Download - How European is European Identification? Comparing Continental Identification in Europe and Beyond

How European is European Identification? ComparingContinental Identification in Europe and Beyond

JOCHEN ROOSEFree University Berlin

AbstractEuropean identification has previously been explained by the selective gains brought by theEuropean integration process, by personal transnational experiences and by the influence ofpolitical programmes aiming at increasing levels of identification. All these explanations imply thatidentification with one’s continent would be specific in extent and distribution across the socialstructure in Europe compared to other continents. These assumptions are tested using InternationalSocial Survey Programme (ISSP) and a longitudinal analysis using Eurobarometer. Results showthat, first, the extent of continental identification in Europe is not higher than in other continents.Second, they reveal that there has been no increase in European identification in recent decades.And third, the socio-structural distribution of continental identification is similar on all continents.Accordingly, explaining European identification with respect to policy output of the EuropeanUnion is questioned by the findings. The results rather hint at the possibility that Europeanidentification is independent of political integration.

Introduction

European identification, understood as closeness felt to Europe, is, by its very nameand essence, European.1 Accordingly, to ask for the Europeanness of European identifi-cation seems to make no sense. However, if we look at this topic from a different angle weare faced with an interesting research question: is identification with one’s continent inEurope, in extent and socio-structural composition, different to that within other conti-nents? The European Union (EU) with its unique regional integration process can beexpected to exert considerable influence on affective bonds to the European continent.Therefore, we would expect a European particularity of continental identification in aglobal comparison. I want to discuss whether this plausible assumption can be empiricallysubstantiated.

The assumption that European integration influences continental identification issignificant for theory building and policy-making alike. In the context of public supportfor European integration, the emotional side of identification is hotly debated (see,for example, Bruter, 2006; Checkel and Katzenstein, 2009; Herrmann et al., 2004;Karolewski and Kaina, 2006; McLaren, 2007; Risse, 2010; Stråth, 2000; Theiler 2012).As identification with the political community is an integral part of support for a politicalsystem (Easton, 1975), it has gained importance in research. However, this research has

1 ‘European identification’ is a specific concept in the huge discussion on European identity (see below). It refers only to thepersonal feeling of attachment to a unit. Other strands in the debate on European identity refer to membership in the socialunit Europe or particularities of Europe in comparison to other regions in the world (Brubaker and Cooper, 2000).

JCMS 2013 Volume 51. Number 2. pp. 281–297 DOI: 10.1111/jcms.12005

bs_bs_banner

© 2012 The Author(s) JCMS: Journal of Common Market Studies © 2012 Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 MainStreet, Malden, MA 02148, USA

focused solely on Europeans. What we are lacking, by and large, is a comparison andanalysis of identification with one’s continent which reaches beyond Europe. Only acomparison with other continents allows us to determine whether the patterns found inEurope are EU-specific, or whether we find levels and socio-demographic patterns ofidentification that are similar throughout the world.

First, theoretical arguments on influences on European identification are summarized.These dominantly, though not undisputedly, suggest that we should expect strongercontinental identification in Europe and a specific pattern of socio-demographic influ-ences on this identification. After briefly introducing the data sources, the first stage of theanalysis describes the distribution of identification with one’s continent. The second steprefers to the temporal development of European identification. Thereby, we can determinewhether the current level of identification, as indicated by the cross-sectional analysis, isthe result of a ‘catch-up’ process, a recent decrease or whether it reflects a stable situation.The final step of analysis compares the patterns of influence on continental identification.Hence, we can discover whether the arguments linking European identification with EUpolicies are plausible, or whether we find the same influences in other parts of the world.

I. Why People Identify with Europe – in Theory

Identity and identification in respect to Europe and the EU are contested concepts(for example, Kohli, 2000, p. 2). With reference to the social identity theory of Tajfel(1978, p. 63), I understand identification as ‘that part of the individual’s self conceptwhich derives from his knowledge of his membership of a social group (or groups)together with the value and emotional significance attached to that membership’. Accord-ingly, I deliberately refrain from discussing whether there is a European identity in asubstantial way and what it would or should be. Rather my interest refers to what peoplethink, whether they feel attached to something like Europe for whatever reasons. Asthis identification with Europe has proved to be influential for attitudes on the politicalintegration of Europe, trust in EU institutions or attitudes towards minorities (for example,Bruter, 2004, 2006; Citrin and Sides, 2004; Marks and Hooghe, 2003; Roose, 2007), it isa research object that deserves our attention.2

To explain the extent of European identification, three main lines of argument can befound. The first approach refers to personal experiences in relation to the EU and meetingother Europeans. The second approach is based on the idea that people who gain from theEU tend to identify with Europe. Finally, there is the idea of European identification as theresult of intentional activities by elites.

Identification by Personal Experience

According to the first approach, personal experiences are crucial for European identifi-cation. The argument is made in two variants. Herrmann and Brewer (2004) refer to theimportance of institutions. According to their socialization model, ‘individuals come

2 It is disputed whether a distinction has to be made between identification with Europe and identification with the EuropeanUnion. While logically the difference is more than obvious, people in their thinking seem to be less sophisticated. Bruter(2006, p. 119) can show that people differentiate between a ‘civic’ and a ‘cultural’ identification. However, both conceptsare highly correlated with each other and both are highly correlated with a general identification referring broadly to Europe.

