To address major global challenges, such as health pandemics and climate change, people need to cooperate in increasingly large and complex groups (Richerson & Boyd, 1998). Partly resulting from this need, regional integration became a global trend in the last century (Hurrell, 1995), for example in Europe (European Union; EU), Asia (Associations of Southeast Asian Nations), Africa (East African Community, ECOWAS, African Union), and South America (Mercosur, ANDEAN Community). Yet, to be effective in solving problems, these regional organizations must be perceived as legitimate by their citizens. That is, the effectiveness and viability of international governance organizations largely depend on whether the public believes that these institutions are entitled to exercise authority, because such legitimacy appraisals safeguard people’s voluntary deference and long-term support (Buchanan & Keohane, 2006; Tyler, 2006; Tyler & Jackson, 2013). Perceived legitimacy constitutes, in that sense, an essential feature of sustainable regional cooperation.
Scholars, most notably from political science, are therefore interested in individual- and societal-level factors which explain when and why people perceive international governance organizations as (il)legitimate (Tallberg & Zürn, 2019). One of the main individual-level factors suggested to be relevant here is social identity (Dellmuth, 2018; Dellmuth et al., 2022; Dellmuth & Schlipphak, 2020). Social identity is however a complex and dynamic concept, involving multiple (nested) group memberships which interact with contextual factors (Ellemers, 1993; Turner, 1999). This article aims to complement prior research on public legitimacy perceptions of regional organizations by elucidating the role of social identity processes.
We focus on the EU, for which declining perceived legitimacy is one of its current key challenges (European Commission, 2017), because the EU is a unique form of supranational cooperation, positioned between a state and an international organization. Since nationality is one of the most important group memberships of EU citizens (Scharfbillig et al., 2021; Sindic, 2011), we are interested in the question how national identities affect people’s legitimacy perceptions of the EU. A survey was conducted to explore how different forms of social identification (i.e., national identification, EU identification, dual national-EU identification) are related to perceived legitimacy of the EU and whether these relations are moderated by the extent to which the EU is seen as a threat to the nation in instrumental (i.e., threat to power) and symbolic (i.e., threat to sociocultural identity) terms. By examining these social psychological dynamics, we seek to contribute to a better understanding of how EU legitimacy perceptions develop. In turn, such an understanding may ultimately inform the ongoing scholarly and societal debate on how to better design the constitutional nature and structure of the EU, including its relationship with member states, so as to enable sustainable cooperation in Europe.
Conceptualizing and Operationalizing Perceived EU Legitimacy
In the broadest sense, legitimacy is about the relationship between an object of legitimacy and an audience, which is characterized by 1) mutual expectations, 2) conformity to these expectations by the object, and 3) assent by the audience (Schoon, 2022). Depending on academic literature and unit of analysis, different definitions and operationalizations exist. As in the current work we are interested in the audience as unit of analysis, legitimacy becomes a perception that an audience (here, EU citizens) holds about an object (here, the EU) or about their relationship with it. Accordingly, perceived legitimacy is defined as “a generalized perception or assumption that the actions of an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions” (Suchman, 1995).
As the concept of legitimacy may be too complex for people to understand, perceived legitimacy is often assessed through related – but conceptually distinct – constructs, such as support, trust, confidence, acceptance, or the felt duty to obey (Schoon, 2022). From these constructs, “support” may be the least adequate to measure legitimacy. To illustrate, according to a recent article, perceived legitimacy is a multidimensional belief system including moral convictions, which are not necessarily but sometimes complemented with self-interest calculations (Dellmuth & Schlipphak, 2020). This article showed that support items mainly capture the self-interest dimension, while trust items may tap into the moral dimension. Support for EU membership, for example, can be driven purely by perceived economic benefits, without the presence of any feeling of attachment to the EU. Outcome satisfaction, however, may not be sufficient for believing in the rightness of authority (Dellmuth & Schlipphak, 2020; Tallberg & Zürn, 2019).
A common approach in the literature on the perceived legitimacy of international organizations is therefore to measure legitimacy through self-reported “confidence”, which is argued to capture a feeling of faith that goes beyond self-interest (e.g., Dellmuth et al., 2019). Closely related to confidence is “trust”, which is often used to measure legitimacy in research on national authorities (e.g., Jackson et al., 2012; Sunshine & Tyler, 2003). In these studies, items measuring the “felt duty to obey” an authority are typically also employed as an indicator of legitimacy. Feeding this back to the broad definition of legitimacy, trust and confidence items capture the conviction that an authority will conform to mutual expectations (e.g., “I have confidence in the EU”), while duty to obey measures capture the feeling of assent to the authority (e.g., “I feel I should accept the decisions made by the EU”).
