The role of outgroups for the formation of
European Identity
University of Amsterdam
Erasmus Mundus Master "Journalism Media and Globalization" Graduate School of Communication
Supervisor: mw. P.H. Penny Sheets Word Count: 9295
Michael Seckler Student-ID: 10583521
Abstract
Over the past years the usefulness of social identity theory (SIT) for research on European Iden-tity has been increasingly acknowledged. This study draws on SIT to explore the role of out-groups for individuals' level of identification with Europe. SIT states that identification with an ingroup increases through the presentation of a salient outgroup. Previous studies furthermore suggest that this effect is especially strong for identities, like the European, which are perceived as intangible and fragile. An experiment including 400 German participants was conducted to test the importance of outgroups for identification with Europe in an empirical manner. Partici-pants received constructed news pieces that either portrayed the EU in clear opposition to Islam, the US or, in case of the control group, did not mention an outgroup at all. Findings show that the conditional nature of self-identification-effects holds true for the case of Europe, as partici-pants were only affected by the manipulation if the ingroup-outgroup-relation was perceived as salient. Furthermore, it was shown that Islam was more effective than the USA in generating self-identification-effects. The paper closes by situating these findings within their political con-text and the fundamental conflict between Europe's conceptual inclusivity and the apparent need of exclusion for the formation of a common identity.
Keywords: Social Identity Theory, Outgroups, European Identity
1. Introduction
"Does the EU require a bogeyman in order to understand itself?" asked 'Der Spiegel' in an article published on May 19th 2014 (Spiegel, 2014, p.115). In the context of the ongoing conflict in Ukraine the article discussed whether Europe finally discovers what it stands for (Spiegel, 2014, p.114). In the face of Putin's nationalist conduct, so the article says, Europe is able to see the values and beliefs that unite them more clearly. Instead of an authoritarian Russia, Europe in the article stands for tolerance as for example expressed through the success of transgender artist 'Conchita Wurst' at the 2014 Eurovision Song Contest. The author finishes with the conclusion
that while the elites have failed to politically unite Europe, the "tv-plebiscite" (Spiegel, 2014, p.116) of the Eurovision Song Contest proves the existence of a sense of unity within the Euro-pean people. Even though the article eventually puts the role of EuroEuro-pean 'others' into perspec-tive, a broader trend holds true: once again, a discussion about the essence of Europe and the EU has been kicked off through the conflict with an outsider. Similar discourses were caused by the conflict about the participation in the United States' War against Iraq in 2003 and the frictions after the publication of the Mohammed cartoons in 2005. While this can be seen as a natural re-sult of identity-dynamics in which meaning generally is constructed through difference, the question arises whether Europe is especially prone to negative identification, that is identifica-tion through differentiaidentifica-tion from others.
In public discourse this question has mostly been answered affirmatively, pointing to Eu-rope's alleged lack of "internal characteristics that can generate a strong sense of self" (Katzen-stein & Checkel, 2009, p.217). The 'Spiegel'-article is a good case in point: After numerous rhetorical questions asking 'what Europe is and what it stands for', the article concludes that it has become "a permanent question", comparing it to "a shy boy with glasses who is being bul-lied by the big guys on the schoolyard" (Spiegel, 2014, p.115/16). However, neither the portray-al of Europe as "uncertain, weak and [...] indeterminate" (Biportray-alasiewicz, 2008, p.71), nor the con-sequential idea of a stronger identification through differentiation is new. What is lacking
though, is an academic investigation that puts the significance of negative identification for Eu-rope to a test. By means of an experimental research design this paper aims to fill this gap and examine the importance of outgroups for European identity in an empirical manner. Specifically, the main question guiding this research is whether and to what degree increased outgroup salience leads to stronger identification with the European ingroup.
While the study principally derives its hypotheses from the logic of Social Identity Theo-ry, there are a number of events in public discourse that suggest a special role of outgroups for European identity, thereby lending an investigation of the issue its political relevance. One could
for example point to the above mentioned case of Jürgen Habermas and Jacques Derrida who saw in the anti-U.S. protests of 2003 the "birth of a European public sphere" and linked the cre-ation of a European identity to the goal of "counterbalancing the hegemonic unilateralism of the United States" (Case, 2009, p.112). Or one could look at more recent events in which Russia, portrayed as an "inherently authoritarian" state (Mozorov, 2012, p.38), helped to establish a sense of a European self. The observation that Europe sharpens its profile in public discourse al-most exclusively against 'others' is highly charged, as such tendencies threaten to undermine EU's ambition of inclusivity and universal values. Case (2009, p.129) explains this by pointing to the "fine line between national chauvinism and the 'building up' of national identities es-poused by Habermas". And Cedermann (2001, p.3) asks whether there is "a trade-off between strengthening European identity and the exclusion of goods, people and states". Put practically, one could ask for example whether resistance against the accession of Turkey to the EU also stems from the attempt to carve out a clearer notion of 'Europeanness'. Such questions are being posed against the backdrop of a broader discussion about the degree to which a postmodern EU is able to overcome the exclusive elements of every identity formation. While a deeper discus-sion of this conflict is beyond the scope of this paper1, it is crucial to note the fundamental
prob-lem: European identity formation entails a dilemma in so far as EU's inclusive and universal am-bitions conflict with the inevitable need for differentiation. As Connolly says: "Identity requires difference in order to be" (Connolly, 1994, p.64). Europe's "postmodern mode of differentiation" (Rumelili, 2004, p.28), however, appears to be at odds with this premise. The findings of this study will have to be contextualized within this broader political context.
As the study is concerned with the functionality of European identity formation, it is nec-essary to organize the vast amount of literature on the matter and separate aspects of identity-re-search which are relevant for the purpose of this paper from the rest. In order to this, the follow-ing section on the theoretical background of the study will be divided into three parts: The first 1 For a detailed discussion see Rumelili (2004)
part will present Social Identity Theory and Self-Categorization-Theory as the guiding concepts for this study. A second part then addresses the particularities of European identity and shows how these complicate the identify-formation process. Finally, a third section provides some background regarding Europe's relation towards the two outgroups that were selected as manip-ulations for this experiment.
2. Theoretical Framework
2.1 Social Identity Theory
According to Kohli (2000, p.114) problems with studying identity begin with the term itself as it is "so exceedingly vague or even vacuous but at the same time seem[s] to capture such impor-tant dimensions of social life". Identity grasps at the very essence of what constitutes an individ-ual or a group. It is for that reason that identity-research has been conducted in numerous aca-demic fields including for example psychology, sociology, psychoanalysis or communication science. Naturally, bringing together such diverse approaches under the same label comes with the danger of harming the significance of the term in research contexts. In fact, Brubaker and Cooper (2000, p.1) argue that even within the social sciences and humanities the word 'identity' has been used imprecisely and sloppily to the point that "the term loses its analytical purchase". In order to carve out a definition of identity that is useful in the context of European outgroups, it is therefore advisable to rely on a clear and operationalizable concept of identity2. As the study
is concerned with group rather than personal identity, it will rely on the prominent Social Identi-ty Theory (SIT) as a guiding concept.
