ETHNO-NATIONAL, RELIGIOUS, IDEOLOGICAL AND SEXUAL DIVERSITY EUROPEAN ELITE AND CITIZEN VIEWS COMPARED MARTINA KLICPEROVÁ-BAKER JAROSLAV KOSTÁL rN o Abstract In contexts of multi-level governance, such as we find in the European Union, where elites are more active in the public sphere, it is particularly crucial to assess whether citizens' views correspond to the views of the elites who claim to represent them. This article compares the views of elites with the views of representative samples of citizens, with a focus on their views on ethno-national, religious and sexual diversity. Findings confirm relationships between elite/citizens views and revealed several rules: Firstly, ethnic and ideological groups which were commonly rejected from neighbourhoods were recognised by elites as relevant for social diversity. Secondly, the most accepted migrant workers by citizens were also viewed as most relevant for social diversity by elites. Finally, sexual diversity manifested a more complex relationship - where gays are most accepted, they are either viewed by elites as highly relevant (Austria, Denmark) or irrelevant for social diversity (Czech Republic, France, Italy, Spain). In countries with high public rejection of gays, LGBT tend to be viewed by elites as very relevant (Turkey, Bulgaria, Estonia). Elite views of relevance push the public to a greater tolerance; public intolerance increases recognition of relevance of marginalised groups. Martina Klicperova-Baker is Research Scholar at the Institute of Psychology, Academy of Sciences of the Czech Republic; e-mail: klicperovabaker@gmail.com. Jaroslav Kostal is Eurosphere Research Fellow at the Institute of Psychology, Czech Academy of Sciences; e-mail: Jaroslav.Kostal@centrum.cz. LO 00 rN 1—I o rN 0"> £ £ 3 a 0 £ * 1 * (R 0 C ! LO 00 Introduction As societies open up to the world, they face increased diversity. Whether ethno-national, religious, ideological or sexual - diversity is an important factor influencing life in the modern, globalised world. It is particularly significant for the EU, a region aspiring to co-existence through supranational citizenship. Diversity as a Challenge. In the academic world, as well as in popular discourse, there is an argument about the extent to which diversity provides cultural enrichment, and at what point it becomes challenging or even endangering. Multicul-turalism - embracing acceptance and understanding of variety - competes with isolationism. Paradoxically, isolationists argue that their philosophy conserves diversity better than the mixing and blending of cultures: in other words, countries should aggressively protect their uniqueness, and minorities should protect theirs (e.g. Milliken 2010). Scholarly studies devoted to effects of diversity, heterogeneity and fractionalisa-tion of societies agree that diversity creates tensions and challenges; the dispute is focused on how these strains fit into a larger picture of human coexistence. Under some conditions, diversity leads to the betterment of societies, but in other cases, the tensions caused have an adverse effect. An expert on democracy, Robert Putnam, pointed out that "immigration and ethnic diversity challenge social solidarity and reduce social capital" (2007, 138). Putnam also noted that these adverse effects (e.g., a loss of trust) not only affect trust in other groups, but, contrary to typical assumptions, also erode trust among the in-group members. Putnam backed up his statements using extensive American data, which documented that both inter-racial trust and trust of neighbours increased with the racial homogeneity of neighbourhoods. Gerritsen and Lubbers (2010), among others, confirmed that this conviction also had relevance to the conditions in the EU, claiming that cultural diversity within the EU decreases levels of trust. Scholars often attempt to soften this generally pessimistic perspective on diverja sity: e.g., Hooghe et al. (2009, 198) summed up the debate by stating that "diversity does not exert the consistent and strong negative effects often attributed to it" and that "fullblown negative relationship between ethnic diversity and generalised trust does not hold across Europe." The real challenge to the generalised negative relationship between social diversity and social capital often comes either from research focused on well-defined, particular aspects, or from studies which attempt a longer perspective. Specifically: ^ a) Studies describing thriving examples of ethnically diverse societies tend to focus on well defined smaller areas; areas with frequent inter-ethnic contacts and hence better mutual knowledge, subjects with a university education, those who have a positive image of other countries, etc. Additionally, successful co-existence is more likely in areas where diversity is limited in scope: such as when languages are similar, religions are close, and there is little variation in socioeconomic status. b) The second type of study that is optimistic about diverse coexistence is the study that considers a long time span. Here we should quote Robert Putnam again (2007, 138-139): In the medium to long run, on the other hand, successful immigrant societ-qq ies create new forms of social solidarity and dampen the negative effects of (R 0 diversity by constructing new, more encompassing identities. Thus, the central challenge for modern, diversifying societies is to create a new, broader ^ sense of "we." This "broader sense of 'we'" includes not only ethnic but also ideational and sexual diversity, and this new sense of "we" is as inspiring for the USA (to which Putnam mostly refers) as it is for the European Union. In its broadest definition, the sense of "we" could expand to include all humanity (McFarland 2011). The international Eurosphere consortium has focused on the expanding topic of diversity from various angles. The Eurosphere working papers alone cover a wide array of topics: from universal perspectives (Sicakkan 2008), methodological issues in diversity research (Perez 2009), liberal responses to diversity (Baubock 2008) and concerns about diversity framing (Huszka 2009), to studies focused particularly on diversity and immigrants (Sicakkan 2009), diversity and gender (Siim 2009, Nielsen 2010), diversity within the EU, and the process of European integration (Mokre and Nielsen 2010, Sata 2010). The Role of Elites in the Public Sphere. There is no doubt that elites play a crucial role in the public sphere - whether they are "policy elites," "opinion elites" (Almond 1960), "power elites" (Mills 1965, Splichal 2002, 165) or, most importantly, the "communications elite" (Almond 1960) and "consensually unified elite" (Diamond 1999, 