In this contribution, we therefore operationalize perceived legitimacy as institutional trust and felt duty to obey. As will be discussed in the method section in more detail, a principal factor analysis revealed these two indicators of legitimacy as different factors, leading us to treat them as separate outcome variables in the main analyses. Thus, our operationalization does not include self-interest calculations nor moral convictions, which we consider to be antecedents rather than elements of legitimacy.
Social Identity and Perceived EU Legitimacy
Legitimacy is closely related to morality and identity. Sharing moral values with other people contributes to a sense of belonging to the same group, and both moral alignment and group identification in turn lead to higher trust in other group members (Ellemers et al., 2013; Ellemers & Van der Toorn, 2015). Similar processes have been shown to lead to trust in political authorities (Berg, 2019). That is, when people feel respected by, and identify with the group an authority represents, they perceive this authority to be more trustworthy and legitimate and are in turn more willing to voluntarily accept and obey it (Tyler, 1997; Tyler & Jackson, 2013; Tyler & Lind, 1992). Relatedly, when people perceive an authority to have an appropriate sense of right and wrong, they are more likely to justify its power exercise and show more obedience (Jackson et al., 2012, 2015). Factors signalling identity and shared values, such as perceived procedural justice, but also receiving positive outcomes, may therefore be sources of legitimacy (Tyler & Blader, 2003).
The idea that stressing a common EU identity increases the perceived legitimacy of the EU can thus be based on previous psychological theories (cf., Sindic et al., 2019). Public opinion studies have, correspondingly, found that identification with the EU is associated with higher support for political unification and integration in the EU (Chalmers & Dellmuth, 2015; Sindic et al., 2019) and lower support for leaving the EU (Van de Vyver et al., 2018).
In addition, the extent to which the EU identity leaves room for national identities may equally be relevant for predicting the perceived legitimacy of the EU. Through European integration, a new, superordinate group membership was imposed on citizens. The integration of a new identity with existing identities can take different forms (Amiot et al., 2007). Prior research has shown that “dual identification”, which entails a form of simultaneous identification where subgroup identities are not de-emphasized but rather recognized and valued within the superordinate group, generates positive social identities (Crisp & Hewstone, 2007; Dovidio et al., 2007; Hornsey & Hogg, 2000a, 2000b; Schmid et al., 2009). With regards to the European context, dual identification would entail a form of common EU identification in which the national identity is positively integrated, resulting in a positive social identity. High dual identifiers could therefore be expected to perceive the EU as more legitimate.
It is however less clear how a strong national identity, which is arguably deeper rooted than an EU identity, is related to legitimacy perceptions of the EU. A pressing question is therefore whether national identities are an obstacle to EU identity and legitimacy, and if so, under which circumstances (Fuchs, 2011). Prior studies have shown that exclusive identification with one’s country is associated with lower support for EU membership (Carey, 2002; Hooghe & Marks, 2005), lower trust in the EU and its institutions (Clark & Rohrschneider, 2019; McLaren, 2007), and higher support for leaving the EU (Cislak et al., 2019). Yet, to measure national and EU identification, these studies often employ the so-called “Moreno” question, which asks respondents to select one of four identification categories (see Materials section), and which is included in every Eurobarometer, the Commission’s public opinion survey. This question has its limitations, however, because it assumes a tension between national and EU identity and fails to capture varying strengths of both identities (Bruter, 2008; Guinjoan & Rodon, 2016). Indeed, when national identity is assessed with continuous measures, findings on the relation between national identity and supportive attitudes towards the EU are more mixed (e.g., Boomgaarden et al., 2011; Carey, 2002). Thus, although it seems that exclusive national identification undermines perceived EU legitimacy, it remains unclear how the relative strength of national identification affects EU legitimacy perceptions.
In forming expectations about the relation between national identification and perceived EU legitimacy, the ingroup projection model is particularly relevant (Bianchi et al., 2010; Wenzel et al., 2007, 2016). According to this model, people tend to project typical ingroup characteristics on the superordinate group, a tendency that is positively related to ingroup identification. More precisely, when people strongly identify with their ingroup, for example their country, they are also more motivated to promote their group’s power, status, values, and distinctiveness (Tajfel, 1974; Tajfel & Turner, 1979). One way of achieving this is by portraying the ingroup as more prototypical of an overarching superordinate identity than other groups falling under this umbrella. Thus, people who strongly identify with their nation may be more likely to project their nation’s norms and values on the EU. Accordingly, given that legitimacy is the perception that the actions of an authority fit people’s existing cognitive system of norms and values, people who highly identify with their nation may perceive the EU to be more legitimate through ingroup projection.