282 Jochen Roose

© 2012 The Author(s) JCMS: Journal of Common Market Studies © 2012 Blackwell Publishing Ltd

to identify with an institution (and the group that it represents) to the extent that theinstitution is salient in their personal lives’ (Herrmann and Brewer, 2004, p. 14). A secondperspective refers to personal interaction. According to this interpretation, frequent ex-periences in a transnational space lead people to hold a European identity. This is arguedfor border regions (Kuhn, 2012a; Roose, 2010, pp. 205ff.), people working within Euro-pean institutions (Wodak, 2004) or more generally for people who are mobile in the EU(Fligstein, 2008; Kuhn, 2012b; Mau et al., 2008). The latter are typically found in thehigher social strata, the more educated with broader language abilities and better profes-sional positions. Quantitative studies can show this relation between education, profes-sional position and European identification (Citrin and Sides, 2004; Dubé and MagniBerton, 2009; Fligstein, 2008, pp. 138ff.). As policies of the EU facilitate mobility withinthe Union and cross-border exchange at the inner EU borders, according to this approachwe should expect higher levels of identification with one’s continent in Europe only, whilesimilar influences to this extent are by and large absent in other continents.3

Identification by Personal Gain

The socialization model by Herrmann and Brewer, as well as other authors arguing forthe significance of interpersonal exchange for European identification, often do notqualify for the kind of experience people have with European institutions or other Euro-peans. However, Favell (2009) describes the ambivalent experiences in different contextsof migration. Not all instances of personal exchange with European foreigners are ex-perienced as pleasurable. The second approach takes this into account, arguing that peoplewho gain from the European integration process tend to feel more European. Already inneo-institutionalism we can find the close link between advantages and mutual loyalty oridentification (Risse, 2005). Fligstein (2008, p. 145; see also, for example, Immerfall andSobisch, 1997, p. 34) interprets the findings of more European identification among theupper social strata as an effect of the improved opportunities resulting from Europeanintegration: ‘[T]he most privileged socioeconomic groups are the most European. [. . .] Allthese groups have opportunities to interact with people from other European countries’.Castano (2004, p. 40, emphasis added) regards this argument as too obvious whenhe writes that ‘few would maintain that self-identification at the collective level is theconsequence solely of perceived material benefits’.

If we accept these theoretical arguments, again we should expect a crucial influence ofEU policies on European identification. However, the resulting level of identification isnot clear. While those who benefit from EU policies should show stronger identificationwith Europe, those who are or feel threatened might be alienated and their Europeanidentification is weakened. Overall these two processes may level out each other, or oneeffect is stronger but we cannot anticipate which of them. However, in respect to thesocio-economic distribution of European identification the arguments are clear. We shouldfind higher levels of European identification among the upper social strata. Again, as thiseffect is directly related to the EU and its policies, the described mechanisms should onlywork in the EU, and on other continents we should find continental identification beingunrelated to, or at least to a lesser extent related to, the respective social strata.

3 Other regional systems like ASEAN or NAFTA do not facilitate cross-border encounters to the extent that the EU does.

How European is European identification? 283


Identification by Elite Construction

A third approach focuses on measures of identity building by elites. Herrmann and Brewer(2004, p. 15) call it the ‘persuasion model’. National elites, who support the Europeanintegration project, and even more so the EU institutions, have been active as Europeanidentity proponents. The establishment of European symbols and narrations is motivatedby the idea of enhancing European identification (for example, Cinnirella, 1996, p. 270).‘The EU from the outset has been engaged in “identity building” and in fostering identitychange’ (Laffan, 2004, p. 84; see also Stråth, 2000).

From the processes of nation-state formation, we have learned how important symbolsand collective narrations are to identity formation (Hobsbawm, 1991; Smith, 1991; Tilly,1975). As Gellner (1968, p. 168) asserts: ‘Nationalism is not the awakening of nations toself-consciousness; it invents nations where they do not exist’. This does not mean that thereference points of shared symbols, beliefs and historical events never existed; rather ithighlights that historical (and other) facts on their own do not create a community. Theyhave to be interpreted and integrated into a narration, which of course neglects and forgetsother facts, therefore resulting in the ‘invention of tradition’ (Hobsbawm, 1983).

Some of the initiatives taken by the younger nation-states have been copied by theEU. For example, the EU has its own anthem in Beethoven’s setting of ‘Ode to Joy’,though usually played without the text. The 9th of May is Europe Day, commemoratingRobert Schuman’s presentation of his plan in 1950 to form a European Community ofCoal and Steel which was established soon after. In many European cities there are eventsand celebrations on this occasion. The European flag is the most widely known EUsymbol (although it was originally the symbol of the Council of Europe). The EU motto‘Unity in diversity’ belongs also to this category of collective symbols and narrations.Even the European security and defence policy can be regarded as an attempt at Europeanidentity formation (Anderson and Seitz, 2006). Writing a European history – in a similarvein to how nation-states have done and still do – has proved to be quite difficult(Leggewie, 2008).

Although the attempts by the EU at forming a common consciousness are by no meansas intense and as successful as the similar activities of nation-states in the 19th century,Europeans are still subject to these activities. Again, this should result in a higher level ofcontinental identification in Europe compared to other continents.

European Identification as a Special Continental Identification

Influences on European identification, as assumed in the literature, are in many casesclosely related to the EU and its policies. Identification by personal experience andidentification by elite construction both lead to the expectation that continental identifi-cation is higher in the EU than on other continents. In respect to the assumption ofidentification by personal gains, the expectation is less straightforward. We may either finda higher or lower level of continental identification in the EU compared to countries onother continents. Still, the pattern should be distinct as this argument also assumes aninfluence of EU policies on the extent of European identification.

Looking at the distribution of European identification across social strata, the argumentof identification by elite construction is neutral, but the two other arguments have a clearassumption: continental identification should be stronger among the upper social strata as

284 Jochen Roose


they profit more from European integration and are more frequently involved in Europeanpersonal experiences. That this assumption is supported by survey data from EU countrieshas been shown time and again (for example, Bruter, 2006; Citrin and Sides, 2004, p. 173;Fligstein, 2008). However, the arguments also lead to the expectation that this pattern isspecific for EU countries.