Generally SIT, developed mostly by Henri Tajfel (1974), is well suited for the current re-search as it is in line with the way this experiment aims to measure European identity, which is 2 A detailed differentiation between essentialist and constructivist approaches to identity is beyond the scope of
this paper and, since identity is measured through participants perceived feelings of belonging to Europe, less relevant for the execution of the experiment. For a detailed discussion of the two approaches see (Brubaker & Cooper, 2000).
on the level of "individual identity proper" (Kohli, 2000, p.122). Instead of looking for the sub-stance of identity in the form of, for example, common history, beliefs or values, individuals' feelings of belonging are the yardstick for the 'existence' of a European identity. In this regard the paper follows Ongur, who suggested the usefulness of SIT for research on European identity. He asserts that
"even without sharing a common culture, a common history, or a common set of tradi-tions, values and aspiratradi-tions, Europeans might – discursively or practically – form in-groups which temporarily allow the ability to construct a social identity" (Ongur 2010, p.133).
So what is a social identity? When Henri Tajfel introduced the idea in 1974 he wanted to show how individuals' self-perception is influenced by membership within different groups. He there-fore defined social identity as "that part of an individual's self-concept which derives from its knowledge of his membership of a social group (or groups) together with the emotional signifi-cance attached to that group" (Tajfel, 1974, p.69). Social identity, in this sense, "entails a deper-sonalized sense of self" (Brewer & Gardner, 1996, p.83). Drawing upon Tajfel's definition of so-cial identity, Turner dealt in his 'Self-Categorization-Theory' with "the cognitive underpinnings of social identity" (Ongur, 2010, p.136). Based on the assumption that "people's self-definitions stem from group identification" (Philippou, 2005, p.294), he was concerned with the psycholog-ical construction and functionality of social identities. His "fundamental assumption is that so-cial identity-building starts with categorizing the soso-cial world into ingroup(s) and outgroup(s)" (Ongur, 2010, p.136). As this study looks at European social identity in the face of potential groups, this dimension of social identity is especially relevant. According to SIT, in- and out-groups are defined primarily through a person's knowledge of belonging to it (Ongur, 2010, p.134). In that sense, they are context-depended and can be created spontaneously. Tajfel proved this through the 'minimal group paradigm' where simple allocation of people into two groups was sufficient to elicit ingroup-favoring behavior (Ongur, 2010, p.134). The random separation of a group into two parts, created momentarily an in- and outgroup and subsequently influenced
people's behavior. From experiments like this, Turner concluded that in order to achieve a posi-tive social identity "groups are driven towards making ingroup-favoring comparisons to attain 'positive distinctiveness' from other relevant groups" (Cinnirella, 1997, p.20). However, while in- and outgroup distinctions can principally be created quickly, it is important to note that in or-der for people to strongly identify with an ingroup, factors like the accessibility and salience of the intergroup relation need to be considered (Rutland & Cinnirella, 2000; Voci, 2006). Self-cat-egorization theory states that only if an ingroup-outgroup-distinction is perceived as salient indi-viduals' social identity will be activated and lead people to develop positive attitudes and feel-ings towards the ingroup (Turner, 1982). National social identity can serve as a prime example in this regard. As a highly accessible category with clear inside/outside-demarcations, the impor-tance of a national social identity for people's self-understanding is reflected in public opinion polls as well as identity research (Cinnirella & Rutland, 2000, Eurobarometer 2013). In a study by Rutland & Cinnirella Scottish students answered a 'who-am-I-test' primarily through refer-ence to their national identity. Eurobarometer results show the same pattern: In 2013, 89 percent of all Europeans considered their national identity to be the most important. The extent to which European (social) identity differs from national social identity will be discussed in the following chapter.
First, however, it is necessary to address another aspect of social identities which will be relevant for this study. Turner not only looked at the relation between in- and outgroups, but also addressed the relation between different ingroups and the way different social identities interact. As Kohli (2000, p.115) emphasizes "a person has never just one social identity as he or she oc-cupies different social positions, is engaged in different fields of activity, and is a member of several different groups at the same time". Different social identities are then organized in "a hi-erarchical system" (Philippou, 2005, p.294), which can be imagined in the shape of a pyramid. From the top to the bottom the categories become less inclusive, so that on top one could find a category as abstract as 'human' and on the bottom as narrow as for example 'chess player'. What
is interesting for the purposes of this study is the way national and supranational social identities interact. Cinnirella (1997, p.21) showed in a study with British and Italian students how Euro-pean and national social identities can either coexist next to each other or be seen as mutually exclusive, depending on whether individuals construct them on the same or different levels of abstraction (Cinnirella, 1997, p.21). Therefore this study not only measured national as well as European identity among participants, but will also spend special attention to the nature of the relation between the two identity-levels.
2.2 European (Social) Identity
Next to these general demarcations regarding our definition of identity, there are a number of particularities for the case of European identity, which need to be taken into account.
From the perspective of individual identity proper surveys show that a strong sense of belonging is still lacking among the European people. The 2013 Eurobarometer survey revealed that four out of ten Europeans continue to see themselves as citizens of their nation only (Eurobarometer 2013). Only six out of ten Europeans indicated to feel at least partly as a European citizen, with the numbers decreasing in every country compared to the 2012 survey. Germany, however, has over the last years always been among the countries with highest share of people identifying as European. In the latest survey 73 percent stated to feel either completely or partly European (Eu-robarometer 2013).
When analyzing such numbers it is important to understand the ways in which European identity formation differs from its national counterparts. Rumelili (2004, p.27) summarizes the main dif-ferences between the EU and nation states as follows:
"Identities in the modern nation-state system rest on the construction of clear and unam-biguous inside/outside and self/other distinctions. In contrast, a postmodern collectivity entails 'moving beyond the hard boundaries and centralized sovereignty characteristics of the Westphalian or 'modern' state towards permeable boundaries and layered sovereign-ty".
The central difference between modern nations and a postmodern Europe therefore lies in its scope. While nation states envisioned themselves as finite entities, defined by a certain territory or people, the postmodern agenda of the EU at least partly decouples questions of belonging and territory. That is, the boundaries of the EU are ever-changing and inherently flexible, making distinctions as to who is 'in' or 'out' difficult to define. As Europe and the EU leave modern un-derstandings of sovereignty, identity and territory behind (Bialasiewicz, 2008, p.74), they also complicate the identity-formation process. Finite categories like a specific territory or group of people are substituted with a conceptually universal set of norms and values. As a starting point for a common identity this inclusivity appears problematic insofar as, following Turner, the con-struction of social identity depends on in- & outgroup categorization and therefore exactly the opposite: exclusion.