Social Identity, EU Legitimacy, and Threat to the Nation
Understanding how individuals relate to groups not only requires insight into individual levels of social identification, but also into how these levels interact with contextual and social factors (Turner, 1999). For example, other groups that threaten the own group’s power, resources, values, or identity elicit defensive responses, especially among people who strongly identify with their group (Branscombe et al., 1999; Rios et al., 2018). In superordinate group situations, this may also concern a threat to the own group’s distinctiveness (Hogg & Terry, 2000). In a similar vein, attitudes towards superordinate bodies are not a mere function of social identification, but are formed in interaction with the perceived threat of superordinate group membership to one’s subgroup identities (Sindic & Reicher, 2009).
Following this reasoning, legitimacy perceptions of the EU may be influenced by the dynamics between people’s social identification and concerns about the EU. While the EU can be a source of a positive “EU identity”, it may also conflict with national identities (Hooghe & Marks, 2009). For example, people have shown to judge and experience the loss of national power and culture due to European integration as highly threatening (Boomgaarden et al., 2011; Licata, 2003). A perceived threat to the national identity has been negatively related to support for EU membership (Carey, 2002; Christin & Trechsel, 2002; Obradović & Sheehy-Skeffington, 2020). It has also been positively related to rejecting the Constitutional Treaty in 2005 and voting “Leave” in the UK membership referendum in 2016 (Hobolt & Brouard, 2011; Van de Vyver et al., 2018).
Yet, as the consequences of these identity concerns for legitimacy perceptions of the EU remain unclear, the aim of this study is to clarify the relation between threat and legitimacy. We expect that experienced threat to a relevant subgroup identity, such as a member state of the EU, will undermine the perceived legitimacy of supranational authority. As indicated above, legitimacy arises in the presence of an authority’s conformity to expectations that people hold about the right way to exercise power (Schoon, 2022; Suchman, 1995). To some extent respecting national interests, autonomy, and values is something that people, especially those with high national identification, should expect from the EU. Indeed, the EU has promised to respect national identities, in particular as inherent in their fundamental political and constitutional structures (see preamble to the Treaty on European Union and Article 4 of this treaty as well as the further respect inherent in the principles of conferral, subsidiarity, and proportionality). In addition to these formal guidelines regarding the relation between the EU and its peoples, the protection of national sovereignty and identity is for many people a deeply engrained moral intuition (Feldman, 2021). When the EU threatens national power and values, it might thus breach mutual expectations about what the EU should do for its members states and member peoples, harming the relationship with its audience. This may in turn undermine the perceived legitimacy of the EU.
Following this line of reasoning, we propose that the relation between social identification and perceived legitimacy of the EU is influenced by the experience of threat to the nation. More specifically, the relation between national identification and legitimacy may depend on people’s perception of threat because strong national identifiers are more likely to project national attributes on the EU (Wenzel et al., 2007). Consequently, strong national identifiers will tend to feel more “betrayed” when they perceive the EU to pose a threat to their country’s interests. As indicated above, for people with a strong EU identification or people whose national identities are positively integrated into the EU identity (i.e., high dual identifiers), generally higher legitimacy perceptions are expected. Whether this relation further depends on the experience of threat remains an open question. On the one hand, high dual identifiers may be less influenced by threat to the nation because they have higher levels of attachment to the EU and accordingly are less affected by a perceived interference of the EU with their country. On the other hand, they may simultaneously have high levels of national identification, meaning that concerns about national power and values may also decrease their perceptions of EU legitimacy.
The Present Research
A survey study was conducted in six EU member states. The main aims of this study were to examine (1) how different levels of social identification (i.e., national, EU, and dual) are related to the perceived legitimacy of the EU, and (2) whether experiencing the EU as a threat to the nation’s power and/or sociocultural identity moderates these relations. We also aimed to obtain descriptive information on social identification and threat experiences among EU citizens. Considering the limitations of the Moreno question, we included not only this question but also other, validated scales of identification in the survey. As there were no specific expectations before data collection and analysis, this contribution provides an exploratory examination of social identity dynamics and threat in determining EU legitimacy perceptions.
Method
Procedure
Data were collected via the online participant recruitment platform Prolific using a self-report questionnaire designed with Qualtrics software.1 Based on nationality and the criterium to be fluent in English, 1200 individuals from Finland, France, Germany, Italy, the Netherlands, and Poland were invited to participate in the study. These countries were selected to cover a more or less balanced spread in terms of their political and economic power, geographical location, and date of accession. When individuals agreed to participate, they were directed to Qualtrics. Before starting the study, which took approximately ten minutes, participants provided informed consent. After the study, participants were thanked, debriefed, and redirected to Prolific, where they were reimbursed with £1. The study was approved by the Psychology Research Ethics Committee of the Faculty of Social Sciences at Leiden University. Data, materials, and supplementary tables and figures are available as Supplementary Materials.