The assumptions of a specifically EU-level and socio-structural pattern of continentalidentification need a qualification in spatial terms. Socialization processes, as they under-lie the mechanisms of identification by personal experience and by elite construction, maytake some time. Accordingly, we may find differences not only between EU members andnon-members but also between old members and new members of the EU. In respect toeconomic gains, but also opportunities for personal experiences, the situation is slightlymore complicated. The European Free Trade Agreement (EFTA) integrates some Euro-pean states de facto into the internal market without becoming members of the EU. Thisapplies for Norway, Iceland, Switzerland and Liechtenstein. These countries also partici-pate in EU exchange programmes. Depending on which of the described mechanisms ismost influential, these European non-EU countries might be subject to similar influenceson continental identification. The test of a special constellation of continental identifica-tion has to take the constellations, relevance of old EU membership and EU membershipas such or EFTA membership into account.

II. Data

Research on European identification has concentrated on the Eurobarometer studies as adata source. The Eurobarometer is a survey regularly carried out by the European Com-mission in all Member States (and for some time spans also in prospective MemberStates). In each country, roughly 1,000 interviews are carried out (with more in the UnitedKingdom and Germany, and fewer in Luxembourg and Malta). A number of questionshave been repeatedly asked, including those regarding European identification. Asthe wording of the questions has changed, a continuous trend is not perfectly available.However, some very insightful comparisons over time are possible. The obvious short-coming of the Eurobarometer for our research question is its restriction to EU MemberStates and prospective Member States.

Another data source includes a question on identification with one’s continent: theInternational Social Survey Programme (ISSP). The ISSP conducts surveys in manycountries in the world. The study in 2003, with a focus on national identity, brings togetherrespondents from 30 countries.4 The country selection is dominated by European coun-tries and cannot claim to cover the globe adequately. However, to test general hypotheseswe do not necessarily need a representative sample on the country level. Rather, if theassumptions outlined above are valid in general, they also should apply to the countries inthe ISSP study.

Measuring identification is difficult. The ISSP offers only one question – ‘How closedo you feel to [continent]?’ – with the response options of ‘very close’, ‘close’, ‘not veryclose’ and ‘not at all close’. In other studies based on the Eurobarometer the question

4 Due to unclear identification of the respective continent or due to missing data, the following countries had to be excludedfrom the analysis: Venezuela, New Zealand, South Africa, Israel, Australia and the Philippines. Germany is analyzedseparately for east and west.



‘How attached do you feel to Europe?’ is used to measure an affective-evaluative dimen-sion of identification (for example, Fuchs et al., 2009). At least for the EU membercountries, a macro-comparison between this established question wording and the oneused in the ISSP is possible. On a macro-level, the correlation between ‘very close’ and‘very attached’ is r = 0.83. The percentages for the upper two values (‘very close’ plus‘close’, and ‘very attached’ plus ‘attached’, respectively) correlate with r = 0.59. Theseempirical results and the similar question wordings support the assumption that also theISSP wording measures the affective-evaluative dimension of identification adequately.

Another problem in international comparisons is that of measurement equivalence(Van Deth, 1998). One could expect that identification with Europe measures somethingdifferent from identification with other continents because of the political integrationprocess. It is quite likely that a considerable portion of the respondents associate the EUwith Europe and identification is therefore influenced by attitudes on the EU. We wouldnot expect such influences on other continents. However, this situation does not constitutea problem for this analysis. First, Bruter (2006) indicates that in the European countriesthat he looked at, identification with Europe is correlated with a cultural identificationunrelated to the EU. This general identification targeted in the ISSP questionnaire shouldbe – at least in this dimension – similar on all continents. Second, for the dimension whichBruter calls ‘civic’, the difference remains. However, it is exactly this difference that thestudy is targeting. This analysis aims to find out whether the results in European countriesdiffer from countries on other continents. Accordingly, one could also regard it as a test ofexternal equivalence (Van Deth, 1998) for the measurement of continental identification.

Gender and age are measured in a straightforward manner. Professional status isoperationalized by income and professional prestige. Income is measured as the log ofincome z-standardized for each country, thereby reflecting the respective income structureof each country. Occupational prestige is measured by the prestige values according toTreiman (Ganzeboom and Treiman, 1996). Education has been measured as the years offull-time schooling including university, but not vocational training.5

III. Continental Identification in Comparison

In Table 1 we find identification with the respective continent per country. For a betteroverview, unweighted averages for groups of countries are represented.6 The level ofidentification with Europe varies considerably among the old EU Member States. Austriais at the top of the range with 30 per cent who feel ‘very close’ to Europe, and GreatBritain is at the opposite end with only 5 per cent feeling ‘very close’ to Europe. Takingthe first two categories together (‘very close’ and ‘close’), Portugal scores even higherthan Austria, while Britain remains at the end. The mean of the countries belonging to thegroup of old EU-15 Member States for feeling ‘very close’ is 16 per cent and for feeling‘very close’ or ‘close’ 54 per cent.

Six countries belong to the eastern accession group of 2004. At the time whenthe survey was carried out they were not yet members of the EU, but the accessiondate was already fixed and, prior to accession, considerable financial transfers took place.

5 For convenience, occupational prestige and age have been divided by 100; years of education was divided by 10.6 It is important to note that these averages do not represent the level of identification for the respective groups as thecompositions of countries are not representative and results are not weighted for population size.

286 Jochen Roose


The results for these countries cover an even wider range from 64 per cent (Hungary) to3 per cent (Latvia) feeling ‘very close’ to Europe. The average of these countries forfeeling ‘very close’ is a considerable 24 per cent, which is even higher than the identifi-cation among respondents from the old Member States.