It is in this context of a completely changed framework of identity formation that the Eu-robarometer-results have to be understood. The EU does not only defy modern inside/outside distinctions, but through its universal scope actively undermines them. To escape this problem intellectuals like Habermas have lobbied for a "constitutional patriotism" (Habermas, 1991) over the last years - that is, a sort of identification with a group based not on territory or ethnicity but on shared values and principles. While patriotism as a modern category refers to positive feel-ings of belonging to the nation state or at least parts of it, Habermas' idea for a European identity can be seen as its supranational and postmodern pendant. However, the Eurobarometer-results show that Europe's value-based identity does not yet trigger the same level of identification as nation states. As the moments highlighted in the introduction have shown, there is an observable peaking in European identification and unity whenever Europe seems to fall back behind its own premises and modern inside/outside demarcations become obvious and a clear outgroup is present. This has led to an ambiguous relation between Europe's conceptual universality and the apparent importance of outgroups for the formation of an European identity.
affect the role of outgroups in the identity formation process. The (perceived) inability to clearly define what Europe is, what it stands for or even where it begins and ends geographically, has led to a tendency of defining Europe negatively, that is in differentiation to an outgroup. As Katzenstein and Checkel (2009, p.224) noted: "However loosely coupled civilizational empires may be internally, they do manage to identify themselves at their fuzzy borders where they en-counter 'the other'". Principally, negative identification is unavoidable and the basic principle not only of identity-formation but any attribution of meaning. On such an abstract level it is obvious that identity "can only be established in relation to what is not, to difference" (Campell, 1992, p.11). Every identity departs from such an observation of difference which is subsequently con-verted into otherness "in order to secure its own self-certainty" (Connolly, 1991, p.64). This dy-namic is also acknowledged by Turner who assumes that self-categorization becomes more salient whenever an ingroup is compared to an outgroup (Rutland & Cinnirella, 2000, p.497). But despite this basic relation in which identity and otherness necessarily consolidate each other, Kohli (2000, p.127) rightfully observes that "everything hinges on the concrete". In this sense it is interesting to validate whether the existence of an 'other' is especially important in the Euro-pean case. Previous studies have suggested two aspects in this regard: Firstly, the two phenome-na - Europe's conceptual fluidity and low levels of identification among European citizens - seem to be related. In a study with Scottish students, Rutland & Cinnirella (2000, p.506) showed how the fact that respondents had "no firm existing notions of what constitutes 'Europeanness'" made Europe as a category of identification less accessible. Based on this observation they, sec-ondly, assume that European identity might be affected more strongly by changes in the inter-group context. They state that "it's feasible that identities that are embryonic and fragile, because their exact meaning or content is both uncertain and open to contestation, are more prone to con-text effects" (Rutland & Cinnirella, 2000, p.506). On the basis of these two observations this study puts forward the following central Hypothesis:
H1: Respondents presented with a story about Europe in clear contrast to an outgroup will be
more likely to self-identify as being European.
Next to this Hypothesis concerning the general importance of outgroups the study aimed to test the status of Islam and the USA in particular. The following part introduces the theoretical con-text on the basis of which the two outgroups were selected.
2.3 Europe and the U.S / Europe and Islam
Conceptually, as has been discussed above, Europe defies the existence of a clear geographical 'other'. In theory, it is not another nation or region that figures as Europe's demarcation point, but its "war-torn past" (Mozorov, 2012, p. 31; Katzenstein & Checkel, 2009). In practice, however we have seen how geographically identifiable 'others' continue to play an important role in pub-lic discourse about European identity. This chapter will contextualize the relation of Europe to two of these European counterparts, the United States and Islam. Despite their different status as a nation state and religion respectively, both entities have figured as prominent counterimages to Europe and therefore been selected as the tested outgroups in this study. Together with Russia and Eastern Europe the selected outgroups are the ones most prominently discussed in the litera-ture as regions of reference to Europe (Kocka, 2005; Katzenstein & Checkel, 2009; Mozorov, 2012) However, while the Eastern European 'other' received most of its attention in the context of political events like the 2004 enlargements or political frictions with Russia, Kocka (2005) emphasizes the historical importance of North America and Islam as more traditional counterim-ages to Europe. He writes that today as much as before these two regions are Europe's central point of reference (Kocka, 2005, p.56). But while both are important, Europe's relation to the two and its mode of differentiation is fundamentally different.
While opposition to America has been present in Europe ever since America was discov-ered (Markovits, 2004, p.67), the relationship to America was always ambiguous. "The enemy from the beginning was not 'America the conquerer', or 'America the imperial republic', but
'America the temptation'" (Markovits, 2004, p.72/73). This principal tension between philo- and anti-Americanism has characterized Europe's relation to America for a long time and holds true until today. The counterimage that America - and especially later, the United States - provided to Europe was never one of sheer opposition, but always entailed an element of admiration. It was arguably exactly this mixture of simultaneous admiration and condemnation, proximity and dis-tance that led to a depiction of America as more than a random 'other', but instead an explicit counterpart to Europe (Diner, 1993, p.31). As a result, Europe developed over the centuries a strong anti-American discourse which was less concerned with what America does than with what it is. Alongside surprisingly persistent dichotomies of depth versus superficiality and indi-viduality versus 'soulless masses', opposition to America established itself as a part of European self-definition. Whether it is the public debate about the free-trade-agreement between the EU and the USA or criticism of America's alleged cultural imperialism: the above mentioned di-chotomies are almost exclusively present when conflicts between Europe and the US are pub-licly discussed. Thus, the principal ambivalence in Europe's relationship with the United States is still in place and was for example reflected through the reactions after the 9/11-attacks. After initial expressions of solidarity ('we are all American now') the support quickly crumbled and turned into active opposition during the anti-war protests of 2003. Today anti-Americanism has again become a "political force affecting all European states" (Katzenstein & Checkel, 2009, p.225) and is by some believed to be stronger than it has been in a long time (Markovits, 2007, p.26). According to Markovits (2007, p.26) one of the particularities of anti-Americanism, as opposed to negative predispositions to other countries, is its equal presence throughout different educational and social classes in Europe. Regarding the predispositions among the tested sample the following sub-Hypothesis is therefore put forward:
H2: Negative predisposition towards the USA will be equally distributed among the different
in-come and education groups of participants.
While Islam has through a growing Muslim population become an important part of Europe, the identity-dynamics between Europe and the Islamic world don't show the same ambivalences as the American case. Instead Muslim countries are often portrayed as being "inherently non-Euro-pean" (Mozorov, 2012, p.38). The opposition towards the Islamic world is described by Kocka (2005, p.56) as "obvious, easy to experience and impossible to deny". As with the U.S., refer-ence to Islam is often used to get a grasp upon the meaning of 'Europeanness'. What is missing in the case of Islam, however, is the flipside of anti-Americanism: a sense of resemblance or even admiration. The unambiguity of the religious differences brings Katzenstein & Checkel (2009, p.224) to the conclusion that "Europe's 'other' has traditionally been located to the East". European discourses about an Muslim 'other' have gained explosiveness over the last decade in the context of radical Islam and its constant presence in the media. The controversy about Euro-pean values following the publication of the Mohammed-cartoons in 2005 is just one example in this regard. As there is no geographical opposition between Europe and Islam, otherness is not directed towards a specific nation or region, but based on "experienced-based 'real communities" (Kohli, 2000, p.128). Technically, Islam can therefore not be considered an outgroup to Europe as there are many European Muslims. Nonetheless, Islam is publicly perceived as more distinct than the US-outgroup. Fuchs et. al. (1995) found in this regard that these more 'visible' groups have become "more important for differentiating between 'us' and 'them' than extraterritorial, imagined communities" (Kohli, 2000, p.128). It is this special status of Islam that makes it the more problematic and politically charged outgroup than the more 'traditional', national US-out-group. Based on these observations and the findings of Fuchs et. al. this third Hypothesis is put forward:
H3: Participants who receive the Islam-manipulation are expected to identify more strongly as
European than respondents in the U.S.-manipulation.