Participants
Responses from a total of 1180 participants were recorded. The data of participants who failed to pass two attention checks (n = 36) or finished the study in less than five minutes (n = 6) were removed from the sample. Furthermore, two participants from Finland were removed from the sample due to missing data. This resulted in a final sample of 1136 participants from six nationalities (Mage = 27.60, SD = 8.98, 58.4% male). Participants were asked to indicate their political orientation on two scales ranging from “Left” (1) to “Right” (100) covering both an economic (M = 43.47, SD = 26.47) and social (M = 28.11, SD = 24.46) dimension of political orientation. As described in more detail and for each of the country subsamples separately in Table S1 and Figure S1 of the Supplementary Materials, participants were relatively young, highly educated, and politically oriented to the left.
Materials
National Identification
National identification was measured using the social identity scale (Ellemers et al., 1999). This validated scale assesses three components of social identity: self-categorization (3 items, e.g., “I am like other [nationality] people”), group self-esteem (4 items, e.g., “I feel good about [nationality] people”), and commitment to the group (3 items, e.g., “I would like to continue belonging to [nationality] people”). The items were measured on a Likert scale ranging from 1 (not at all) to 7 (very much). For the current analyses, we used the composite of all items (α = .90).
EU Identification
EU identification was measured with the same ten items as national identification, adjusted to the EU identity (e.g., “I am like other people in the EU”; α = .89).
Dual Identification
To measure dual identification, we used one item: “I feel I belong to both [nationality] people and people in the EU”, which was answered using a Likert scale ranging from 1 (not at all) to 7 (very much).
Moreno Question
The Moreno question, derived from the Standard Eurobarometer, asked: “Do you see yourself as:”, with four answer categories: [nationality] only; [nationality] and European; European and [nationality]; and European only. These four answer options represent four levels of social identification. The first option concerns exclusive national identification; the second and third options both concern dual identification but differ in the precedence of either the national or EU identity; the final option concerns exclusive EU identification.
Threat to National Power and Sociocultural Identity
We created 12 items to measure threat to national power and sociocultural identity. Items were answered on 7-point Likert scales ranging from 1 (strongly disagree) to 7 (strongly agree). An exploratory factor analysis revealed a two-factor solution (see Table S2, Supplementary Materials, for items and details of the analysis), one factor that could be interpreted as “national power threat” and one factor that could be interpreted as “sociocultural identity threat”. However, since these factors were strongly correlated (r = .77), since we had similar predictions for both types of threat, and to simplify the analyses, both scales were combined into one composite variable, i.e., “threat to nation”. Additional regression analyses for the separate forms of threat showed highly similar results to the results of the analyses for the composite scale.2
Perceived Legitimacy
The perceived legitimacy of the EU was measured with nine items covering two subscales: institutional trust and felt duty to obey. We adapted four previously-used items reflecting institutional trust and five items reflecting felt duty to obey (Johnson et al., 2014; Sunshine & Tyler, 2003). Answers could be provided on 7-point Likert scales ranging from strongly disagree to strongly agree. We conducted exploratory principles axis factoring, using oblique promax rotation, to assess the factor structure (Carpenter, 2018) (see Table 1). A scree plot suggested that two factors should be retained. The final solution revealed one factor with four items to assess institutional trust (eigenvalue = 2.94; α = .90) and one factor with six items to assess felt duty to obey (eigenvalue = 3.17; α = .91). We therefore decided to run the regression analysis separately for both subscales.
Table 1
Item | Factor 1: Duty to obey | Factor 2: Trust | ||
---|---|---|---|---|
1. I have confidence in the EU | .87 | |||
2. The EU is trustworthy | .99 | |||
3. The EU acts within the law | .77 | |||
4. Most EU civil servants do their job well | .70 | |||
5. People should obey laws made by the EU even if it goes against what they think is right | .81 | |||
6. People should obey laws made by the EU even if they will not be caught for breaking it | .61 | .17 | ||
7. People should do what the EU tells them to do even if they disagree with it | .96 | |||
8. People should do what the EU tells them to do even if they don’t like the way the EU treats them | .90 | |||
9. I feel I should accept the decisions made by the EU | .63 | .30 | ||
Model fit indices | ||||
TLI | RMSEA | χ2 | df | p |
.95 | .10 | 229.46 | 19 | < .001 |
Note. Results from a three-factor solution revealed that a third factor had an eigenvalue lower than 1, hence two factors were retained. Factor loadings in bold indicate onto which final factors the item loaded, based on a recommended cut-off level of .32 (Carpenter, 2018).