Table 1: Feeling Close to Continent (%)

Country Veryclose

Close Not veryclose

Not atall close

N

Austria 29.5 41.5 25.4 3.6 992Denmark 18.2 35.8 34.9 11.2 1,255Finland 7.4 32.2 47.1 13.2 1,211France 20.4 34.0 29.1 16.5 1,522Germany (east) 10.2 46.9 36.9 6.0 401Germany (west) 13.3 47.9 33.1 5.6 803Great Britain 4.5 22.7 41.2 31.6 813Ireland 10.8 33.7 38.6 16.9 1,046Netherlands 14.2 32.8 35.3 17.7 1,694Portugal 26.0 41.9 25.7 6.4 1,542Spain 17.3 60.6 17.2 4.9 1,192Sweden 14.2 34.5 41.8 9.4 1,103EU-15* 15.5 38.7 33.9 11.9Czech Republic 20.3 51.4 22.0 6.2 1,204Hungary 63.6 30.2 5.2 1.1 1,007Latvia 3.4 16.2 41.6 38.8 925Poland 19.5 43.5 32.3 4.7 1,199Slovak Republic 17.8 51.7 21.2 9.3 1,047Slovenia 21.2 45.5 23.6 9.7 1,057EU-NMS-10 (2004)* 24.3 39.8 24.3 11.6Bulgaria 32.4 40.1 17.7 9.8 950EU-NMS-2 (2007)* 32.4 40.1 17.7 9.8Norway 19.7 40.2 33.0 7.1 1,322Switzerland 20.7 58.3 18.9 2.1 1,031EFTA (not EU)* 20.3 49.3 26.5 4.6Russia (European part) 3.1 9.6 33.4 54.0 2,039Europe (not EFTA)* 3.1 9.6 33.4 54.0Canada 19.2 44.0 27.2 9.6 1,031United States 27.1 38.4 27.5 7.0 1,165North America* 23.1 41.2 27.4 8.3Chile 36.0 36.5 21.6 6.0 1,467Uruguay 22.3 32.2 34.5 11.0 1,089Venezuela 30.0 28.3 23.5 18.2 1,162South America* 29.4 32.3 26.5 11.7Japan 16.1 46.5 28.2 9.2 901South Korea 6.0 24.4 45.6 24.0 1,286Taiwan 1.3 10.1 34.5 54.2 1,827Asia* 7.8 27.0 36.1 29.1South Africa 38.7 31.4 19.2 10.7 2,357Africa* 38.7 31.4 19.2 10.7

Source: ISSP 2003 and author’s own calculations.Note: * All averages are unweighted means which do not represent valid informa-tion for the group of countries.



In the two EFTA countries which are not members of the EU, but are part of theISSP data set (that is, Norway and Switzerland), the average level of closeness felt toEurope is not lower than the level among the EU-15 countries. Every fifth person in thesecountries feels ‘very close’ to Europe and another half of the population feels ‘close’ toEurope.

Feeling ‘close’ to Europe is considerably less common in the European part ofRussia – the only available European country that is part of neither EFTA nor the EU.Only 3 per cent feel ‘very close’ to Europe and only another 10 per cent feel ‘close’ toEurope. From these results we could conclude that economic advantages provided bythe European free market (including the EFTA countries) and the identity-building activi-ties of the EU have an effect. The Russians and their low identification with Europe arethe crucial case in point. However, as soon as we leave Europe, the picture changescompletely.

The second highest level of continental identification is found in South Africa, where39 per cent feel ‘very close’ to Africa and another 31 per cent feel ‘close’. Also, in SouthAmerica identification with the continent is very high. In Chile, 36 per cent feel ‘veryclose’ to the continent and another 37 per cent feel ‘close’. The numbers for Venezuela area bit lower, but the share of people feeling ‘very close’ there is still higher than in anyEU-15 Member State. In the EU, only Hungary, with its exceptionally high rate of 64 percent of respondents feeling ‘very close’ to Europe, shows stronger identification with itsrespective continent.

It is also instructive to take a look at those countries where closeness to the continentis seldom felt. At the very bottom of the scale we find the Taiwanese and the Russians. Inboth countries more than half of the population feels ‘not at all close’ to its respectivecontinent. Other countries in which more than a fifth of respondents do ‘not feel at allclose’ to his or her continent are South Korea and some EU members: 32 per cent of theBritish and 39 per cent of Latvians say that they do ‘not at all feel close’ to Europe.

The differences between countries call for more specific explanations than just mem-bership in the EU or the EFTA. Controversies on EU policies in general (for example, inthe case of Great Britain) or in the context of the accession process (for example, inLatvia) come to mind. The lower level of continental identification with Europe among theolder Member States might be a result of controversial EU politics. However, there areother countries with similar characteristics in respect to EU policies that show differentresults. The theoretical arguments led us to expect a specific level of identification amongEU or EFTA members. Identification by personal experience and by elite constructionwould imply a higher level of identification, which cannot be found. Identification bypersonal gains was undecided in respect to a higher or lower level, but still would haveimplied a typical EU or EFTA level. The descriptive results for the countries available donot support these assumptions.

IV. European Identification over Time

There is a possible way to explain why the EU might have an (increasing) effect onidentification with Europe and yet we still might not see the expected differences in aglobal comparison. We could assume that the EU already has increased European iden-tification among those people living in Europe and that their level of identification as

288 Jochen Roose


found in the 2003 survey is already the result of this process. This would be plausibleif the continental identification in Europe was low in a global comparison in earlierdecades. In fact, such an assumption might be reasonable when we take into accountthe bloody wars in Europe in the first half of the 20th century and even in the centuriesbefore.