Overall, the presence of USA and Islam as an outgroup to Europe, both historically and more re-cently, ensures that the respective ingroup-outgroup-relation is highly accessible for participants.
3. Methodological Approach
3.1 Research Design
As outlined above the paper explores the existence of a causal relationship between the salience of European outgroups and self-identification as European. For this reason an experimental search design was chosen as "one of the great strengths of experimental research is that the re-searcher can control the environment and manipulate particular variables of causal interest with great precision" (Halpenrin & Heath, 2012, p.167). While this increases the internal validity, ex-ternal validity suffers due to the artificial setting which is "unrelated to real-world experiences and contexts" (Bryman, 2012, p.55). However, in the context of this paper such a trade-off seems acceptable, as functionality of European identity formation lies at its focus. To demon-strate whether the perception of a clear outgroup leads to stronger identification with Europe, the experiment was constructed as follows. In a first step participants were asked to answer a short pre-manipulation questionnaire, consisting of 21 items that entailed demographics as well as op-erationalizations of concepts like media use, political knowledge or predisposition towards out-groups. After this participants were randomly assigned to one of the two manipulation groups or the control group, where they were presented with one of three constructed news articles (see Appendix). The randomized allocation resulted in a distribution of 130 and 131 participants in the respective manipulation groups and 139 participants in the control group. All participants re-ceived an identical, second set of questions after having read through the article. This post-ma-nipulation questionnaire consisted of another 15 items which asked for the level of identification with the EU and Germany as well as ethnicity and religious background of participants.
3.2 Sampling
The sampling was carried out through the external company 'Research Now' that provided a quasi-representative sample of 400 German participants. As the experiment focuses on Germany
as a case study and articles as well as the survey were written in German, the sample was limited to German people. It was guaranteed that the sample mirrors a representative sample in terms of distribution among gender, age and regions in Germany. But because participants are part of a volunteer panel for Research Now, and self-select into the study, they cannot be taken as truly representative of German citizens. The sample validity compared to a traditional lab sample, however, is great increased. Participants received compensation through 'Research Now' upon completion of the survey. Furthermore, the random allocation to the experimental groups guar-anteed an equal distribution among the groups.
3.3 Manipulations
Participants received different manipulations in each of the three groups in the form of a con-structed news piece which was written by the researchers (see Appendix). The goal was to in-crease the "ingroup-outgroup-distinction" (Voci, 2006, p.73) in the two manipulation groups. In contrast the article in the control group did not mention any European outgroup in order to guar-antee a neutral value on the independent variable. As the manipulation articles presented Europe and outgroups as salient, it was necessary to have an article with only the EU salient to have a systematic point of comparison. To establish outgroup-salience as the only difference, the arti-cles were constructed as follows. All three artiarti-cles kicked off with a paragraph about a (hypo-thetical) research on public opinion towards Europe and the EU. The paragraph was priming neither pro- nor anti-European sentiments as it mentioned dissatisfaction with Europe as well as growing feelings of belonging to Europe among citizens. While the control group read only this paragraph, the articles in the manipulation groups continued with a report about 'Europe-wide'-demonstrations against the influence of the USA/Islam on Europe. In order to guarantee that the outgroup itself is causing the change in the DV, and not, for example, political events or actions related to it, the manipulation articles were identical with only the words for 'USA' and 'Islam' differing. The articles drew upon the motif of protest to create the impression of a factual and
growing opposition between Europe and USA/Islam respectively. They therefore included hypo-thetical data about diverging value systems, fictional protesters complaining about the respective outgroup and finally a political commentary by a fictional professor for international relations. Participants were debriefed at the end of the experiment, and told that the articles were con-structed for the purposes of the study.
3.4 Operationalization of the Dependent Variable 'Identification with Europe'
As has been shown above, European identity is complex and differs in several regards from na-tional identity. These circumstances have created a variety of measurements through which Eu-ropean identity can be operationalized for research purposes. In the face of these numerous at-tempts, it has been argued that the concept generally is too complex for accurate measurement (Smith, 1992, p.221). Sinnott, however, answers that measuring identity is by no means "a lost cause". In his study he compares commonly used measurements and gets to the conclusion that European identity is best measured through "identification ratings" (Sinnott, 2006). These mea-sures include ranking on a scale from national to European - thus putting the two forms of iden-tity against one another - as well as general attachment to Europe. Both types of measures were included in the post-questionnaire (see Appendix). Next to this rating, which asks for attachment to different geographical units, the identity-measures drew upon operationalizations of social identity which have previously been tested and shown to be reliable. However, also here re-searchers criticize that social identity has been "conceptualized and operationalized [...] in nu-merous and sometimes quite disparate ways" (Jackson & Smith, 1999, p.120). As Jackson and Smith identify this variation as the reason for inconsistent results in social identity studies, it seems necessary to give the different dimensions inherent in social identity operationalizations a closer look.
Jackson and Smith identify four different dimensions of social identity, which describe different types and levels of identification with a particular ingroup: perceptions of the
inter-group context (1), attraction to the ininter-group (2), interdependency-beliefs or common fate (3) and depersonalization (4). Their argument goes like this: Each of these dimensions of social identity is related differently to aspects like intergroup bias or intragroup cooperation. The fact that they have been used as operationalizations for the same concept is responsible for the inconsistent findings. A person with high attraction to the ingroup is for example less likely to act on behalf of the group than a person who scores high on 'depersonalization' (Swann et. al., 2011, p.919). Swann et. al. (2011) point in this regard to the difference between 'identification' and 'identify fusion' and state that mere identification comes with lower readiness for action. The four dimen-sions of social identity and their respective operationalizations allow to differentiate between different types and degrees to which participants identify with an ingroup and will be considered in the analysis of the results.
While generally the seven items measuring social identity from Cinnirella (1997) were adopted, these measures were complemented with individual questions wherever they did not satisfy the four dimensions of Jackson and Smith (1999) discussed above. For example, in order to have an item for the first dimension, intergroup context, the following statement was added "I feel closer to Europeans than for example Americans or Chinese people", to which participants could indicate their level of agreement on a seven-point scale. Reliability for the seven items from Cinnirella resulted in a cronbach's alpha-value of .88 and a value of .75 for the remaining four items. Cinnirella's items included for example questions like 'to what extent to you feel Eu-ropean?' (M=5.25, SD=1.56) or 'to what extent do you feel strong ties with other European peo-ple?' (M=4.62, SD=1.51) Based on a reliability test, one item, asking whether participants could imagine to live in another European country, was excluded. The seven items of Cinnirella were subsequently combined and averaged to create a single variable for European identity (M=4.71, SD=1.20).
3.5 Controls & Additional Elements
Next to this general structure regarding the independent and dependent variable the experiment included a number of additional items that controlled for political predispositions as well as gen-eral information about respondents. Next to the usual demographics (age, gender, educational background, income) this included measurements of political orientation, predisposition towards different groups (including the tested ones), media use, political knowledge, religion and EU-support. All of these measures were included as they have previously been shown to have a moderating influence on political opinion and identity formation processes (Schuck&deVreese, 2009; Eurobarometer 2013). An important additional measure looked for participants identifica-tion with Germany to control whether high level of naidentifica-tional identificaidentifica-tion inhibits identificaidentifica-tion with Europe. In order to be able to directly compare European and German identity, both were measured by using the same seven-item set of questions by Cinnirella. Reliability tests of the scale showed a cronbach's alpha-value of .91; the items were then averaged to create a single in-dex of German identity (M=5.03, SD=1.32).