218). Risse (2010, 234) pointed out that "the European Community of communication is almost exclusively populated by elites rather than by civil society." Margolis and Mauser (1989, 87) observed that public opinion "is dependent on elite initiatives that are linked to the public via the mass media and other means" - hence the course of events in this arena is controlled by the elite, while the citizens are limited merely to attempts to limit that control, and may struggle not to feel alienated (Knobloch 2011). Risse, among others, noted a cultural cleavage, a democratic deficit due to the fear of the elites to start a public debate, rock the boat and wake a sleeping giant (Risse 2010, 240-242). Statham (2010, 292) stresses that it is the "overdomination by elite actors of the Europeanised debates" which constitutes the substance of the EU public sphere deficit. The Public Sphere as an Arena of Citizens and Elites. The natural differences between the views of elites and citizens have been widely acknowledged and documented. Among others, Papadopoulos (1995) demonstrated the clash of views on political referenda between ruling elites (who consider them disruptive) and citizens (for whom they symbolise empowerment). Nissen (2003) and Diez-Medrano (2003) proved different levels of support for EU integration among "ordinary" citizens and among intellectual, political and local elites in various European countries. Diversity and minorities can be classed as sensitive issues, and potential disparities between citizen and elite views are worthy of exploration. The notion that cultural conflicts are largely the creations of intellectual elites, as opposed to representing real problems bothering average citizens, has been largely dismissed (e.g., Yates 2001, criticising the lighthearted approach to diversity taken by Smelser and Alexander 1999). It has also been argued that average citizens suffer from the burdens of diversity more than "elites who tend to be both morally and materially insulated from the common people" (Devine 1996). This view can also be illustrated in an abundance of real-life vignettes: for example, during the controversy over a difficult coexistence at the boundary between a Roma project and family houses (the case of Maticni street in Usti nad Labem in the Northern Czech Republic), some citizens accused president Havel, who defended the Roma, of elitist pseudo humanism. They suggested that he should purchase or rent one of the family houses and try to live there himself. After all, if the general citizenry is underrepresented in the public sphere, we should know what views are underrepresented. How can we compensate for this imbalance in order to foster democracy in the EU? This study aims to shed light the issue, on the path towards answering some of these questions. Method The core of this study is a comparison of citizen and elite attitudes to diversity across various European countries, searching for relationships and imbalances between their views. Design and Survey Items. Two initial data sets were employed: the first was comprised of elite views. It was created at our international Eurosphere consortium (the respondents and the procedure are described below). The Eurosphere projects focused on multiple facets of diversity in current European society. Its extensive survey was introduced by a question "In your own notion of diversity, which groups do you believe are relevant today for defining a diverse society?"(Question Qv1, variables Qv1_1 to Qv1_17, answered by 725 respondents). Responses were categorised into a list of 18 diversity categories, including ethnic, migrant, ideological, class, disability, gender, sexual, linguistic, social economic and age groups. Data were collected in 2008/2009. At the same time, though approaching the question from a different angle, tolerance to diversity was the subject of an international survey by the European Values Study (EVS). In question Q6 (variables v46-v60) EVS presented respondents with a list of 14 groups/minorities (ethnic, religious, sexual, etc.) and asked the question "Could you sort out any (of this list) whom you would not like as neighbours?" Responses from countries which also participated in Eurosphere constituted the source for our second data set. When we wrote this study in 2011, this set included the available £ data from 22,128 respondents participating in the fourth wave of EVS.1 As soon as we had the opportunity to broaden the sample using the updated data edition of the fourth EVS wave, we did so. During the revision phase of this issue, we expanded the data set to N=25,196 and included Norway in the data wherever feasible. These represented two different approaches, target groups and different wording of questions, yet both were based on a similar underlying issue: the diversity of ^ citizens in Europe at the present time. The subject matter overlapped but was not quite identical; for comparison we had to drop from each data file those items which had no adequate counterparts. For example, Eurosphere asked about generational diversity, cultural and language groups, shifting and territorial belonging, but EVS did not; on the other hand, EVS questioned people with a criminal record, drug addicts and heavy drinkers, emotionally imbalanced and AIDS patients, as well as large families, but Eurosphere did not. Despite above mentioned differences, we compiled comparable data from 16 countries in total. They are presented in a condensed form, as country percentages, in Table 1. These figures, as well as their qq standardised z-transformations, were the subject of our analysis. c > Table 1: Eurosphere and EVS Responses. Eurosphere Elite Responses (percentage by country; positive answers to the question "Which groups are relevant today for defining a diverse society?") vs. European Values Study (citizen re- 00 sponses, percentages by country; affirmative answers to the question "Could you sort out any (of this list) whom you would not like as neighbours?"; significantly higher percent values are highlighted in the columns) Eurosphere: Elite responses European Value Study: Citizen responses Ethnic groups (people identifying with a specific ethnic group) V1 7 Ideological groups (people identifying with a specific ideology) V1 8 Migrant groups (people coming from non-European countries) V1_10 Religious groups (people identifying with a specific religion) V1 13 Sexuality groups (e.g., gays, lesbians, transsexuals, homosexuals, etc) V1_14 N Left wing extremists v48 Q6C Right wing extremists v50 Q6E Muslims v53Q6H Jews v58 Q6M Immigrants/foreign workers v54 Q6 People of different race v47 Q6B Roma v59 Q6N Gays v57 Q6L N 1 Austria 73.08 46.15 