We do not have data on continental identification for the years when the forerunnerof the EU – the European Coal and Steel Community – was founded in the early 1950s.However, one should keep in mind that neither the economic advantages of a free marketnor the activities of the EU in furthering European identification had become significantfor the ordinary population at that time. The single market project gained momentum withthe Cassis de Dijon ruling by the European Court of Justice in 1979 and the intensifiedactivity of the European Commission following the White Paper on the internal market in1985 (see, for example, Peterson and Bomberg, 1999, pp. 60ff.). Exchange programmessuch as ERASMUS or INTERREG were initiated in the 1990s. The European anthem, theEurope Day and the European flag were chosen in 1985 and 1986, with the motto onlyfollowing in the year 2000. Likewise, other activities that could be regarded as part of thisidentity formation by EU institutions mostly began in the 1990s and after. So the relevanttime span in which the EU’s activities should have had the greatest influence on identi-fication – according to theories – is in the 1980s at the earliest, but mainly we should focuson the ensuing two decades.

From the Eurobarometer we have data on the evolution of European identification. Inthe 1990s a question asked the respondent to weigh his or her belonging to the respectivenation-state against belonging to Europe. Asking how a person will see themselves in thenear future, the respondents could choose either their nationality, Europe or both incombination with either mentioning Europe or nationality first.7 Figure 1 shows thedevelopment over time of answers to this question. The options most often chosen areeither the respective nationality first and Europe second or nationality only. Answering‘Europe first and nationality second’ or ‘only European’ occur very seldom as the lines atthe bottom indicate. In the early 1990s and more recently the answer ‘nationality first andEuropean second’ was chosen most often. However, sometime in between (1996–99 and2001), the answer most often chosen was the respective nationality only. In the years1992–2004 there is no obvious trend indicating an increase in European identification.An obvious rise in European identification, indicating influence by the individual gainsavailable from the internal market, or by the other attempts to increase European identi-fication, cannot be found.8 Therefore, we can rule out the possibility that the current extentof European identification is the result of a considerable increase after a very low initiallevel. Also, a pronounced decrease as a reaction to controversial EU policies is not foundin these results. Instead, we can conclude that the extent of European identification in theEU is not peculiar to Europe, but within the usual range of continental identification foundaround the world.

7 The wording is: ‘In the near future, will you see yourself as [. . .]’, with the options ‘Nationality only’, ‘Nationality andEuropean’, ‘European and Nationality’ and ‘European only’.8 The result remains identical when we look at another version of the question. In some studies of the Eurobarometerrespondents were asked how often they feel not only national identification but also European, with the options ‘often’,‘seldom’ or ‘never’. Changes in response between 1982 and 2005 for the ten states which were already members in 1982are moderate, and again with no clear trend (Roose, 2007).



V. Socio-structural Differences in Continental Identification

Besides differences between the extent of individual identification, we should also expectspecific variation in continental identification according to socio-demographic character-istics. As argued above, within the EU the highly educated with higher professional statusand income should identify with Europe to a larger extent than respondents from lowersocial strata.

We again turn to the ISSP data set but now compare the influence of socio-economicvariables on continental identification. Our theoretical arguments led to the expectationthat the influences should be different in Europe compared to countries on other conti-nents. Transposed to a statistical analysis, the question is whether the data are compatiblewith identical coefficients for all country samples or whether we do in fact find a modelof influences which is compatible with the European country samples only. The depen-dent variable – identification with one’s continent – is measured in four steps. In the firstapproach I calculated ordered logistic regressions – one for each country sample. Theidentification with one’s continent was explained by the available socio-economic ques-tions regarding household income, occupational prestige, education, gender and age. Thequestion was whether there is a set of coefficients that fits to all country samples. Thismeans the values for the coefficients should be within the 95 per cent confidence intervalfor the logistic regression of each country.

Figure 1: Self-Concept: European versus National, 1992–2004

Source: EB 37.0, EB 40.0, EB 42.0, EB 43.1, EB 44.1, EB 46.0, EB 47.1, EB 49.0, EB 50.0, EB 52.0, EB 53.0, EB 54.1,EB 56.2, EB 57.1, EB 58.1, EB 59.1, EB 60.1, EB 61, EB 62.0 and author’s own calculations.

290 Jochen Roose


The result is somewhat mixed. In fact there is no set of coefficients for these fivevariables which is compatible with all logistic regressions. For each coefficient, we findsome countries that deviate. The number of country samples with coefficient confidenceintervals that do not cover the best fitting value are between three and seven out of 29samples. However, no country deviates in more than two coefficients, and for all coeffi-cients the deviant countries stem from all groups – that is, EU-15, other European andnon-European countries. That means deviation from the typical pattern does not follow adistinction between Europe versus the rest of the world. The assumption of a specificEuropean pattern of socio-economic influences on continental identification is notsupported by this evidence.

Comparing patterns of socio-economic influences by country regressions is suboptimalbecause of the inductive approach fitting the coefficients to the data. The question here isdifferent, though, as we are looking for a set of coefficients acceptable (though possiblynot optimal) for each country sample and identical for all countries. The appropriatestatistical test for homologous influence patterns is structural equation models com-paring multiple groups. These procedures are able to fit coefficients to several samples,allow the specification of identical influences across country samples between the avail-able variables for the socio-economic position and different influences among thesevariables, and they produce general fit measures which allow us to judge whether theresult is acceptable or not. The application of structural equation modelling with anordinal dependent variable turned out to be too complex in this case and resulted innon-identification. The situation is different though if we rely on the linearity of thedependent variable.9

The hypothesis to be analyzed by the group comparison is whether there is one specificset of coefficients that fits all country samples at the same time.10 The extent to which thisis possible will be indicated by general fit measures. In a simple structural equation model,the influence of occupational prestige, income, age, education and gender on an individu-al’s identification with the respective continent is modelled. Some of the variables can beexpected to be highly correlated. Occupational prestige, education and income are obvi-ously closely connected. Additionally, age can be expected to have an influence on incomeand, due to the expansion of education, also on the educational level.11

What should we expect from the group comparison when we assume a specificinfluence of socio-economic variables on continental identification in the EU and adifferent pattern outside the EU? First, we should expect that we will be unable to fit amodel with fixed path coefficients for all groups.12 Second, we should expect that a modelwhich has optimal fit for all countries would fit poorly for the EU or EFTA members.