4. Results
Looking at the central Hypothesis H1 results show that there is no statistically significant differ-ence between the control and manipulation groups in terms of identification with Europe. Table 1 shows participants' mean level of identification with Europe, taken from the seven items by Cinnirella (2000), across the different experimental groups.
EU-Identification N Mean Std.-Deviation
Manipulation USA 130 4.61 1.24
Manipulation Islam 131 4.76 1.20
Control Group 139 4.76 1.15
Contrary to expectations participants did not identify more strongly as Europeans after having read an article that portrayed the EU in opposition to an outgroup. A one-way analysis of vari-ance (ANOVA) confirmed this (F(2, 397) = .68, p = .508). In contrast, the mean score for the control group is almost similar to the one of participants in the Islam-manipulation and even higher than for participants who received the US-manipulation. The exclusion of participants who failed to answer the manipulation-check variable correctly (n=66) did not cause a signifi-cant change in this tendency. Overall, Hypothesis 1 was falsified, in so far, as participants in the control group did not score significantly lower on the different European identification items than participants in either of the manipulation groups.
However, evidence of the manipulation's effect can be observed once moderating factors are taken into consideration. Tested were moderators like political orientation, anti-immigration-attitudes and negative predispositions, which indicate a higher salience of the perceived ingroup-outgroup-relationship. Among them, these factors showed a moderate correlation as political ori-entation to the right was correlated with both, anti-immigration-attitudes r(392) = .253, p< .01 as well as existing negative predispositions r(390) = .375, p< .01. It was assumed that for partici-pants who, for example, strongly oppose immigration especially the Islam-outgroup would be more accessible. Once immigration-attitudes were taken into account, the picture changed ac-cordingly. A new variable was created out of two questions, on which higher scores on a sev-en-point scale indicated stronger aversion towards immigration (M=4.11, SD=1.56). Based on the distribution on this item, mean values between the two items that were higher than 5 were classified as holding anti-immigration attitudes. According to this classification 29 percent (n=104) of all participants were coded as holding anti-immigration-attitudes. Table 2 shows the ANOVA-results for the manipulation’s effect on EU identification, separately for the two groups.
Anti-immigration attitudes No anti-immigration attitudes EU-Identification N Mean Std.-Deviation N Mean Std.-Deviation Manipulation USA 25 4.77 1.23 93 4.64 1.19 Manipulation Islam 38 4.86 1.33 78 4.69 1.20 Control Group 41 4.34 1.42 89 4.92 .99
As the table shows mean scores in the group with low anti-immigration attitudes are more or less balanced for the experimental groups and the highest in the control group F(2, 257), = 1.95, p= .217. Among participants with anti-immigration-attitudes however, the mean scores show the expected manipulation effect. For this group, there is a tendency to identify stronger as Euro-pean when Europe is presented in opposition to an outgroup F(2, 101) = 1.60, p = .207. While an intergroup-comparison for that group does not satisfy the criteria for statistical significance, the findings suggest that the manipulation's effect depends on pre-existing dispositions. This im-pression is further supported when other moderators relevant to the salience of an outgroup are considered. One of those is political orientation, which was measured using a ten-point scale go-ing from politically left to politically right (M=4.76, SD=1.92). On the basis of this item a new variable was computed in which values from 1-3 were coded as 'left', 4-7 as 'neutral' and 8-10 as 'right'. Table 3 shows the relation between manipulation group and mean identification with Eu-rope for each of these three groups.
Table 2: Mean identification with Europe by manipulation groups for participants with and without anti-immigration attitudes
Politically left Politically neutral Politically right EU-Identification N Mean Std.-Deviation N Mean Std.-Deviation N Mean Std.-Deviation Manip. USA 32 4.72 1.51 90 4.59 1.14 8 4.41 1.37 Manip. Islam 40 4.70 1.30 81 4.74 1.16 10 5.09 1.22 Control Group 38 4.92 .99 87 4.64 1.14 14 4.74 1.63 ANOVA F(2, 107) = .35, p = .702 F(2, 255) = .42, p = .657 F(2, 29) = .48, p = .623
As can be seen from these tables, the further participants self-identified on the political right, the stronger they were affected by the Islam-manipulation. In the 'politically right'-group respon-dents with the Islam-manipulation identified as more European than the other groups. Even though this is merely a tendency, the findings suggest that the more participants self-identified as politically right the stronger was the effect of the Islam manipulation in triggering a self-iden-tification as European. Interestingly, the moderating effect of political orientation held only true for the Islam-manipulation and not for the US-manipulation.
A similar pattern can be found when looking at general predisposition towards the relevant out-groups as a moderator. Just like political orientation, outgroup predisposition was measured by means of a 10-point scale on which participants could indicate how favorable they feel towards a list of several groups. For Islam and USA a new variable was created in which values of 3 or lower were coded as 'negative predisposition'. Assuming that only negative predispositions will cause higher outgroup salience (Rutland & Cinnirella, 2000), all other values were coded as 'no predisposition'. Comparing participants with and without predisposition regarding their affection by the manipulations shows that, again, an existing predisposition is only relevant for the Islam-manipulation. Table 4 shows that while participants without negative predisposition felt the most European in the control group, participants with a negative predisposition towards Islam felt the
Table 3: Mean identification with Europe by manipulation groups and political orientation
most European in the Islam-manipulation.
Negative predisposition towards Muslim
No negative predisposition towards Muslims
EU-Identification N Mean DeviationStd.- N Mean Deviation
Std.-Manipulation USA 42 4.35 1.23 88 4.73 1.22
Manipulation Islam 44 4.57 1.29 87 4.85 1.15
Control Group 40 4.29 1.34 99 4.95 1.01
ANOVA F(2, 123) = .60, p = .550 F(2, 271) = .87, p = .419 Post Hoc Test:
Control Group and Manipulation Islam
p= .551 p= .823
As mentioned, the differences are too small to be statistically significant. However, the observ-able tendencies regarding the Islam-manipulation suggest that if Islam as an outgroup is salient, a presentation in opposition to Europe triggers stronger identification with the European in-group. In this light the assumption that the Islam-manipulation will cause a stronger ingroup identification than the US-manipulation (H3), has to be assessed ambiguously. While generally the numbers show no statistical significant difference between the two outgroups, there is a ten-dency towards Islam being for Germans the more 'effective' counterimage within the European identity formation process.
For the US-manipulation a similar effect was not found. Even among participants with negative predispositions towards Americans, identification with Europe was still higher in the Islam- than in the US-manipulation. While this indicates a lower salience of the USA as an out-group, it was indeed proven that negative predispositions towards Americans are more equally distributed among educational groups than in the case of Muslims (H2). Participants with and without higher education have almost similar predispositions towards Americans, which was
Table 4: Mean identification with Europe by manipulation groups for participants with & without a negative predisposition towards Islam
measured on a 10-point-scale with low values indicating negative predispositions (higher educa-tion: M=6.75, SD=2.02; no higher educaeduca-tion: M= 6.59, SD=2.36). In comparison the means for predispositions towards Muslims differ more clearly (higher education: M=5.36, SD=2.39; no higher education: M= 4.87, SD=2.63). Taken together these findings suggests that while nega-tive predispositions towards Americans are more widespread among societal groups, they are at the same time not as strong as in the case of Muslims.