42.31 55.77 53.85 52 47.28 61.52 30.07 16.89 22.52 17.15 30.20 23.31 1510 2 Belgium 50.00 16.67 50.00 37.50 37.50 24 25.31 37.64 14.45 3.84 6.16 5.37 26.11 6.69 1509 3 Bulgaria 84.21 31.58 24.56 63.16 36.84 57 43.20 44.40 18.80 14.07 16.80 20.00 47.47 51.47 1500 4 Czech Republic 76.32 7.89 7.89 15.79 7.89 38 33.22 36.02 29.16 11.42 28.83 21.58 54.81 22.35 1821 5 Denmark 66.04 18.87 28.30 52.83 54.72 53 5.37 16.85 10.95 1.73 5.71 3.92 32.18 4.78 1507 6 Estonia 47.22 30.56 19.44 44.44 38.89 36 31.16 28.99 32.61 21.67 31.09 23.85 44.40 47.69 1518 7 Finland 59.62 13.46 21.15 46.15 34.62 52 17.90 23.81 22.49 4.67 15.34 8.73 48.68 11.38 1134 8 France 34.29 5.71 8.57 17.14 8.57 35 13.32 28.65 7.53 2.33 4.26 3.40 25.18 5.66 1501 9 Germany 58.33 35.42 33.33 52.08 37.50 48 37.06 65.88 23.71 5.49 10.51 4.14 26.89 15.47 2075 10 Hungary 80.00 18.18 9.09 34.55 47.27 55 11.43 12.62 10.97 6.35 15.14 8.92 38.47 29.28 1513 11 Italy 66.07 .00 92.86 16.07 .00 56 26.93 30.41 21.33 11.39 15.08 14.75 59.78 20.34 1275 12 Netherlands 66.67 14.29 33.33 42.86 40.48 42 58.17 67.57 18.40 7.85 15.06 10.88 29.34 10.62 1554 13 Norway 17.07 4.88 12.20 9.76 9.76 41 14.50 31.83 13.21 2.84 5.96 5.05 24.50 5.60 1090 14 Spain 16.67 10.42 45.83 16.67 10.42 48 15.53 19.27 12.60 2.47 4.07 3.93 25.47 5.27 1500 15 Turkey 90.20 13.73 .00 76.47 27.45 51 62.37 61.79 5.91 67.41 47.69 42.03 66.07 89.14 1206 16 United Kingdom 54.05 16.22 27.03 32.43 32.43 37 25.30 29.92 12.75 3.20 14.61 5.77 33.31 10.70 1000 Total 60.69 18.21 29.10 39.72 30.48 725 31.10 39.13 17.68 13.51 17.49 13.59 39.28 25.05 25196 Per-country mean 58.85 17.78 28.53 38.43 29.92 16 29.36 37.43 17.80 11.60 16.25 12.53 38.36 22.63 16 Standard deviation 21.9 12.00 21.76 18.85 16.77 16.29 17.29 7.88 15.51 11.34 10.15 13.11 22.15 Eta/Cramer's V .432 .316 .506 .394 .385 .366 .367 .213 .535 .330 .332 .284 .571 Participants. Eurosphere, with 130 researchers in 16 countries, carried out an interview survey focused on diversity, European Union, and the European public sphere. The focus of interview data collection was on elites - significant members of major political parties, actors within social movements, NGOs, think tanks and the media (for a detailed description of the methodology, please see Sicakkan 2008). To minimise a selection bias, institutions were selected according to a general key (e.g., the parties selected included two major parties and a maverick party; social movements included national as well as transnational institutions; think tanks had to include advocacy think tanks, universities without students, and contract research organisations; broadcast media were to include public service as well as commercial broadcasters; print media comprised a major daily as well as minor periodicals etc.). The interviewees selection key instructed the Eurosphere researchers to choose three to seven members from each institution (e.g., from political parties it was an organisational leader, an opinion leader, two internal opposition leaders, and three internal "group" leaders; from think tanks, an organisational leader, a research leader and a prominent researcher; from the media, a representative from the Chief Editor's office, a news-section editor and up to two news-section journalists). Although not all the teams fulfilled the originally planned quota of respondents (because the saturation point for information about some organisations was reached before the quota was filled, and in a few cases, the elites that the Eurosphere team contacted were not accessible), a satisfactory number of 725 elite respondents from diverse backgrounds were interviewed (N column in Table 1), and answers were coded and entered into a central database administered by Norwegian Social Science Data Services and the University of Bergen. Data Treatment. Initial analyses of Eurosphere and EVS data were conducted separately. Means and variances were analysed by ANOVA and t-test to determine a level of international variance within the data. In parallel, frequency counts were subjected to contingency analyses (CHI2 and adjusted residuals at sig. level <.01, Cramer's V for measuring the intensity of the relationship and a sign test for measuring higher than expected frequency on adjusted residuals). The second step focused on analyses of Eurosphere and EVS data together to determine the degree of agreement between them. This stage included: • Testing the independence of both samples pairwise, using Student t-tests of independent samples for percentages; • Searching for patterns and testing closeness of relationships between the two samples using ANOVA and correlation analysis (Pearson's r and scatterplots, Spearman's rho and Eta, as well as ALSCAL distance models); • Finally, due to the risk of bias in a small sample, verification of previous results A by non-parametric rank-order tests for two independent samples: the MannWhitney test was used to check for parallel design of data (null hypothesis signaled that both samples came from the same underlying distribution). We also calculated transposed standardised z-percentages (zp) to control for an individual bias within both samples: Z=(x-Xbar)/SDx Zp=(Z+Min(Z)/(Max-Min (Z))*100 where x.=score, SDx=standard deviation; Xbar=mean over the whole range of Q multiple dichotomies (i.e., 18 variables in Eurosphere and 14 variables in EVS); Min=minimum and Max=Maximum. The following data matrix was computed for multidimensional scaling (ALSCAL): >i u >j v v >j v' where D..=Euclidean distance and a,b,c,d are the four variables required for ^ calculation of the distance between A and B (e.g., a=Eurosphere view of ethnicity in cn Relationships between Eurosphere and EVS data Having verified that both Eurosphere and EVS data manifest statistically significant international variance, we progressed to analysis of both data files together. Table 2: Ethnic, Migrant, Ideological and Sexuality Diversity: Relevance and Intolerance (percent scores by country). Superimposition of the elite view on the relevance of various groups to defining social diversity (Eurosphere N = 725, variables starting with V1_) and citizens' rejections of having neighbours from the relevant groups (EVS N = 22,128, variables v47-v59) Ethnic groups (V1_7) Migrant groups (V1_10) Religious groups (V1_13) Roma (v59) Jews (v58) Muslims (v53) Different race (v47) Foreign workers (v54) Ideological groups (V1_8) Left-wing extremists (v48) Right-wing extremists (v50) Sexuality groups (V1_14) Gays (v57) 1 2 3 4 5 6 7 8 9 10 11 12 13 1 Austria 73.08 42.31 55.77 30.20 16.89 30.07 17.15 22.52 46.15 47.28 61.52 53.85 23.31 2 Belgium 50.00 50.00 37.50 26.11 3.84 14.45 5.37 6.16 16.67 25.31 37.64 37.50 6.69 3 Bulgaria 84.21 24.56 63.16 