In a first step, the group comparison is fitted to all countries for which data are available(Figure 2). This means coefficients were calculated that are identical for all countrysamples, including countries from the EU as well as from other continents and Europe

9 For example, Bentler and Chou (1987) regard four categories as acceptable to be treated as interval.10 As the structural equation models are calculated separately for each country sample, the unrepresentative countryselection is not a problem.11 These correlations can be found as double arrows connecting the error terms of the respective variables in Figure 2.Because these relations will vary across the countries but are not in the focus of this analysis, these error correlations areallowed to vary across countries,12 This means that the coefficients of causal influence (equal to regression coefficients in regression analysis) are equal forall groups – that is, country samples.



beyond the EU. The model fit is satisfying, which is a surprising result considering that wehave a very heterogeneous sample of countries being analyzed.13

The coefficients in the model confirm that people with better professional positions,higher income and better education tend to identify more with their continent. Addition-ally, older people tend to identify more with their continent.14 According to the coeffi-cients, men tend to identify less with their continent, but this finding is not significant atthe 0.1 per cent level. These results come as no surprise. What is remarkable though is thatthe influences do not seem to vary across the countries around the world.

To verify this finding, we can make an additional test. The sample of availablecountries we have used is dominated by EFTA members. To verify our results with astricter test we can determine whether the coefficients in the model are also valid for thenon-EFTA countries when analyzed on their own. Table 2 compares the Goodness of Fitmeasures for structural equation models between the countries belonging to the olderEU-15 Member States, the countries belonging to the EU-27 Member States and theavailable non-EFTA countries, where the coefficients were fixed to the respective valuecalculated for the whole sample above. The Goodness of Fit measures for the EU-27countries indicate an acceptable quality of the model which is similar to the overall model.This is no surprise because these countries make up the largest share in the sampleanalyzed. The model fit for the countries from the EU-15 members is about as good.However, the Goodness of Fit measures for the model calculated for the non-EFTAcountries are very good as well. So the results for the theoretical model (see Figure 2) inthe six non-EFTA countries are very similar to the results for older and newer EU MemberStates.

13 The global fit measures are: RMR = 0.023, GFI = 0.978 (usual threshold is >0.95), AGFI = 0.951 (>0.90), CFI = 0.928(>0.90), NFI = 0.918 (>0.90), RMSEA = 0.015 (< 0.10). For recommended thresholds of model fit see, for example, Hu andBentler (1995).14 All of these coefficients are significant on the 0.1 per cent level.

Figure 2: Path Analysis for Closeness Felt to One’s Continent: All Countries with Fixed Effects

-0,027

0,115

0,420

0,053

0,241 e2 occupational

prestige

income

age

education

male

identificationwith

continent e1

e3

e4

e5

e6

Source: ISSP 2003 and author’s own calculations.

292 Jochen Roose


The same test has been carried out for a set of coefficients that is derived from a groupcomparison of the EU-15 countries only (not reported). In the first step, identical coeffi-cients for all samples from available EU-15 countries were calculated and then fixed fora group comparison of samples from EFTA countries and from non-EFTA countries. Theresult of this group comparison is the same. The fit measures indicate acceptable modelfit for all these groups. That means the data gathered on continental identification innon-EFTA countries are compatible with an influence structure that is derived from EU-15countries or from all countries available. This does not mean that in all cases thesecoefficients are the best fit for the data, but the data give us no safe ground to conclude thatthe influence structures of socio-economic variables on continental identification differbetween the available EU-15 countries, EU-27 countries, EFTA countries or countries onother continents. The assumption that continental identification in the EU isdue to the effects and opportunities of European integration and EU policies is not wellsupported by these results.

Conclusions

The EU is unique within the world. On no other continent do we find such an endeavourof uniting established nation-states in a new political system. This political and economicprocess should be accompanied by a change in attitudes and identification of Europeancitizens. ‘Through its institutions (economic, legal, or educational), the EU can require ormotivate major swaths of behavioral change, and it would be foolhardy to believe suchbehavioral changes are isolated from identity changes’ (Breakwell, 2004, p. 26).

There has been ample research on European identity and its development withinthe EU. However, a comparison between the continental identification of Europeans andof people in other parts of the world with their respective continents had not beenpreviously carried out. In theory, it is very plausible that the level and structure ofEuropean identification is specific to Europe and the EU. Identification through personaltransnational experiences, through personal gains due to EU policies or elite construction,

Table 2: Comparison of Goodness of Fit Measures for PathAnalysis Models Explaining Closeness Felt to the RespectiveContinent

Model for EU-27countries

Model for EU-15countries

Model for non-EFTAcountries

RMR 0.028 0.030 0.031GFI 0.976 0.976 0.973AGFI 0.954 0.953 0.945CFI 0.923 0.920 0.917NFI 0.911 0.908 0.908RMSEA 0.018 0.022 0.030

Source: ISSP 2003 and author’s own calculations.Notes: Coefficients are fixed to the values of the reference model for all countrysamples; RMR = Root Mean Square Residual; GFI = Goodness of Fit Index(usual threshold >0.95); AGFI = Adjusted Goodness of Fit Index (>0.90);CFI = Comparative Fit Index (>0.90); NFI = Normed Fit Index (>0.90);RMSEA = Root Mean Square Error of Approximation (<0.10).



based on collective symbols and narratives, would lead us to expect the EU population toidentify more strongly or at least in a specific way with Europe.

