this by making autochthony claims or that this ideological belief increases national identification. Although testing an alternative mediation model rendered similar results, we argue that this general autochthony belief (or ideology) serves as a legitimizing myth for further exclusion of out‐group members. Our reasoning goes in line with previous empirical evidence on the association between identity and ideology. For instance, using correlational and experimental design, Rios Morrison and Ybarra (2009) showed that group identification moderated the link from threat to SDO, thus showing that strong in‐group identification is a condition that motivates people to endorse certain types of ideology. Hindriks et al. (2014) also found that national identification was a moderator of the relationship between SDO and prejudice. As such, it seems more likely theoretically that higher national identifiers resort to the arguments of autochthony to exclude newcomers than that those who believe that first comers are generally more entitled would therefore start identifying more with their nation (and less with global identity). Nonetheless, future research should explore these associations using longitudinal and experimental data in order to examine the causality of the associations found.

Moreover, the data available do not rule out the possibility of other hidden variables affecting the correlations found in this study. Shepherd et al. (2017) found that threat has an indirect effect on negative emotions (e.g., anger) and prejudice, which in turn predict intentions to participate in collective action against disadvantaged groups. Additionally, individual characteristics such as personality traits (e.g., agreeableness) should be taken into account, given that they could have a positive impact on prosocial behavior and therefore reduce willingness to protest against refugees. Perceived efficacy and perceived injustice are also two factors that play a role in collective action intentions (Van Zomeren, Spears & Leach, 2008). Future research would benefit from studying the effects of autochthony on collective action intentions while including the abovementioned factors in order to determine the extent to which autochthony plays a role in collective action relative to other relevant mechanisms.

The use of convenience samples further hinders the possibility of comparing the findings across national contexts. Given that the data were collected partly among students and partly among a wider population by means of snowball sampling, there might be a selection bias: Participants were not randomly drawn from the population of every country but were dependent on the respondents who were assessed first (Atkinson & Flint, 2001). Yet, despite these inconveniences, the data have the strength of having been collected during the time in which the “refugee crisis” was hotly debated, covering 11 European countries, and containing a sample size sufficiently large to test our hypotheses. Further, although 53% of our participants were undergraduate students and 48.6% identified with the political left, almost the half of the sample were non‐students and 47.6% identified with political center or right wing, this demonstrating that our sample was still reasonable diverse. Nevertheless, future research should use representative data of the country's population and larger sample sizes per country, enabling a more accurate comparison between them.

Also, in order to properly test for measurement invariance across the 11 countries, we ideally should have used country as a grouping variable. However, given the small per‐country sample sizes, we believe we have chosen the best possible approach to our data by testing invariance across three groups: two countries with large enough samples—a Western European country (Germany) and a Southern European country (Croatia)—and a third group consisting of the rest of the participants pooled together.

Furthermore, in this study we examined the effect of national identification on autochthony and collective action towards refugees without making a distinction between civic and ethnic understandings of national belonging. Previous research indicates that ethnic understanding of national identity is associated with higher adherence to autochthony, and therefore lower acceptance of immigrant rights, whereas a civic understanding is related to a weaker support for autochthony and higher acceptance of minority rights (Verkuyten & Martinovic, 2015a). This suggests that civic and ethnic understandings of national belonging are differently related to autochthony, and therefore could have different effects on intentions to participate in collective action against refugees. Future research should include both conceptions of national belonging and examine their effects on collective action. Additionally, given the inconclusive results regarding the effect of European identification on autochthony beliefs and collective action, future research should help unpack the meaning of European identity, and further explore how European identification is related to autochthony beliefs and exclusionist reactions in different national contexts.

Finally, we believe that our research contributed significantly to the literature on refugees because we focused on specific forms of collective action in response to the arrival of a refugees, such as signing a petition preventing refugees from entering the country, as well as types of threat relevant in the context of the arrival of refugees, such as ownership threat and security threat. However, both autochthony and threat were measured regarding the country of residence of participants, whereas some aspects of collective action referred to protesting against the inflow of refugees to Europe in general. Thus, future research could compare the effects of autochthony and threat at both the European and national level on exclusionary responses towards voluntary and involuntary migrants.