47.47 14.07 18.80 20.00 16.80 31.58 43.20 44.40 36.84 51.47 4 Czech Republic 76.32 7.89 15.79 54.81 11.42 29.16 21.58 28.83 7.89 33.22 36.02 7.89 22.35 5 Denmark 66.04 28.30 52.83 32.18 1.73 10.95 3.92 5.71 18.87 5.37 16.85 54.72 4.78 6 Estonia 47.22 19.44 44.44 44.40 21.67 32.61 23.85 31.09 30.56 31.16 28.99 38.89 47.69 7 Finland 59.62 21.15 46.15 48.68 4.67 22.49 8.73 15.34 13.46 17.90 23.81 34.62 11.38 8 France 34.29 8.57 17.14 25.18 2.33 7.53 3.40 4.26 5.71 13.32 28.65 8.57 5.66 9 Germany 58.33 33.33 52.08 26.89 5.49 23.71 4.14 10.51 35.42 37.06 65.88 37.50 15.47 10 Hungary 80.00 9.09 34.55 38.47 6.35 10.97 8.92 15.14 18.18 11.43 12.62 47.27 29.28 11 Italy 66.07 92.86 16.07 55.55 12.90 17.20 14.75 15.55 .00 25.60 27.40 .00 28.50 12 Netherlands 66.67 33.33 42.86 29.34 7.85 18.40 10.88 15.06 14.29 58.17 67.57 40.48 10.62 14 Spain 16.67 45.83 16.67 25.47 2.47 12.60 3.93 4.07 10.42 15.53 19.27 10.42 5.27 15 Tu rkey 90.20 .00 76.47 71.97 61.86 .00 33.91 45.40 13.73 67.16 67.50 27.45 90.05 16 United Kingdom 54.05 27.03 32.43 37.90 6.20 14.10 9.20 15.10 16.22 27.90 29.50 32.43 24.10 Views on Ethnic Diversity. How prominently is ethnic diversity perceived by the elites and to what degree are citizens willing to live in ethnically diversified neighbourhoods? Table 2 presents a combination of both views. The first column for each country represents the Eurosphere results, indicating that the representatives of national elites viewed ethnic groups (ethnicity) as a prominent factor, country A, b=EVS intolerance of foreign workers in country A, c=Eurosphere view of ethnicity in country B, d=EVS intolerance of foreign workers in country B). Both samples were tested for independence of percentages and standardised z-percentages. The percentages from Table 1 were used for testing as country scores and country standardised z-scores, i.e., as elements of one merged sample of 16 countries or 2 x 16 countries. Analyses were carried out using SPSS version 18. u S 5 0 £ * i * (R G C fN similar to general diversity. The relevance of this factor was reported by a majority of elite respondents in most countries - most notably by Turkish elites - but was less prominent in Central Europe and was only marginal in more cosmopolitan Spain/Catalonia and France. How do these elite views compare to what public opinion (in the EVS results) tells us about acceptance of various ethnicities? Columns 4 through 8 in Table 2 illustrate great variation in the willingness to live with ethnically diverse neighbours between each country. Roma were particularly ostracised: rejected by majority of citizens in Turkey, Italy, and the Czech Republic. Relationships between elite views on the relevance of ethnic groups and citizen xenophobia were analysed by examination of scatter plots and by statistical scrutiny. All five scatter plots (one for each relevant EVS category) followed a similar pattern, i.e., a positive correlation of the "elite opinion on ethnic group relevance" with citizen intolerance toward people of different race, toward immigrant and foreign workers, and to Roma, Jews and Muslims respectively. Hence, ethnic relevance was tied to intolerance. A modest increase in citizen intolerance was related to a steep increase in the relevance of ethnic groups judged by national elites. Pearson's R correlation coefficient expressed the relationship between relevance and unwillingness to share the neighbourhood with Roma (r=.752), with people of different race (r=.602), with immigrants/foreign workers (r=.592), and with a combined index of five diverse groups (r =.543). However, we have to take into account the low number of countries compared, and the fact that these coefficients tend to be falsely enhanced due to the existence of separate clusters and outliers - these serve to corrupt Pearson's correlation. Therefore, an additional statistical measure - conversion of raw scores - was employed. A statistically significant correlation was confirmed only for the relationship between "relevance of ethnic groups" (by elites) and "intolerance toward immigrants/foreign workers" (by citizens), depicted in Figure 1. Figure 1: Relevance of Ethnic Diversity and Intolerance of Immigrant and Foreign Workers. Scatter Plot (percent) Relevance of ethnic groups to defining diversity (by elites) and intolerance of immigrants/foreign workers (citizens) E £ 50 40 30 20 10 0 20 40 60 80 Relevance of ethnic groups (by elites) 100 The outliers stand out in the graph: Turkey on the top right (least tolerant citizens and most relevance), Spain on the lower left (most tolerant citizens, least CO relevance), whilst the third outlying country is Estonia in the middle, high above ^ the regression line (almost half of elites recognising relevance of ethnicity, and a third of citizens rejecting neighbouring immigrant workers). Table 3: Correlations and Significance of Non-parametric Tests between Euro-sphere and EVS Countries: Final Relevant Results (significant correlations supported by non-parametric tests are highlighted; Eurosphere N=725, variables starting with V1_; EVS N = 22,128 variables v47-v59) EVS and Eurosphere combined Left-wing extremists+ ideological diversity v48 and v1_8 Right-wing extremists+ ideological diversity v50 and v1_8 Foreign workers/immigrants+ non-European migrant diversity v54 and v1 10 Roma+non- European migrant diversity v59 and v1 10 Gays+ sexuality diversity v57 and v1_14 Immigrants/foreign workers+ ethnic diversity v54 and v1_7 Different race+ ethnic diversity v47 and v1_7 Jews+ religious diversity v58 and v1 13 Muslims+ religious diversity v53 and v1_13 Country scores Pearson's r x x x x .592 (x) .602 (x) .768 (x) x Country scores Eta .391 x .384 x x x x x (1) x (2) p of F<.01 Standardised zp-scores Eta x x (7) x .808 (4) .885 .312 x x (1) x (2) p of F<.01 Country scores Rho x x -.546 x (4) x x x x x p <.01 Standardised zp-scores Rho x x x x x x x x x p <.01 Sig. of country scores Mann-Whitney .056(a) .004(a) .050(a) .000(a) .305(a) .000(a) .000(a) .019(a) .000(a) H0 (same sample) refuted if p<.05 Sig. of standardised zp-scores Mann-Whitney 1.000(a) .106(a)(7) .000(a) .116(a) .000 (a)(5) .126(a) .126(a) .000(a) .001(a) H0 (same sample) refuted if p<.05 a) no ties x = not significant or false (i.e. not supported by n-par tests) correlation (1) for religious diversity (Jews) Eta without outlier (Tk) is .727 (.726 for standardised zp-scores); however n-par tests refuted dependence of both samples (sig. Mann-Whitney=.000 ) (2) for religious diversity (Muslims) Eta without Tk is .533 (.857 for standardised zp-scores); however dependence of both samples was refuted (Mann-Whitney sig.