The data available have not shown a clear difference between continental identificationin Europe or the EU countries and countries on other continents. The differences betweencountries are sometimes large and much larger than the differences between continents.Country-specific explanations for continental identification, relating to the specific historyand national discourses seem to be crucial (for example, Díez Medrano, 2003; Marcussenand Roscher, 2000). A difference between EU or EFTA countries compared to countrieselsewhere in respect to continental identification could not be found. In addition, theassumption of a specific pattern of socio-economic influences on continental identificationin Europe could not be substantiated. The data we have are compatible with the assump-tion of identical influences for all available countries.

Before reaching conclusions from these results some words of caution must be men-tioned. Measuring identification is a challenge (Bruter, 2006). Relying only on one surveyquestion (like most other quantitative analyses) is therefore a weak foundation. Addition-ally, the data are subject to all the problems of international comparison (Van Deth, 1998).And we should keep in mind that the country selection, especially beyond Europe, is notoptimal. There are some reasons not to build very tall theoretical buildings on such unsafeground. Still, ignoring empirical information, especially as it is the only information wehave on an important question, is also not an advisable strategy.

There are some lessons to be learned. First, explanations of identification withEurope referring to the effects and outcomes of EU policies are not the only possibility.According to the presented results, identification with one’s continent underlies mecha-nisms that are of general impact in many countries around the world. At least thepersistent finding of the upper social strata identifying more with Europe is not suffi-cient support for explanations relating to gains or opportunities stemming from EUpolicies as this pattern can be similarly found on other continents. Possibly, it is insteada general cosmopolitan perspective, learned among higher social strata, which widensthe view to the continent as a whole (cf. Hannerz, 1990; Beck and Grande, 2004; Mauet al., 2008). The idea of cognitive mobilization presented decades ago by Inglehart(1970) and Deutsch (1961) is more in accordance with our findings than references toEU policies.

Second, we could conclude for attitude research with respect to the EU that it ispossible that identification with Europe is not simply part of a general syndromeof attitudes on the EU, but rather might be a disposition sui generis. According to thefindings presented here, it may be that it is not so much the evaluation of the EU thatresults in a European identification, but the identification that has an influence over theevaluation.

Third, for the European integration process and elite strategies to support this process,these findings could bear fundamental consequences. European identification, with itsinfluence on support for EU integration and EU policies (see, for example, Citrin andSides, 2004; Fuchs et al., 2009; Roose, 2007), seems to be not closely related to selectivegains. The effort to bind citizens emotionally to Europe by successful policy output mightbe much less effective than many hope. In addition, the activities to enhance identificationby introducing common symbols of unity and a common historical narration have beenwithout sweeping effect so far.

294 Jochen Roose


Finally, researchers of European identity should not forget to widen their view beyondthe EU. Some trends that seem specific to the EU experience may turn out to be part ofglobal phenomena unrelated to the unique European integration process.

Correspondence:Jochen RooseInstitute for SociologyFree University BerlinGarystrasse 5514195 BerlinGermanyemail: [email protected]

References

Anderson, S. and Seitz, T.R. (2006) ‘European Security and Defense Policy Demystified:Nation-Building and Identity in the European Union’. Armed Forces and Security, Vol. 33,No. 1, pp. 24–42.

Beck, U. and Grande, E. (2004) Das kosmopolitische Europa (Frankfurt am Main: Suhrkamp).Bentler, P.M. and Chou, C.-P. (1987) ‘Practical Issues in Structural Modeling’. Sociological

Methods and Research, Vol. 16, pp. 78–117.Breakwell, G.M. (2004) ‘Identity Change in the Context of the Growing Influence of European

Union Institutions’. In Herrmann, R.K., Risse, T. and Brewer, M.B. (eds).Brubaker, R. and Cooper, F. (2000) ‘Beyond Identity’. Theory and Society, Vol. 29, No. 1,

pp. 1–47.Bruter, M. (2004) ‘Civic and Cultural Components of a European Identity’. In Herrmann, R.K.,

Risse, T. and Brewer, M.B. (eds).Bruter, M. (2006) Citizens of Europe? (Basingstoke: Palgrave Macmillan).Castano, E. (2004) ‘European Identity: A Social-Psychological Perspective’. In Herrmann, R.K.,

Risse, T. and Brewer, M.B. (eds).Checkel, J.T. and Katzenstein, P.J. (eds) (2009) European Identity (Cambridge: Cambridge Uni-

versity Press).Cinnirella, M. (1996) ‘A Social Identity Perspective on European Integration’. In Breakwell, G.M.

and Lyons, E. (eds) Changing European Identities (Oxford: Butterworth Heinemann).Citrin, J. and Sides, J. (2004) ‘More than Nationals’. In Herrmann, R.K., Risse, T. and Brewer,

M.B. (eds).Deutsch, K.W. (1961) ‘Social Mobilization and Political Development’. American Political

Science Review, Vol. 55, No. 3, pp. 493–514.Díez Medrano, J. (2003) Framing Europe (Princeton, NJ: Princeton University Press).Dubé, S. and Magni Berton, R. (2009) ‘How Does Income Influence National and European

Identity?’ In Fuchs, D., Magni-Berton, R. and Roger, A. (eds) Euroscepticism (Opladen:Barbara Budrich).

Easton, D. (1975) ‘A Re-assessment of the Concept of Political Support’. British Journal ofPolitical Science, Vol. 5, No. 4, pp. 435–57.