4.2. CONCLUSION

We can conclude that autochthony beliefs are an important motivation to exclude refugees and a relevant link between group identities and collective action intentions against refugees. Whereas national identification is positively related to the exclusion of refugees via higher autochthony beliefs and the effects of European identification on collective action are inconclusive, global identification can decrease exclusionary reactions towards refugees in Europe, particularly among people who feel threatened. Promoting and strengthening social categories that embrace and include disadvantaged out‐groups is of crucial importance in order to decrease hostile responses towards refugees in Europe.

CONFLICT OF INTEREST

The authors have no conflict of interest to declare.

AUTHOR CONTRIBUTION

Paola Hasbún López, Borja Martinović, and Magdalena Bobowik designed the research, analyzed the data, and wrote the article. The remaining authors collected the data in their countries. All the authors have given their final approval of this version of the article.

ETHICAL STATEMENT

We confirm that research reported in this manuscript has been conducted in accordance with the APA Code of Conduct and the authors’ national ethics guidelines. Informed consent was obtained from the participants, participation was voluntary, and anonymity has been guaranteed. Results are reported honestly. The submitted work is original and not (self‐)plagiarized, and authorship reflects individuals’ contributions.

TRANSPARENCY STATEMENT

The data, codebook, and syntax are available online via Open Science Framework. See footnotes 5 and 6 for the links.

APPENDIX 1.

1.1.

Table A1 Collective action intentions items

Item 1	Sign a petition to restrict the number of refugees arriving to [country]
Item 2	Participate in a demonstration against refugees
Item 3	Support a policy aiming at expelling refugees from Europe

Table A2 Autochthony items

Item 1	The earliest inhabitants of a country are more entitled than newcomers to decide about important national matters.
Item 2	Every country belongs primarily to its first inhabitants.
Item 3	The earliest inhabitants of a country should have the most right to define the rules of the game.
Item 4	“We were here first” is an important principle for determining who decides on what happens in a country.
Item 5	The ones who arrived first in a country can be considered more rightful owners of the country than those who arrived later.

Table A3 Threat items

Item 1	Refugees pose a threat to the [country] culture.
Item 2	Refugees are increasing the tax burden on [country] citizens.
Item 3	Due to the arrival of many refugees, native [country] are losing their say in political matters.
Item 4	[Country] customs are threatened by refugees.
Item 5	Refugees pose a health risk for the [country]
Item 6	I am worried that the refugees will pretty soon start telling the [country] how to run this country.
Item 7	The refugees’ religion is incompatible with the dominant religion in Europe.
Item 8	With the increase of refugees comes the increase of crime rates in the [country]
Item 9	Native [country] are slowly losing the [country] to refugees.

Table A4 Identification items

National identification
Item 1	I strongly feel [nationality]
Item 2	My [national] identity is an important part of me
European identification
Item 1	I strongly feel European
Item 2	My European identity is an important part of me
Global identification
Item 1	I feel like my fate and future are bound up with all of humankind
Item 2	I feel that I am related to everyone in the world as if they were my family
Item 3	I think of myself as a citizen of the world

APPENDIX 2. Measurement invariance between Croatia, Germany and the remaining countries pooled together

2.1.

First, an unconstrained (configural) model was fitted in which the factor loadings, intercepts and means were allowed to be different between the three sets of countries. The model fitted the data reasonably well after allowing for the error terms of two pairs of items (item 6 and 9 of threat, and item 2 and 3 of collective action) to covary, CFI = 0.952, TLI = 0.944, RMSEA = 0.054, SRMR = 0.036. The Santorra‐Bentler Chi‐square test was significant, χ²(711) = 2,025.19, p < 0.001, but this is common with large samples. Theoretically, items 6 and 9 of the threat measure refer to ownership threat, namely, fear of losing the country to refugees, so it is likely that these two items share more commonalities between each other than with other items of threat. Similarly, items 2 and 3 of collective action refer to expelling refugees who are already in the countries, whereas item 1 is related to restricting refugees arriving in the future.