=.019 ) (3) for migrant groups (people of different race) Eta without outliers (Tk and It) is .554 (and for standardised zp-scores Eta=.924 without Tk and It); however both samples were independent (for country scores MannWhitney sig.=.013 and for standardised zp-scores sig.=.055) (4) for migrant groups (Roma) rho without outliers (Tk and It) is -.615 (and for standardised zp-scores Eta=.808 including Tk and It); and both samples were coming from the same underlying dimension (for country zp-scores Mann-Whitney sig.=.101 and including It and Tk sig=.116) (5) for sexuality diversity (gays) without outlier (Tk) intensity of relationship is very high (Eta=.938 for standardised zp-scores) and also similarity in pattern of zp-responses is very high (Mann-Whitney sig.=.178) (6) for ideological diversity (left-wing extremists) Eta without outlier (Tk and Nl ) is .234 (Spearman's rho not significant); samples were not independent (sig. Mann-Whitney=.27) (7) for ideological diversity (right-wing extremists) Eta without outlier (Tk and Nl ) is .444 (not significant Eta=.340 for standardised zp-scores) (Spearman's rho not significant for both scores without outliers); the same sample provenience confirmed for standardised zp-scores (sig. Mann-Whitney for scores=.034, for standardised zp-scores=.101 ) To refine the analysis, instead of raw scores (derived from original percentages), we worked with converted and z-transposed zp-scores (based on zp-percentages which control for individual bias, improve comparability and enable further statistical operations. Zp-calculations confirmed that the correlation between variables is statistically significant although rather low (Eta .312). See Table 3. Views on (Non-European) Migrant Diversity. How relevant do elites perceive migrant diversity to be, and to what degree are citizens willing to live in neighbourhoods with migrants? (Values depicted in Table 2, Columns 2, 4, 7, and 8). Table 2 demonstrates that representatives of national elites generally did not view migrant groups as a particularly relevant factor in social diversity (on average by country only 28 percent did so). One country was an exception, seeing migrant groups as extremely relevant: this was Italy, which was experiencing particularly strong waves of immigrants at the time. The opinions of citizens on coexistence with migrants were rather mixed. In the previous section we noted citizens' deep reluctance to share neighbourhoods with Roma. On average, almost 40 percent of respondents voiced their aversion to having a Roma neighbour. However, here the context was migrant groups, and Roma are now only partially migrant. In countries where coexistence with Roma is most problematic, i.e., in the post-communist countries, Roma were forced to settle down by the previous Communist regimes. Still, some Roma are ready to move from the former Communist countries and emigrate in search of better living conditions. Within Europe, once again, the favourite target for emigration is Italy. At the time of our Figure 2: Migrant Diversity and Intolerance toward Roma: Multidimensional Distance Model between Standardised and Transposed Zp-scores by ALSCAL. (Dimension 1: Decreasing relevance of migrant diversity from left to right, along with a decreasing positive difference from left to centre and an increasing negative difference from centre to right between elite and citizen opinions) study, a wave of Roma immigrants settling in Italy stirred political debates about the right of free movement within the current boundaries of the European Union. LO The most recent data proved that negative attitudes towards Roma neighbours in ^ Italy had increased, from 52 percent in 1999 to 60 percent in 2010. Whether due to the exodus of Roma, or to immigrants from Africa, Italy turned out to be an outlier in the context of Eurosphere data on migrant groups. Foreign workers and people of different race tend to be more readily accepted than Roma in all studied countries. Increased rejection rates toward Roma are obvious in Eastern Europe - in Turkey, Estonia, the Czech Republic and Bulgaria. Somewhat surprisingly, rejection rates of foreign workers and people of different race in Italy are not high despite the overwhelming recognition of the relevance of migrants by Italian elites. Is there a similar tendency of positive correlation between the opinions of elites and citizens apparent in the previous section on ethnic diversity? Curiously, the scatter plots suggested the opposite, a negative relation between (elite perceived) relevance of migrant groups and (citizen) intolerance of foreign workers, as well as of Roma or different races. To state in brief, migrant relevance correlated with tolerance towards these groups, intolerance correlated with irrelevance. Yet, there was an exception: the Italian data did not here match the general pattern of other countries. For the Italians, even in the context of migrant groups, undesirability (not desirability) correlated with relevance. Statistics including non-parametric tests partially confirmed this model (see Table 3, significant results for Roma (Eta for zp-scores=.808). Figure 2 presents a multidimensional distance model based on zp-scores (ALS-CAL). Elite views on the relevance of migrant diversity range from the lowest intensity in Hungary (only 9.1 percent elite relevance alongside 38.5 percent citizen intolerance) to the highest in Italy (92.9 percent of elite relevance alongside 52.9 percent citizen intolerance). Several meaningful clusters of countries can be identified based on the intensity of opinion scores, as well as on the probability of congruence of elite and citizen views. The upper right corner hosts Bulgaria, Finland, the Czech Republic and Hungary. The odds of citizen views to elite views are between 1.9 and 6.9, and the consequent probability of congruence is 14-52 percent; the intensity of elite opinions is distinctly below that of citizen views, by 23-47 p.p. difference. The United Kingdom, the Netherlands, and Norway in the upper left segment attain equal to high probabilities of congruence 50-88 percent, and the intensity of elite opinion is slightly below that of citizens (by 4-12 p.p.). Another cluster is formed by France, Denmark, and Estonia (on the bottom right) where the opinions of both the elites and the citizens have a low intensity and variable difference (4-25 p.p.) Austria and Germany (bottom centre) also have a very low difference (6-12 p.p.) and probability of congruence (71-81 percent). Views on Religious Diversity. Elites expressed considerable variation in their assessment of the extent to which religious groups are relevant to national diversity. More traditional Turkey took the lead with as many as three quarters of elite