Results
Identification and Threat Across Countries
The average ratings of all levels of identification were relatively high, with EU identification (M = 5.33, SD = 0.97) being higher than national identification (M = 4.83, SD = 1.14), except in the Finnish and Dutch sample (see Table S3, Supplementary Materials, for identification descriptives in country sub samples). Figure 1 compares the dual identification scores with the scores on the Moreno question, which measures four categories of dual identification. When only looking at the Moreno question, the findings suggest that across countries, a dual identification representation where national identity had precedence over EU identification was the most common identification (see Figure 1). Only a few participants, ten of which came from the Polish sample, identified exclusively with the EU (n = 27). Yet, when also considering the levels of dual identification (M = 5.13, SD = 1.57), it becomes clear that this does not imply that dual identification in general in the entire population was low.
Figure 1
The distribution of the threat scores for all country subsamples are depicted in Figure 2. The average ratings of threat scores in the entire sample suggest a fairly low level of threat to national sociocultural identity (M = 2.42, SD = 1.25), and a slightly higher level of threat to national power (M = 2.98, SD = 1.51). One-way analyses of variance (ANOVA) revealed small to medium significant differences between countries in power threat, F(5, 1130) = 6.89, p < .001, = .03, and identity threat, F(5, 1130) = 9.539, p < .001, = .04 (see Tables S4 and S5, Supplementary Materials, for the results from Tukey HSD pairwise comparisons).
Figure 2
The average rating of perceived legitimacy of the EU for the entire sample was 4.62 (SD = 1.15). An ANOVA on perceived legitimacy showed a small significant difference between countries, F(5, 1130) = 5.35, p < .001, = .02. Perceived legitimacy was lowest in the Polish sample (M = 4.40, SD = 1.17) and highest in the Italian sample (M = 4.88, SD = 1.03), ΔM = 0.48, 95% CI [-0.81, -0.15], p < .001 (see Table S6, Supplementary Materials, for the results from all Tukey HSD pairwise comparisons). Another ANOVA showed a large significant difference on perceived legitimacy between categories of dual identification, F(3, 1132) = 77.36, p < .001, = .17. Exclusive national identifiers had lower legitimacy perceptions (M = 3.84, SD = 1.33) than dual identifiers with a precedence of the national identity (M = 4.74, SD = 1.04), ΔM = -1.21, 95% CI [-1.47, -0.96], p < .001, and than dual identifiers with a precedence of the EU identity (M = 4.99, SD = 0.97), ΔM = -1.78, 95% CI [-2.09, -1.47], p < .001.
To examine the relationships between the main variables, Pearson correlations were calculated (Table 2). These showed that perceived legitimacy had a small positive association with national identification and strong positive associations with EU and dual identification. National identification correlated higher with the duty to obey subscale of legitimacy, while EU and dual identification correlated higher with the trust subscale. Furthermore, power and sociocultural identity threat were strongly related to each other and negatively related to the perceived legitimacy of the EU.
Table 2
Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
---|---|---|---|---|---|---|---|---|---|
1. Age | – | ||||||||
2. Political orientation economic | -.08** | – | |||||||
3. Political orientation social | .11*** | .43*** | – | ||||||
4. National identification | .15*** | .08** | .18*** | – | |||||
5. EU identification | .04 | -.16*** | -.27*** | .19*** | – | ||||
6. Dual identification | .04 | -.11*** | -.16*** | .34*** | .65*** | – | |||
7. Threat to nation | .10*** | .28*** | .49*** | .13*** | -.58*** | -.38*** | – | ||
8. Perceived legitimacy: trust | -.05 | -.14*** | -.29*** | .05 | .63*** | .50*** | -.66*** | – | |
9. Perceived legitimacy: duty to obey | .04 | -.12*** | -.16*** | .07* | .45*** | .39*** | -.50*** | .66*** | – |
10. Perceived legitimacy: total | <.01 | -.14*** | -.24*** | .06* | .58*** | .48*** | -.63*** | .88*** | .94*** |
*p < .05. **p < .01. ***p < .001.
With regard to the demographic variables, an ANOVA revealed small significant differences between different levels of education regarding perceived legitimacy, F(1, 1134) = 32.10, p < .001, = .03, although perceived legitimacy did not linearly increase with every higher level in education (see Table S7, Supplementary Materials, for the results from Tukey HSD pairwise comparisons). Political orientation on both the economic and social dimension were negatively related to perceived legitimacy. Age was unrelated to perceived legitimacy.