While so far European identity has been treated as one variable and mean scores were calculated from the seven items suggested by Cinnirella (2000), the discussion of operationaliza-tions above has shown that these items can reflect different dimensions of social identity. A com-parison between the mean scores on all EU-identification items suggests that participants' rela-tion to Europe can be dissected by means of these dimensions. The item aiming at the intergroup context of European social identity, ('I feel closer to fellow Europeans than, say, to people from the USA or China'), achieved the second highest agreement (M=5.29, SD=1.63). In contrast the item which asked for a degree of depersonalization ('When you hear someone who is not Euro-pean criticize the EuroEuro-pean's, to what extent do you feel personally criticized?') achieved the lowest mean score of all items (M=3.67, SD=1.76). The differentiation between these two ques-tions appears crucial, as they allow to separate participants who merely looked favorable at the ingroup from those who actually felt a strong connection to it. An ANOVA-test showed that, if the 'strong feelings of depersonalization'-item is included as a moderator, the level of identifica-tion with Europe differs significantly between manipulaidentifica-tion and control groups F(2, 62) = 3.90, p = .025. Participants who showed signs of depersonalization, that is a stronger shift from their personal to their social identity (Jackson & Smith, 1999), were therefore more strongly affected by the manipulations.
Next to these findings regarding the central hypotheses and the manipulation effects, an-other observation can help to contextualize the role of outgroups within the broader process of European identity formation among Germans. As was discussed before it is crucial to see
whether national and European identity are constructed on mutually exclusive levels of abstrac-tion. For the Germans participating in this study this was not the case as 'mean national identity' and 'mean European identity' showed a moderate correlation r(398) = .36, p < .01. When com-pared among the experimental groups this correlation is only significant for participants in the US-manipulation r(128) = .479, p< .01 and the Islam-manipulation r(129) = .458, p< .01. For participants in the control group there was no statistically significant relation between national and European identity r(137) = .107, n.s. The biggest difference between mean national and Eu-ropean identity can therefore be found among participants in the control group. This indicates that even if the manipulations did not cause a significant increase of identification with Europe, they might have shifted participants' feelings of belonging to a more equal balance between Eu-rope and Germany.
5. Discussion
This study sought to shed light on the importance of outgroup-perception for individuals' level of identification with Europe. Overall, results show that the effect of the manipulations was not straightforward but dependent upon pre-existing opinions towards the outgroup. As participants perceived Islam as the more salient outgroup, manipulation effects were stronger among partici-pants who received the Islam-manipulation.
Before going into a more detailed analysis of the individual hypotheses, a general obser-vation regarding identity dynamics among participants needs to be contextualized. In line with earlier findings (Eurobarometer, 2013) European and national identity seem to be constructed by Germans on different levels of abstraction. If they were not, the two social identity measures would have a shown a negative correlation as was for example the case with British students in a study by Cinnirella (1997). Instead the two levels were positively correlated, suggesting that participants did not perceive their national identity to be threatened by European identity. Quite
contrary, the correlation shows that European identity might actually be enhanced through stronger identification on the national level3. For the following discussion this observation is
crucial in so far, as an exclusive national identity may have inhibited people from identifying with Europe (Cinnirella, 1997) and therefore skewed the manipulation effect.
Despite the absence of exclusive national identities, results show that Hypothesis 1 did not hold true. At large, the assumption that a portrayal of Europe in opposition to an outgroup will lead to increased identification with the European ingroup was not confirmed. Based upon previous studies it was expected that especially the ambiguities and intangibility of European identity would make participants more prone to self-identify with the ingroup, once the relation to an outgroup is presented as salient. While such a relation could not be verified for the whole sample, the inclusion of specific moderators showed that the fundamental mechanism is in place. When certain conditions were met the news articles in the manipulation groups did lead to increased identification with Europe. This differentiation supports the assumption that "context effects on self-categorization are not necessarily straightforward" (Rutland & Cinnirella, 2000, p.495) but dependent upon several factors. Self-categorization Theory predicts that only an out-group that is considered to be distinct on salient dimensions, as for example religion or race, will affect individuals' level of identification (Cinnirella & Rutland, 2000). The results of this study further support this assumption as a manipulation-effect could only be observed when partici-pants' pre-existing opinions suggested a high salience of the outgroup in question.
When looking at which factors determine whether an intergroup context is perceived as salient, the findings suggest that especially two factors are important: negative predispositions towards the outgroup and feelings of depersonalization towards the ingroup. The moderating in-fluence of negative predispositions was proven in so far as the inclusion of general predisposi-tions, political orientation and immigration attitudes resulted in stronger manipulation effects. Particularly where these moderators included a negative perception of the outgroup (like
nega-3 The direction of this causal relation is based on the assumption that national identity usually precedes European
tive predisposition towards outgroup or anti-immigration attitudes) participants were more re-ceptive to self-categorization effects. While the correlation between the three factors showed that all of them play a role in making an outgroup more salient, results show that especially anti-immigration attitudes increased the manipulation effect. This is in line with findings by Rutland & Cinnirella (2000), who emphasized that accessibility of an outgroup alone will not affect self-categorization unless the outgroups are perceived negatively by ingroup-members. For the rea-sons outlined in chapter 2.4 it was reasonable to assume that the EU-Islam/USA-relation was sufficiently accessible to participants. Results show however, that participants were only affect-ed by the manipulations if indeaffect-ed both, accessibility of the outgroup and negative praffect-edisposi- predisposi-tions towards it, were high.
The second factor, depersonalization, is defined by Voci (2006, p.74) as the transition from personal to social identity through which "people perceive and act as group members and not as [...] independent individuals". A comparison of the different dimensions of social identity revealed comparatively low scores on that level of participants' identification with Europe. Fur-thermore it was shown, that once the moderating4 influence of depersonalization was taken into
account, people with strong feelings of depersonalization felt significantly more European in the manipulation groups than in the control group. The fact that participants generally scored higher on the ingroup-attraction-dimension and lower on the depersonalization-dimension seems to be in line with the broader observation that "most people believe in [Europe] without feeling deeply attached to it" (Katzenstein & Checkel, 2009, p.226). This lack of attachment is relevant in our context as Voci (2006, p.86) found that the effect on group processes was "fully mediated by de-personalization". The same effect was found in this study, where strong feelings of depersonal-ization led to a stronger manipulation effect. Low scores on this specific dimension of social identity could therefore be one reason for the weak manipulation effect in the full sample.
4 While Voci (2006) looked at depersonalization as mediator between 'comparative fit' (perceived homogeneity of
a group) and group processes, this study considered depersonalization as a moderator for the relation between intergroup context and identification as European.