respondents recognising the relevance of religious groups, while at the other end of the spectrum are secular Czechs, as well as the French, and the Catholic countries Italy and Spain. There is probably not a great deal of religious diversity in these societies - see Table 2, Columns 3, 5 and 6. u s 5 0 £ * 1 * (R O C (¡B Citizens also varied in their acceptance/rejection of various religions in their neighbourhoods. Perhaps most striking is the intensity with which Turks distance themselves from the Jews while, obviously, embracing fellow Muslims - only the French could compete with their level of pro Muslim embrace. On the other hand, Muslims were accepted with a lot of caution in Central Europe (Estonia, the Czech Republic and Austria). Note that Turks were not asked about Muslims in EVS wave 3, hence we do not have data in that category. The data pattern tends to be inconsistent, implying that there is no clear relationship between the opinion of the elite and public opinion. The Pearson coefficients suggest there might be a positive relationship between religious relevance and intolerance of Muslims (.200, sig.=.492), and of Jews (.768 and even higher without outlying Turkey), however, these correlations are falsely enhanced and not corroborated by statistical scrutiny. Neither ANOVA, Spearman's rho, nor correlations with transformed country scores (and especially non-parametric rank-order analyses), could confirm the statistical significance of the correlations cited above. Ideological Diversity. The representatives of national elites generally did not see ideological groups as significantly contributing to national diversity (see Table 2, columns 9, 10, and 11). In Italy, France and the Czech Republic, the relevance of ideological groups was reported as being negligible, while understandably rather different views were expressed by elites in Germany, Austria, Bulgaria and Estonia where sensitivity to ideology was high, as was citizen vigilance against right-wing extremism (except in Estonia). This vigilance was also shared by most citizens of Turkey and the Netherlands. In general, there was more apprehension about right-wing rather than left-wing extremism, except in East European post-communist countries (and in Turkey), which signalled that the spectre of communism still exists. A relationship was found between the elite perceived ideological relevance and public intolerance of left-wing extremism. Furthermore, a comparable intensity of relevance and intolerance was observed in Austria, Estonia and Germany (Figure 3) and prevalence of elite views of relevance over citizen intolerance in Hungary and Denmark. The rest display a more common prevalence of citizen intolerance (including uncor-related outliers Turkey, the Czech Republic, the Netherlands). Figure 3: Relevance of Ideological Diversity and Intolerance toward Left-wing Extremists (country scores and percent) I Relevance of Ideologica l diversity (v1_8) □ Public intolerance toward Left-wing extremists (v48) 100 80 60 40 20 Sexual Diversity. On average, less than a third of the national elites of the respective EU countries indicated that sexually diverse groups play an important role. There were few "average" voices since the elites had a tendency to report ^ either a considerable or negligible relevance to sexual groups. On the other hand, citizens' voices were much more varied: from an embracing attitude toward gay people (single digit rejections in Denmark, Spain, France and Belgium) to an almost total rejection, with hardly any gay-friendly neighbourhoods, in Turkey (see Table 2, Columns 12 and 13). Pearson's coefficient approximates to zero because of outliers (especially Turkey), and because countries appear to form several clusters. The relationship could correspond more to a curvilinear rather than a linear function. Low rejections of gay people (i.e., high acceptance) seem to be related to both high and low relevance of sexuality groups, while at the same time, high rejection seems to be associated with medium levels of relevance. Four distinct country categories can be identified (see Table 4). The highest positions on the rejection scale are held by three Eastern European countries (Turkey, Bulgaria and Estonia) where gay people are rejected by approx. 50 percent citizens or more. Even so, gay people in these countries have relatively high relevance. Five countries can be characterised by their similar level of relevance but high acceptance of gay people (Belgium, the Netherlands, Finland, Germany and UK). The remaining two groups embrace gay people but differ dramatically in the gay-groups relevance rating. The data suggest that in Denmark, Austria and possibly Hungary gay groups are very intensely relevant (possibly very active within the public sphere as LGBT representatives). In contrast, we have France, Spain, the Czech Republic, Norway and Italy, where gays are embraced with low relevance; one can assume that they are accepted and integrated, having attained most of their rights. Being gay, then, is as normal as having a different colour of eyes; it is not a political issue in these countries. Table 4: Sexual Diversity - Relevance and Intolerance: Main Categories RELEVANCE OF SEXUALITY GROUPS VERY LOW RELEVANCE LOW RELEVANCE HIGH RELEVANCE VERY HIGH RELEVANCE GAYS ACCEPTED France, Spain, Italy Czech Republic -- Belgium, Germany, Netherlands, UK Denmark, Austria (Hungary) GAYS UNDESIRABLE -- -- Turkey, Bulgaria, Estonia -- Statistical scrutiny included conversion to standardised zp-scores. Subsequent correlations of variables became significant (Eta=.885 and even higher without outlying Turkey and rho=.554) and this was also supported by a non-parametric test. As illustrated by Figure 4, the pattern of relationship between zp-scores appears as a complex combination of ordinal (linear line) and nominal (undulating value) variables. While citizens' intolerance decreases, elite views of diversity fluctuate close to the citizens' views. The shape of the relationship may be a combination of linear and polynomial curves. Figure 4: Sexual Diversity - Relevance and Intolerance: Scatter plot of Zp-scores Discussion Our study drew upon two surveys distinct in their approach and methodologies. Combining disparate resources is inspiring, but the treatment of the data is challenging. Our solution to complexity was to analyse the data both from a qualitative point of view (utilising graphic layouts of histograms and scatter plots, most of which could not be included here for lack of space), as well as in taking a meticulous approach to statistical data treatment (z-standardisation, using both parametric and non-parametric methods). Non-standardised