Predicting Legitimacy From Identification and Threat
We conducted two basic multiple linear regression models where the three types of identification (national, EU, dual) were regressed on the subscales of perceived legitimacy of the EU. To examine whether the relation between identification and legitimacy is moderated by threat to the nation, threat was included as an interaction term with each of the identification predictors. Since we were not interested in the question which variable could explain the most variance in the outcome variables, the models were not hierarchical. All continuous variables were mean-centred.
As some of the main variables (i.e., political ideology, EU identification, threat to the nation, institutional trust, felt duty to obey) were highly correlated, we conducted an overall factor analysis to test whether these variables were empirically distinct or loaded on the same dimension. The results confirmed that the items that measured the two subscales of legitimacy loaded on a different factor than the items that measured EU identification and threat to the nation (see Figure S2, Table S8, and Table S9, Supplementary Materials). Trust was thus an indicator of legitimacy, and closer related to legitimacy than to EU identification. Political ideology loaded on the same factor as the items assessing threat to national power. Finally, multicollinearity was not an issue in any of the regression models (all VIF’s > 1.12 and < 3.03).
Country was included as a control variable, as significant differences between countries on threat and legitimacy were found. We did not control for age, which was unrelated to legitimacy, nor for education, because differences between education levels on legitimacy were small and prior research suggested that education does not causally influence political attitudes about the EU (Kunst et al., 2020). Controlling for political orientation on both economic and social dimensions resulted in negative regression coefficients for these variables, while they were positively correlated to both legitimacy subscales; the other regression weights remained similar. As this suggests a suppression effect, which may be caused by the high correlation with the threat variable, the political orientation control variables were removed from the analyses.
The model with institutional trust as outcome variable (see Table S10, Supplementary Materials) had good predictive value, F(12, 1124) = 121.10, p < .001, adjusted R2 = .56. It showed that institutional trust was positively predicted by EU identification, b = 0.28, 95% CI [0.22, 0.34], SE = 0.03, p < .001, and dual identification, b = 0.13, 95% CI [0.08, 0.19], SE = 0.03, p < .001, and negatively predicted by threat to the nation, b = -0.44, 95% CI [-0.49, -0.39], SE = 0.03, p < .001. In addition, it revealed an interaction effect between national identification and threat, b = -0.07, 95% CI [-0.10, -0.03], SE = 0.02, p = .001), indicating that the relation between national identification and institutional trust in the EU is different at different levels of threat (see Figure 3). Simple slope analyses showed that national identification was positively related to institutional trust when threat was low, 1 SD below the mean; b = 0.07, 95% CI [<0.01, 0.13], SE = 0.03, p = .04, and negatively when threat was high, 1 SD above the mean; b = -0.06, 95% CI [-0.12, -0.01], SE = 0.03, p = .03. A t test for the difference between slopes of national identification for low and high threat confirmed that this was a small significant difference, t(321) = 3.06, p = .002, d = 0.17.
Figure 3
The model with felt duty to obey as outcome variable (see Table S11, Supplementary Materials) also had good predictive value, F(12, 1124) = 45.50, p < .001, adjusted R2 = .32. Like for institutional trust, felt duty to obey was positively predicted by EU identification, b = 0.16, 95% CI [0.09, 0.23], SE = 0.04, p < .001, and dual identification, b = 0.11, 95% CI [0.04, 0.18], SE = 0.03, p = .001, and negatively predicted by threat to the nation, b = -0.35, 95% CI [-0.41, -0.28], SE = 0.03, p < .001. In addition, it revealed interaction effects between national identification and threat, b = -0.06, 95% CI [-0.10, -0.01], SE = 0.02, p = .024, and between dual identification and threat, b = 0.11, 95% CI [0.04, 0.17], SE = 0.03, p = .001, indicating that the relations between national and dual identification and duty to obey are different at different levels of threat (see Figure 3). Simple slope analyses showed, similar as for trust, that national identification was positively related to duty to obey when threat was low, 1 SD below the mean; b = 0.08, 95% CI [<0.01, 0.16], SE = 0.04, p = .060, and negatively when threat was high, 1 SD above the mean; b = -0.03, 95% CI [-0.11, 0.04], SE = 0.04, p = .370. These slopes were however not significant, nor was the difference between them, t(321) = 1.94, p = .053, d = 0.11. Simple slope analysis for dual identification showed that it was only significantly positively related to duty to obey when threat was high, b = 0.22, 95% CI [0.13, 0.31], SE = 0.05, p < .001, but not when threat was low, b = 0.01, 95% CI [-0.09, 0.10], SE = 0.05, p = .900. The difference between these slopes was significant but small, t(321) = 2.97, p = .003, d = 0.16.