Favell, A. (2009) ‘Immigration, Migration and Free Movement in the Making of Europe’. InCheckel, J.T. and Katzenstein, P.J. (eds).

Fligstein, N. (2008) Euroclash (Oxford: Oxford University Press).Fuchs, D., Guinaudeau, I. and Schubert, S. (2009) ‘National Identity, European Identity and

Euroscepticism’. In Fuchs, D., Magni-Berton, R. and Roger, A. (eds).



Ganzeboom, H.B. and Treiman, D.J. (1996) ‘Internationally Comparable Measures of Occupa-tional Status for the 1988 International Standard Classification of Occupations’. Social ScienceResearch, Vol. 24, pp. 201–239.

Gellner, E. (1968) Thought and Change (London: Weidenfeld & Nicolson).Hannerz, U. (1990) ‘Cosmopolitans and Locals in World Culture’. Theory, Culture and Society,

Vol. 7, No. 2, pp. 237–51.Herrmann, R.K. and Brewer, M.B. (2004) ‘Identities and Institutions: Becoming European in the

EU’. In Herrmann, R.K., Risse, T. and Brewer, M.B. (eds).Herrmann, R.K., Risse, T. and Brewer, M.B. (eds) (2004) Transnational Identities (Lanham, MD:

Rowman & Littlefield).Hobsbawm, E.J. (1983) ‘Introduction: Inventing Traditions’. In Hobsbawm, E.J. and Ranger, T.

(eds) The Invention of Tradition (Cambridge: Cambridge University Press).Hobsbawm, E.J. (1991) Nations and Nationalism since 1780 (Cambridge: Cambridge University

Press).Hu, L.-T. and Bentler, P.M. (1995) ‘Evaluating Model Fit’. In Hoyle, R.H. (ed.) Structural

Equation Modeling (Thousand Oaks, CA: Sage).Immerfall, S. and Sobisch, A. (1997) ‘Europäische Integration und Europäische Identität’. Aus

Politik und Zeitgeschichte, Vol. 47, No. B10, pp. 25–37.Inglehart, R. (1970) ‘Cognitive Mobilization and European Identity’. Comparative Politics, Vol. 3,

No. 1, pp. 45–70.Karolewski, I.P. and Kaina, V. (eds) (2006) European Identity (Berlin: LIT).Kohli, M. (2000) ‘The Battlegrounds of European Identity’. European Societies, Vol. 2, No. 2,

pp. 113–38.Kuhn, T. (2012a) ‘Europea ante Portas: Border Residence, Transnational Interaction and Euro-

scepticism in Germany and France’. European Union Politics, Vol. 13, No. 1, pp. 94–117.Kuhn, T. (2012b) ‘Why Educational Exchange Programmes Miss Their Mark: Cross-Border

Mobility, Education and European Identity’. JCMS, Vol. 50, No. 6, pp. 994–1010.Laffan, B. (2004) ‘The European Union and Its Institutions as “Identity Builders” ’. In Herrmann,

R.K., Risse, T. and Brewer, M.B. (eds).Leggewie, C. (2008) ‘A Tour of the Battleground: The Seven Circles of Pan-European Memory’.

Social Research, Vol. 75, No. 1, pp. 217–34.Marcussen, M. and Roscher, K. (2000) ‘The Social Construction of “Europe”: Life-Cycles of

Nation-State Identities in France, Germany and Great Britain’. In Stråth, B. (ed.) Europe andthe Other and Europe as the Other (Bruxelles: Lang).

Marks, G. and Hooghe, L. (2003) National Identity and Support for European Integration. (Berlin:Wissenschaftszentrum).

Mau, S., Mewes, J. and Zimmermann, A. (2008) ‘Cosmopolitan Attitudes through TransnationalPractices?’ Global Networks, Vol. 8, No. 1, pp. 1–24.

McLaren, L. (2007) ‘Explaining Mass-Level Euroscepticism’. Acta Politica, Vol. 42, No. 2,pp. 233–51.

Peterson, J. and Bomberg, E. (1999) Decision-Making in the European Union (Basingstoke:Palgrave Macmillan).

Risse, T. (2005) ‘Neofunctionalism, European Identity and the Puzzles of European Integration’.Journal of European Public Policy, Vol. 12, No. 2, pp. 291–309.

Risse, T. (2010) A Community of Europeans? (Ithaca, NY: Cornell University Press).Roose, J. (2007) ‘Die Identifikation der Bürger mit der EU und ihre Wirkung für die Akzeptanz von

Entscheidungen’. In Nida-Rümelin, J. and Weidenfeld, W. (eds) Europäische Identität (Baden-Baden: Nomos).

Roose, J. (2010) Vergesellschaftung an Europas Binnengrenzen (Wiesbaden: VS).

296 Jochen Roose


Smith, A.D. (1991) National Identity (London: Penguin Books).Stråth, B. (2000) ‘Multiple Europes: Integration, Identity and Demarcation of the Other’. In

Stråth, B. (ed.).Tajfel, H. (1978) Differentiation between Social Groups (London: Academic Press).Theiler, T. (2012) ‘Does the European Union Need to Become a Community?’ JCMS, Vol. 50,

No. 5, pp. 783–800.Tilly, C. (ed.) (1975) The Formation of National States in Western Europe (Princeton, NJ:

Princeton University Press).Van Deth, J.W. (1998) ‘Equivalence in Comparative Political Research’. In Comparative Politics

(London: Routledge).Wodak, R. (2004) ‘National and Transnational Identities: European and Other Identities

Constructed in Interviews with EU Officials’. In Herrmann, R.K., Risse, T. and Brewer, M.B.(eds).



Download - How European is European Identification? Comparing Continental Identification in Europe and Beyond

Top Related