and standardised data served as sources of complementary insights into the patterns of survey results, each highlighting different aspects of a complex picture. The theme has certainly not been exhausted, but to keep this article concise we did not extend the analyses - to socio-demographic details, for example. Rather, we presented the initial direction and approach taken in our procedures. However, we also plan to extract "elites" from the citizen samples, and to compare the views of elites and non-elites. We did not devote any special attention to the possible effects of social desir-y ability and political correctness. We can assume that they skewed elite and citizen responses in opposite directions: sophisticated elites, living mostly in state capitals, might have projected more relevance to minorities than they objectively should have done, while at the same time, representative samples of citizens might have been hesitant to reveal all their prejudices, and they might have downplayed the diversity factor. This could have influenced some of the observed discrepancies or concord between the views of elites and citizens. The original survey questions for elites and citizens differed from one another in addition to their contexts, preventing us from drawing shortcut conclusions about the level of concord. Still, our results confirm the assumption of cross-national differences, and the division of public sphere, where elites and citizens stress different aspects of diversity, and have different worries and aspirations.2 This comparative study has a limitation; namely, the imperfect fit of the categories which were compared. For example, ideological diversity is not exhausted merely by left and right-wing extremism, and religious diversity is certainly not exhausted by questions relating only to Jews and Muslims. We also did not explore all possible relationships between these views. Some ^ meaningful relationships that we detected were not incorporated into our analy- c > ses: for example, religious relevance (assessed by elites) correlated not just with citizen intolerance of diverse religions, but also with acceptance or rejection of C^ gay people. ^ The Eurosphere questions for elites were complex, and we assumed that elites are used to dealing with complex terminology. Something could have been lost in the transfer from the elites to our interviewers, but they can also be categorised as elites by their rank, so hopefully corruption was minimised. The key concept of "relevance to social diversity" is complex, having at the very least a dual meaning: it contains both an aspect of diversification (which groups make society diverse and fragmented?) as well as an aspect of inclusion (which groups managed to have their voices heard?). Further analyses should recognise this twofold aspect. Still, overall, the Eurosphere and EVS data appear to provide a good knowledge base, and our analyses proved that their data differentiate significantly along the important variables which we were to study. Both resources - the elites as well as the citizen samples - expressed a variety of opinions to allow us to study diversity across a very wide range: from diversity as a challenge (the prevalent view of the elites) to an ordinary citizen's view, which includes irrational phobias alongside legitimate fears. Conclusions Our study presents two perspectives on diversity: a concrete one, expressed by citizens who were asked to consider having a diverse neighbour, and a more abstract viewpoint expressed by national elites. We gathered, arranged and analysed empirical data with a particular attention to the relationship between views expressed by elites and citizens, and to their agreements and incongruities. One example of this is the fact that in both the Netherlands and Turkey, elites do not see ideological diversity groups as particularly relevant, but the citizens of both these countries tend to intensely ostracise left- and right wing extremists. We also identified patterns or typologies of diversity according to the distribution of data. In some cases we could also detect gradient of relationships: for example, in case of ethnic diversity, where a modest increase in citizens' intolerance was related to a steep increase in the relevance of ethnic groups, as judged by national elites. Standardisation of scores enabled us to study the level of agreement of elites and citizens in individual countries, as well as projection of clusters of dis/agreeing countries. Hypothetical Patterns of Diversity. Four main patterns of diversity between social relevance and rejection/acceptance were observed: A. Positive correlation of rejection with relevance and acceptance with irrelevance. This applied especially in the contexts of ethnic, ideological and possibly religious diversity. For example, ethnic diversity could be recognised most clearly in countries where minorities were most rejected. Conversely, minorities that were most accepted appeared to be the least relevant in the ethnic diversity context. A typical example of this is the case of immigrants/foreign workers in Europe. Well-integrated minorities do not form political pressure groups; conversely, a high level of fear in citizens may be associated with high publicity of ethnic crime. B. Reverse relationship: Correlation of acceptance with relevance and rejection with irrelevance. This pattern was typically observed in migrant diversity: diverse soci- ü o C > II o eties where migrant workers are most accepted also accord the highest relevance to migrant groups. This model may have its limit if the number of migrants rises above a certain threshold, and the embracing attitude may be replaced by increasing fear. C. Model of converging/diverging perspectives. Disparities between public and elite views seem to mutually influence each other in a converging process. For example, public intolerance of minorities seems to "push" elites toward recognition of the higher relevance of the problem. On the other hand, the recognition of relevance by the elites appears to push the public towards greater tolerance. A combination of converging and diverging perspectives may be witnessed, for example in the case of gay people: in liberal societies they tend to be accepted by the public, but the elites tend to differ in their views on relevance. Gay people are either recognised as a highly significant minority (and LGBT activists participate in the society) or sexual orientation is not viewed as an issue at all (being socially irrelevant), since gay people have equal social rights and do not need to be accorded any special status. Congruence and Clashes between Citizen and Elite Voices. Our analysis