Additional analyses showed that most of the relations held in a robustness check conducted in the most left-oriented part of the sample.3
Discussion
With this study, we aimed to provide insight into the relation between different levels of national and EU identification and perceived legitimacy of the EU, taking into account the role of experienced threat to the nation’s power and sociocultural identity. Identifying with the EU was positively related to perceived EU legitimacy, which extends previous studies showing that higher EU identification explains higher support for the EU (Chalmers & Dellmuth, 2015; Sindic et al., 2019; Van de Vyver et al., 2018). Interestingly, national identification was also positively related to legitimacy, which may be explained by ingroup projection. High national identifiers are more likely to project ingroup characteristics on the superordinate EU category (Bianchi et al., 2010; Wenzel et al., 2007, 2016), and may consequently perceive the EU as more legitimate because it aligns with national norms and values. This result may at first seem contradictory to prior findings that suggest that national identification leads to less trust in the EU (e.g., Clark & Rohrschneider, 2019; McLaren, 2007). Yet, these studies typically measure national identification as more or less exclusive from EU identification with the Moreno question, while we measured it as weak to strong identification with a continuous scale. Therefore, our results are not contradictory but complementary, indicating that only an exclusive form of national identification but not strong national identification as such undermines the perceived legitimacy of the EU.
However, our findings suggest that a strong national identity does not positively predict perceived legitimacy of the EU when the EU is seen as a threat to the nation. This effect was small but consistent with the idea that national identity is linked to legitimacy through ingroup projection processes, as threat jeopardizes the belief that the EU aligns with national norms and values and serves national interests. This finding may explain why earlier research findings on the relation between national identification and trust in the EU were mixed (e.g., Boomgaarden et al., 2011; Carey, 2002). Our findings about national and EU identification are also interesting in light of research on the perceived legitimacy of international organizations more generally, which has found that legitimacy of these organizations is predicted by a global identification but not by national identification (Dellmuth et al., 2022). National identities may thus be more relevant to legitimacy perceptions of the EU than to those of other, less state-like forms of international collaboration.
Two implications of the suggested ingroup projection processes should be noted. First, ingroup projection dynamics may harm harmonious relations within the EU in several ways. When other groups (e.g., other member states) claim to be part of the superordinate group (e.g., the EU), this can be perceived as an attempt to impose their values on the ingroup, which triggers threat (Gómez et al., 2013). In addition, groups which find themselves highly prototypical of the superordinate group (e.g., Germany) may be less open to diversity, which decreases inclusiveness of groups with diverse values (Wenzel et al., 2016). Second, and related to the latter point, since people from larger and more powerful, or “higher-status”, member states may engage more in ingroup projection, our findings may have been different if people from smaller and less powerful member states would also have been included in the sample. In that case, national identification may be less positively related to legitimacy perceptions of the EU.
The results also provide new insights into the meaning of dual identification and how this form of identification relates to legitimacy. We found that national and EU identification are not mutually exclusive, consistent with prior findings (Boomgaarden et al., 2011). Although the Moreno question showed that most participants with a dual identity identified first with their country and then with the EU, dual identification on the continuous measure was high and positively related to EU legitimacy perceptions. Exclusive national identifiers had the lowest legitimacy perceptions compared to the other dual identification categories, corroborating earlier findings that exclusive national identification leads to more negative attitudes towards the EU (Clark & Rohrschneider, 2019; Hooghe & Marks, 2005; McLaren, 2007). It thus seems that the previous finding that some level of superordinate identity in combination with a high level of subgroup identity is sufficient to promote a sense of dual identity (Simon & Ruhs, 2008) also applies to a superordinate EU identity. When participants experienced threat, high dual identifiers showed a stronger felt duty to obey the EU than low dual identifiers. No relation between dual identification and duty to obey was found when experienced threat was low. Although it should be noted that this effect was small and not found for the trust subscale of legitimacy, it may indicate that dual identification can positively predict perceived legitimacy of the EU even when threat to the nation is experienced. This could be because people whose self-concept includes positively integrated national and EU identities have different expectations from the EU than people with low dual identification. For example, they may be less concerned with interference of the EU with national interests, because they consider these as part of the larger EU interests and expect the EU to act in the interest of the EU as a whole.
Finally, the experience of threat from the EU to national power and sociocultural identity was low, which may be attributed to certain aspects of the used sample, which consisted of relatively highly educated and liberal people, who identified on average stronger with the EU than with their country. Nevertheless, threat to one’s nation was a strong, consistent, and universal (i.e., across counties) predictor of the perceived legitimacy of the EU. Few differences were found between the two types of threat examined. This suggests that power and sociocultural identity are both important aspects of the national identity which, when threatened, trigger defensive responses resulting in lower legitimacy perceptions of the EU. Thus, the nature of the threat experience, which may be driven by local political narratives, may not be as important as the general feeling of threat to national identity that seems to undermine perceived legitimacy in all countries examined here.