Taken together we can therefore conclude the following regarding the validity of H1: While the overall results showed no significant effect on European identification, the inclusion of moderators puts further evidence to the conditional nature of self-categorization. It seems that also in the case of European identity "processes involving social perceptions cannot be thought of as absolute, but have to be considered in specific judgmental contexts" (Voci, 2006, p.87). Despite the ongoing tendency in public discourse to rely on negative identification of Europe, it seems therefore less likely that strong identification with Europe among citizens will arise solely through such differentiation to others.
Next to this Hypothesis regarding the general role and importance of outgroups for Euro-pean identity, H2 and H3 were concerned with the effectiveness of the two selected outgroups specifically. Looking firstly at the Hypothesis regarding the USA we have seen that negative predispositions were indeed more equally distributed among educational groups than was the case with negative predispositions towards Muslims (H2). However, the US-manipulation nev-ertheless proved to be less effective in causing a self-categorization-effect. The fact that the US-manipulation did not trigger stronger ingroup-identification could suggest two things. A first in-terpretation would be that the Europe-USA-relation was simply not accessible enough for partic-ipants to cause a self-categorization effect. If this is the case negative feelings towards the USA might very well be present, but would not be perceived in an EU-USA intergroup context and therefore not affect self-categorization. However, previous studies suggest a second interpreta-tion, which is also in line with Self-categorization Theory. According to the principle of func-tional antagonism "salience of an intergroup context will inhibit the expression of social identi-ties at another level of abstraction" (Rutland&Cinnirella, 2000, p.504). If for example a national intergroup-context is perceived as salient, individuals are less likely to show self-categorization-effects on the level of European identity (Rutland & Cinnirella, 2000, p.515). The US, other than Islam, might have primed some participants to think on the level of nation states rather than Eu-rope, and thereby weakened or 'diverted' the self-categorization-effect. The data shows slight,
however statistically non-significant, evidence for this assumption as 27 percent of participants in the US-manipulation identified as 'only German' compared with only 20 percent in the Islam-manipulation. As the degree to which participants identified themselves through a religious so-cial identity was not measured, it could not be verified whether the Islam-manipulation had a similar effect and increased participants' self-identification as, for example, Christian.
Moving to the Islam-outgroup results have shown that, as H3 assumed, the Islam-manip-ulation was generally more effective than the US-manipIslam-manip-ulation. Again, this holds true especially for subsets of the sample when the moderating effect of predispositions is taken into account. While for the total sample the mean identification with Europe was only marginally higher in the Islam-manipulation than in the US-manipulation, this trend strengthened through the inclu-sion of moderators. When aspects like ideology or negative predispositions toward the out-groups were taken into account, participants showed a stronger self-categorization effect in the Islam-manipulation. Aside from effects on self-categorization the results show that Islam as a negatively connoted outgroup was more salient than the USA. While 31 percent (n=126) of par-ticipants held a negative predisposition towards Muslims, only 9 percent (n=35) had similar feelings towards Americans. This further supports the assumption that "the boundary that citi-zens define between 'us' and 'them' no longer runs between [...] nation states [but] is instead con-structed against the foreign ethnic immigrant communities within Western Europe" (Kohli, 2000, p.128). While this assumption also holds true with regard to national identity, the findings of this study suggest that a supranational European identity might be especially prone to prac-tices of othering that run along non-national categories, such as race or religion.
From a political perspective the salience of Islam as an European outgroup is controver-sial insofar as racial and religious elements push the creation of a clear 'other'; a notion that is at odds with Europe's conceptual self-understanding. Several researchers have shown how Eu-rope's aspired openness is defied in political practice through disparate political and discursive approaches towards the different border regions of the EU (Kuus, 2004; Favell, 2009; Browning
& Joenniemi, 2008). They show how the loss of a fixed territory, gives higher importance to the way borders are constructed and perceived in public discourse (Browning & Joenniemi, 2008). As has been seen, the particularities of European identity complicate the creation of clear inside-outside distinctions, so that Europe "situates its external others [...] not in directly oppositional, but in liminal partly-self/partly-other positions" (Mozorov, 2009, p.32). Kuus (2004) has identi-fied this dynamic for example with regard to Eastern Europe in which the relation between Western Europe and the accession countries is coined by "shades of otherness" (Kuus, 2004, p.479) rather than clear dichotomies. However, with regard to Europe's south-eastern border, re-searchers have shown how these more flexible and fluid identity dynamics harden and the bor-der, both politically and discursively, thickens (Browning & Joenniemi, 2008). Politically, EU's geo-strategy towards the south contributes to "a rather Westphalian conception of a fortress Eu-rope", in which the border is perceived as static and "the outside [...] as largely unreformable" (Browning & Joenniemi, 2008, p. 31). Discursively, the border between in- and outgroup is of-ten times upheld by means of the racial and religious elements of Europe's relation to that re-gion. As Favell (2009, p.189) notes
"The payoff for a Europe, which is uneasy primarily about a neighbouring Islamic South rather than a less familiar Balkan East, is clear. Most Poles, Romanians and Ukrainians are white and Christian; most Turks, Moroccans, Senegalese, or Somalians are not“.
It is within this context that the observed salience of Islam as an European outgroup becomes a politically highly charged topic. The racial and religious elements of the relation between the EU and Islamic countries facilitate the creation of clear oppositions and therefore practices of othering. This leads to the paradoxical situation in which the region, where Europe is the least able to live up to its premises of inclusivity and non-static borders, is simultaneously the region which is most effective in triggering a strong identification with Europe.
6. Conclusion
The central goal of this paper was to assess the role of outgroups for the formation of European identity. Inspired by previous studies and public discourse on the matter it was assumed that people's identification with Europe will rise with increased salience of an outgroup. Drawing on Social Identity Theory an experiment was conducted to put this assumption to a test. The find-ings resulted in two central observations. Firstly, it was shown that also for the case of European identity the effect on ingroup-identification is conditional. Feelings of depersonalization towards the ingroup as well as negative predispositions towards the outgroup were identified as neces-sary preconditions for an outgroup to cause higher identification with Europe. This is in line with findings on national social identity and puts the assumption of an stronger self-identifica-tion effect for the case of Europe into perspective. The intangibility of European identity and its susceptibility for negative identification seem not to override the conditional nature of self-cate-gorization effects.
Secondly, the study looked at the individual status of USA and Islam within European identity formation. Here it was found that Islam was more effective than the US in triggering self-identification with the European ingroup. These findings suggest that Europe, as a suprana-tional entity, tends to find its 'other' mainly in non-nasuprana-tional contexts. This, as the discussion of the case of Islam has shown, comes with interesting implications for the broader discussion about European identity. Islam, through the racial and religious components within the inter-group relation, allows for the creation of inside-outside distinctions that a postmodern Europe aims to leave behind. The observation that Islam seems to become exactly through such distinc-tions a source of European identification is therefore problematic. It points to the general con-flict that identification with Europe seems to happen the easiest, wherever EU's own premises are left behind and a clear outgroup is visible.
With regard to some limitations, the findings of this study have to be assessed carefully. As participants were exclusively German, the results still have to be validated in different
na-tional contexts. Especially the findings for Islam and the US are valid only for the case of Ger-many, due to nation-specific predispositions towards certain outgroups. The findings that con-cern the general role of outgroups for European identity, on the other hand, are believed to be less dependent on the national context. The study provided first empirical insight into the impor-tance of outgroups for self-identification on the European level. Future research is asked to vali-date these findings in greater contexts, paying special attention to the factors conditioning the ef-fect of outgroups on identification with Europe.