of congruence between the voices of the elites and citizens was particularly focused on the intensity and constellations of citizen/elite opinions. To ensure maximum comparability, significance was tested with standardised zp-scores. We found different communication models: a) Balanced, with relative equilibrium between the views of elites and citizens (e.g., elite views of the significance of ideological diversity and citizens' intolerance of left-wing extremists). b) Imbalanced, with prevalence of citizen voices (public opinion) or prevalence of elite voices (e.g., the citizens had a more intense opinion than elites when they voiced their attitudes to Roma; at other times, the elites were more vocal than the public opinion, e.g., about the relevance of sexuality groupings). The imbalanced model was more common. From a wider perspective, it may appear as if public opinion puts pressure on £ the elites, leading them to attempt to push through their view of differentiation (e.g., coexistence with Roma). Conversely, the elites' views were more pronounced than those of citizens,' as if the elites were educating their fellows toward an inclusive tolerance and the embracing of minorities (e.g., about the significance of the equal voice of gay people). These relationships and models may be helpful in the further study of diversity, diversification and integration within the EU, and for enlightening the path towards European democratic citizenship. Acknowledgement The authors wish to thank Eurosphere www.eurospheres.org for their universal support, the European Commission for funding WP6 and the European Values Study for making their data available at http://www.europeanvaluesstudy.eu/. This study was conceived as a part of the research plan of the Academy of Sciences of the Czech Republic: Human Beings in the Context of a Life Long Development, AV 0Z 70250504. Notes: 1. The United Kingdom, Italy, Turkey and Norway have not made their data available at that time, in q their case we used data from wave three (except Norway which did not take part in the 3rd wave). t—I 2. This can be illustrated by many examples, e.g., interesting discrepancies in religious sphere. The Danish elites refer to religious groups as very relevant for social diversity, yet a mere 11 percent of Danes express intolerance towards Muslims. Conversely, the Czech elites are relatively disregarding of the relevance of religion; yet as many as a third of Czech citizens would prefer not to have a Muslim neighbour. References: Almond, Gabriel Abraham. 1960. The American People and Foreign Policy. New York: Praeger. Baubock, Reiner. 2008. Eurosphere Working Paper Series, Online Working Paper No. 6: Beyond Culturalism and Statism. Liberal Responses to Diversity. Devine, Philip E. 1996. Human Diversity and the Culture Wars: A Philosophical Perspective on Contemporary Cultural Conflict. Westport, Connecticut: Praeger. Diamond, Larry. 1999. Developing Democracy: Toward Consolidation. Baltimore: The Johns Hopkins University Press. Diez-Medrano, Juan. 2003. Ways of Seeing European Integration: Germany, Great Britain, and Spain. In: M. Berezin and M. Schain (eds.), Europe without Borders, 169-196. Baltimore, MD: Johns Hopkins U Press. Gerritsen, Debby and Marcel Lubbers. 2010. Unknown is Unloved? Diversity and Inter-population. European Union Politics 11, 2, 267-287. Hooghe, Marc, Tim Reeskens, Dietlind Stolle and Ann Trappers. 2009. Ethnic Diversity and Generalized Trust in Europe: A Cross-National Multilevel Study. Comparative Political Studies 42, 2, 198-223. Huszka, Beata. 2009. Eurosphere Working Paper Series, Online Working Paper No. 28: Framing of Diversity and the EU by the Hungarian Print Media. Knobloch, Katherine. 2011. Public Sphere Alienation: A Model for Analysis and Critique. Javnost-The Public 18, 4, 21-38. Margolis, Michael and Gary A. Mauser. 1989. Public Opinion as a Dependent Variable: A Framework for Analysis. Political Communication and Persuasion 6, 2, 87-108. McFarland, Sam G. 2011. Presidential Address: The Slow Creation of Humanity. Political Psychology 32, 1, 1-20. Milliken, Eveline J. 2010. Toward Cultural Safety: An Exploration of the Concept for Social Work Education with Canadian Aboriginal Peoples. Dissertation Abstracts International, A: The Humanities and Social Sciences 71, 1, 3-34. Mills, C. Wright. 1965. The Power Elite. Oxford: Oxford University Press. Mokre, Monika and Helene Pristed Nielsen. 2010. Eurosphere Working Paper Series, Online Working Paper No. 30: Interrelations between Gender, Diversity and European Integration: Perceptions among NGOs and Political Parties. Nielsen, Helene Pristed. 2010. Eurosphere Working Paper Series. Online Working Paper No. 31. A Common Cause? Anti-racist and Gender Equality Activists in Europe. Nissen, Sylke. 2003. Who Wants Enlargement of the EU? Support for Enlargement among Elites and Citizens in the European Union. Sociologicky casopis 39, 6, 757-772. Papadopoulos, Yannis. 1995. Analysis of Functions and Dysfunctions of Direct Democracy: Top- down and Bottom-up Perspectives. Politics & Society 23, 4, 421-448. Putnam, Robert. 2007. E Pluribus Unum: Diversity and Community in the Twenty-first Century. The 2006 Johan Skytte Prise Lecture. Scandinavian Political Studies 30, 2,137-174. Risse, Thomas. 2010. A Community of Europeans? Transnational Identities and Public Spheres. Ithaca: Cornell University Press. Sata, Robert. 2010. Eurosphere Working Paper Series, Online Working Paper No. 29: The Boundaries of Diversity and the Extent of Europeanization - the Case of Hungary. Sicakkan, Hakan Gurcan. 2008. Eurosphere Research Notes, research note 9: A Proposal on Guidelines for Selection of Actors and Respondents. Sicakkan, Hakan Gurcan. 2009. Methodological Issues of Comparative Research on European Media Public Sphere - Experiences from EUROSPHERE. Unpublished Lecture Manuscript at Eurosphere PhD Course at Tallinn University. Sim, Biirte. 2009. Eurosphere Working Paper Series, Online Working Paper No. 17: Gender and Diversity in the European Public Spheres. Smelser, Neil J. and Jeffrey C. Alexander. 1999. Diversity and Its Discontents: Cultural Conflict and Common Ground in Contemporary American Society. Princeton, N.J.: Princeton University Press. Splichal, Slavko. 2002. Principles of Publicity and Press Freedom. Lanham, Md: Rowman & Littlefield. Statham, Paul. 2010. What kind of Europeanised Public Politics? In R. Koopmans and P. Statham (eds.), Making of a European Public Sphere. Media Discourse and Political Contention, 277-306. Cambridge, UK: Cambridge University Press. Yates, Joshua J. 2001. Debunking Diversity's Discontents. Hedgehog Review 3, 1, 106-116. S I 0 £ * 1 * (R G C « fN