In a second step, we constrained the loadings to be equal across groups, thus fitting a metrically invariant model. Although this model had a significantly worse fit compared to the configural model based on the Chi‐square difference test (Santorra‐Bentler Δχ²(36) = 193.18, p < 0.001), it still had a reasonable model fit that was similar to the fit of the unconstrained model when other indices were considered: CFI = 0.946, TLI = 0.940, RMSEA = 0.056, SRMR = 0.050. A large total sample can lead to the over‐rejection of measurement invariance tests if the χ² is the only criterion used (Putnick & Bornstein, 2016), and changes in the fit indices tend to be less sensitive to sample sizes (Cheung & Rensvold, 2002).

APPENDIX 3.

3.1.

Table C1. The main paths in the moderated mediation model broken down by country

	BE (N = 106)	CR (N = 468)	UK (N = 126)	FR (N = 60)	GE (N = 356)	GR (N = 146)	HU (N = 155)	IT (N = 100)	NL (N = 118)	SE (N = 157)	SP (N = 117)
	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)
AU → CA	.124 (.079)	.178*** (.039)	.094 (.071)	.197† (.102)	.071* (.031)	.123 (.079)	.096 (.066)	−.051 (.059)	.170* (.075)	.024 (.074)	−.030 (.054)
AU*Threat → CA	.048† (.026)	.024 (.018)	.032 (.041)	−.116 (.097)	.049* (.021)	−.102** (.037)	.017 (.027)	−.065* (.027)	.132*** (.022)	.029 (.037)	.129*** (.030)
EI → AU	−.091 (.076)	.077 (.056)	−.024 (.083)	.057 (.183)	−.008 (.063)	−.185† (.097)	−.106 (.111)	.238* (.113)	.253** (.081)	.026 (.077)	.050 (.099)
GI → AU	−.561*** (.089)	−.228*** (.060)	−.202* (.099)	−.036 (.158)	−.257*** (.062)	−.166 (.104)	.017 (.099)	−.414*** (.114)	−.470*** (.076)	−.241** (.090)	−.451*** (.109)
NI → AU	.523*** (.131)	.289*** (.052)	.074 (.092)	.174 (.179)	.258*** (.057)	.722*** (.094)	.369*** (.091)	.108 (.114)	.399*** (.202)	.352*** (.075)	.289*** (.070)
Threat → CA	1.004*** (.108)	.561*** (.045)	.593*** (.096)	.349* (.142)	.706*** (.049)	.564*** (.099)	.538*** (.065)	.778*** (.081)	.470*** (.081)	.816*** (.095)	.525*** (.080)

Abbreviations: AU, Autochthony; BE, Belgium; CA, collective action; CR, Croatia; EI, European identification; FR, France; GE, Germany; GI, global identification; GR, Greece; HU, Hungary; IT, Italy; NI, National identification; NL, Netherlands; PO, Political orientation; SE, Serbia; SP, Spain; UK, United Kingdom.

^† p < .10, *p < .05, **p < .01, ***p < .001.

APPENDIX 4.

4.1.

Table D1. Simple slopes analyses for the interaction term between autochthony and threat

	High threat			Low threat
	B	SE	p	B	SE	p
BEa	.230	(.079)	.003	.017	(.079)	.828
CR	.217	(.039)	<.001	.140	(.039)	<.001
FR	.056	(.102)	.587	.339	(.102)	.001
GEa	.132	(.031)	<.001	.010	(.031)	.757
GRa	−.008	(.079)	.918	.255	(.079)	.001
HU	.121	(.066)	.067	.071	(.066)	.283
ITa	−.798	(.059)	<.001	.697	(.059)	<.001
NLa	.372	(.075)	<.001	−.032	(.075)	.672
SE	.062	(.074)	.403	−.014	(.074)	.849
SPa	.133	(.054)	.013	−.192	(.054)	<.001
UK	.139	(.071)	.052	.050	(.071)	.481

^aCountries in which the interaction term was significant.

Hasbún López P, Martinović B, Bobowik M, et al. Support for collective action against refugees: The role of national, European, and global identifications, and autochthony beliefs. Eur J Soc Psychol. 2019;49:1439–1455. 10.1002/ejsp.2608