Limitations and Future Research
The large sample size with participants from different nationalities allowed us to explore social psychological processes underlying perceived legitimacy of the EU beyond country-level differences. Some limitations should be noted. First, as indicated, our sample was relatively young, well-educated, and oriented towards the political “left”, which may be due to our inclusion criterion of being fluent in English. At the same time, this limitation may strengthen our findings, as this may have provided a relatively conservative test, for the expected relations between threat and EU legitimacy were even found in a fairly EU-supportive sample. Nevertheless, future research should strive for more representative samples, also including citizens from other member states. Especially considering the potentially complex intergroup relations based on the ingroup projection model, it would be interesting to further investigate the role of social identity for perceived EU legitimacy in countries that may be perceived as less prototypical of the EU.
With regard to measurements and operationalizations, our measurement of threat to the nation relied on participants’ self-reports. This approach may have failed to capture more unconscious experiences of threat, and future research is well-advised to extend this study by employing other measures. For example, threat can be elicited with experimental manipulations (Rios et al., 2018) and measured more directly by assessing cardiovascular reactivity (Blascovich et al., 2001; Scheepers & Ellemers, 2005, 2018). In addition, there remains discussion about how to conceptualize and measure perceived legitimacy (Dellmuth & Schlipphak, 2020). In the present study, we operationalized perceived legitimacy as institutional trust and felt duty to obey. Our findings suggest that although these subscales represent different dimensions of perceived EU legitimacy, they are generated by similar processes related to social identity. A notable difference, however, was that there seemed larger variation between countries on levels of institutional trust than on levels of felt duty to obey.
Another limitation is that the cross-sectional nature of the study limits us in interpreting the causality of the relationships. Although we conceptualized social identity and threat as predictors of perceived legitimacy of the EU, legitimacy perceptions may also influence identification and threat experiences (see e.g., Verhaegen et al., 2017). Relatedly, due to the nature of our study, we could not investigate the role of political ideology in this process. Political ideology was so strongly related to threat, in particular to threat to national power, that including it in the regression model led to unstable estimates of the relationship between political ideology and legitimacy. Participants with a more right-oriented political ideology, both regarding economic and social issues, reported high levels of threat to their nation. This relation was so strong that in the overall factor analysis political ideology loaded on the same factor as threat to the nation’s power. Prior research has shown that political orientation does not have a linear but rather a U-shaped effect on attitudes towards the EU, as the left and right ends of the political spectrum have shown greatest Euroscepticism, but with different underlying motives (Hooghe & Marks, 2005; Lubbers & Scheepers, 2010; van Elsas & van der Brug, 2015). Importantly, however, adding political ideology did not remove the main and interaction effects in the regression models, suggesting that relationships between social identification and legitimacy are stable and not influenced by political ideology. It is for future studies with longitudinal and experimental designs to elucidate the relationship between political ideology and social identification in predicting legitimacy, and to define causes of experienced threat.
Finally, considering the large sample size, significant results were obtained for relatively small effects. Although small effects can make a large difference when they are integrated in well-informed interventions at the population level (Cohen et al., 2011), it is important to keep the relatively modest size of our effects and the exploratory nature of our study in mind when interpreting the results. Another issue that cannot be entirely ignored when interpreting some of the other (small) effects reported, is the issue of common method variance, meaning that some of the relationships may be due to their shared psychometric characteristics and response tendencies among participants, as most of the measures involved similar scales.
Conclusion
Safeguarding public perceived legitimacy is key for the effectiveness and sustainability of regional cooperation, such as in the EU. We hope that the current contribution provides a first step for more research on how social identity affects these perceptions. Our findings suggest that people’s subjective experiences of a threat from the EU to their nation pose a risk to the perceived legitimacy of the EU. While a strong national identity in itself does not seem to harm perceived EU legitimacy, it may do so when people experience threat to their national identity. Successful integration of national identities within the superordinate EU identity may protect legitimacy when experienced threat is high. Taken together, it could be important to better recognize and respect national subgroups and their values in order to improve EU legitimacy. Yet, considering the exploratory nature of this study and the small effects, more research is needed to understand these processes and the potential pitfalls of promoting the EU as a superordinate group, and to investigate whether these social psychological processes are EU-specific or general and therefore informative for other models of regional integration across the world. It would also be important to imbue these studies with insights from political science and EU law to further optimize external validity and to inform possible interventions.