References:
Bialasiewicz, L. (2008). The Uncertain State(s) of Europe. European Urban and Regional Studies 15(1), pp. 71-82.
Brewer, M. B., & Gardner, W. L. (1996). Who is this “we”? Levels of collective identity and self representations. Journal of Personality and Social Psychology, 71, pp. 83–93.
Browning, C. and Joenniemi, P. (2008). Geostrategies of the European Neighbourhood Policy. European Journal of International Relations 14(3), pp.519- 551.
Brubaker, R. & Cooper, F. (2000). Beyond 'Identity'. Theory and Society, 29(1), pp.1-47.
Bryman, A. (2012). Social research methods (4th edition). Oxford: Oxford University Press.
Case, Holly. (2009). Being European: East and West, In: P. J. Katzenstein & J. T. Checkel (eds.). European Identity (pp. 111-131). Cambridge: Cambridge University Press.
Cederman, L. E. (Ed.). (2001). Political boundaries and identity trade-offs, In: Cederman L. E. (ed). Constructing Europe's identity: the external dimension (pp.1-34). Colorado: Lynne Rienner Publishers.
Cinnirella, M. (1997). Towards a European Identity? Interactions between the national and European social identities manifested by university students in Britain and Italy. The British Psychological Society, 36, pp. 19-31.
Connolly, W. E. (1991). Identity-Difference. Democratic negotiations of political paradox. Minneapolis: University of Minnesota Press.
Diner D. (2003). Feindbild Amerika. Über die Beständigkeit eines Ressentiments. Berlin: Propyläen.
European Commission. (2013). Eurobarometer report fall 2013, pp. 1-73.
Favewell, A. (2009). Immigration, migration and free movement in the making of Europe, In: P. J. Katzenstein & J. T. Checkel (eds.). European Identity (pp. 167-189). Cambridge: Cambridge University Press.
Fuchs, D., Gerhards, J. & Roller, E. (1995). ’Nationalism versus eurocentrism? The construction of collective identities in Western Europe, In: M. Martiniello (ed.). Migration, Citizenship and Ethno-national Identities in the European Union (pp. 165-78). Aldershot: Avebury.
Habermas, J. (1991). Staatsbürgerschaft und nationale Identität: Überlegungen zur europäischen Zukunft. St. Gallen: Erker.
Halperin, Sandra & Oliver Heath. (2012). Political Research. Methods and Practical Skills. Oxford: Oxford University Press.
Jackson, Jay W. & Eliot R. Smith. (1999). Conceptualizing Social Identity: A new framework and evidence for the impact of different dimensions. Personality and Social Psychology Bulletin 25 (1), pp. 120-135.
Katzenstein, P. J. & J. T. Checkel. (2009). Conclusion – European Identity in Context, In: P. J. Katzenstein & J. T. Checkel (eds.). European Identity (pp. 213- 227). Cambridge: Cambridge University Press.
Kocka, J. (2005). Verflechtung und Differenz. Zum Problem der Grenzen Europas, In: T. Meyer, C. Winkler & S. Donig (eds.). Europäische Identitäten - Eine europäische Identität? (pp.52-58). Baden-Baden: Nomos.
Kohli, M. (2000). The Battlegrounds of European Identity. European Societes, 2 (2), pp. 113-137.
Kuus, M. (2004). Europe’s Eastern Expansion and the reinscription of otherness in East Central Europe. Progress in Human Geography 28(4), pp. 472- 489.
Markovits, A. S. (2004). Amerika dich hasst sich’s besser. Antiamerikanismus und Antisemitismus in Europa. Hamburg: Konkret Literatur.
Markovits, A. S. (2007). The Americanism Mindset, In: B. O’Connor (ed.):
Anti-Americanism. History, Causes, Themes. (Volume 1: Causes and Sources (pp.23-40). Oxford: Praeger.
Morozov, V., & Rumelili, B. (2012). The external constitution of European identity: Russia and Turkey as Europe-makers. Cooperation and Conflict, 47(1), pp. 28-48.
Ongur, H. O. (2010). Towards a social identity for Europe? A social psychological approach to European identity studies. Review of European Studies, 2(2), pp. 133-143.
Philippou, S. (2005). Constructing national and European identities: the case of Greek‐Cypriot pupils. Educational Studies, 31(3), pp. 293-315.
Rumelili, B. (2004). Constructing identity and relating to difference: understanding the EU's mode of differentiation. Review of International Studies, 30(01), pp. 27-47.
Rutland, A., & Cinnirella, M. (2000). Context effects on Scottish national and European self‐ categorization: The importance of category accessibility, fragility and relations. British Journal of Social Psychology, 39(4), pp. 495-519.
Schuck, A. R., & de Vreese, C. H. (2009). Reversed Mobilization in Referendum Campaigns How Positive News Framing Can Mobilize the Skeptics. The International Journal of Press/Politics, 14(1), pp. 40-66.
Sinnott, R. (2006). An evaluation of the measurement of national, subnational and supranational identity in crossnational surveys. International Journal of Public Opinion Research, 18(2), pp. 211-223.
Smith, A. D. (1992). National identity and the idea of European unity. International Affairs, 68, pp. 55–76.
Swann, W. B., Gomez, À., Brooks, M. L., Buhrmester, M. D., Vázquez, A. & Jetten, J. (2011). On the nature of identity fusion: Insights into the construct and new measures. Journal of Personality and Social Psychology 100 (5), pp. 918-933.
Tajfel, H. (1974). Social identity and intergroup behaviour. Social Science Information 13 (65), pp. 65-93.
Turner, J. C. (1978). Social comparison, similarity and ingroup favoritism. In: H. Tajfel (ed.). Differentiation between social groups (pp. 235-250). London: Academic Press, .
Turner, J. C. (1982). Towards a cognitive redefinition of the social group. In H. Tajfel (ed.). Social identity and intergroup relations (pp. 15–40). Cambridge: Cambridge University Press.
Voci, A. (2006). Relevance of social categories, depersonalization and group processes: Two field tests of self-categorization theory. European Journal of Social Psychology, 36(1), pp. 73-90.
Weldes, J., Laffey, M., Gusterso, H. & Duvall R. (1999). Introduction: Constructing insecurity, In: J. Weldes,M. Laffey, H. Gusterso, & R. Duvall (eds.). Cultures of Insecurity (pp. 1-33). Min-neapolis: University of Minnesota Press.
Appendix: 1) Pre-questionnaire:
Q1) Please indicate your age. Q2) What is your nationality? Q3) What is your gender?
Female Male
Q4) Please indicate your highest level of education completed.
Primary School
High School or equivalent
BA or three years vocational training or equivalent Master's or postgraduate training
Q5) What is your average monthly household income?
< 1150€ 1150€ - 1800€ 1800€ - 2600€ >2600€
Q6) Did you vote in the European Parliamentary Elections that were held in May 2014?
Yes No
Q7) (Only if Q6=no): Why didn't you vote?
I didn't have time
I didn't know about the election I don't care about the election I am against voting
