Abstract
More often than not, European identity is portrayed as one coherent identity for people across the whole continent. However, why should European identity not vary across countries? In this paper, we examine contents and structures of European identities in nine very different European countries. We draw on theoretical and normative approaches and differentiate between a political and a cultural component. In contrast to other studies, we consider national variation in the determination of the contents of both dimensions. Our findings suggest that European identities vary considerably across the continent. Certainly, there is no single picture of a European identity. A two-fold distinction between political and cultural elements prevails in some countries, whereas in other countries political and more pragmatic issues dominate the people's ideas about what it means to be European. Finally, we discuss some of the implications of these findings for policy makers at both the national and European level.
Notes
Moral, pragmatic and cosmopolitan issues add on to the variety of ‘top-down’ approaches towards European identity (Delanty, Citation2002).
This can be seen in the outcome of Bruter's (Citation2005) analyses. For instance, the different explanatory power of the civic identity factor ranges from 63% in the United Kingdom to 51% in France. Factor loadings, i.e. a statistical measure for the association between a manifest measure and the latent construct (civic European identity) vary considerably across countries and explain variation in the measurement to different degrees. For instance, the civic factor and its fifth indicator (‘Does being a “Citizen of the European Union” mean anything for you?’ if I am correctly interpreting his study) are differently related across the three countries. The latent factor explains 64% of the variation in the UK, whereas only 25% are explained in The Netherlands.
To separate ‘true’ differences from those implied by measurement makes it necessary to analyse data in every detail. That is, to control for as many conditions as possible in the production of the survey (e.g. translation issues), in the process of data gathering (e.g. sampling, interview situations) and finally, the analyses. With secondary data such as the Eurobarometer, we have no chance to rule out errors which occurred in the first two stages. What we can do, however, is to analyse the data and interpret the outcome.
Important criteria for evaluation of the robustness in confirmatory factor analysis are goodness-of-fit indicators such as the χ2/df-ratio, which should be 2 or smaller, RMSEA (root mean square error of association) of 0.05 or below and GFI (goodness-of-fit index) of 0.95 or higher. Models which fit these criteria can be regarded as good, though because of the strictly confirmatory character of the method they can also be described as rather ‘conservative’ in the sense that they react to smallest empirical deviations from a theoretically imposed model. We should thus work with rather ‘small’ models of identification with Europe because only a smaller selection of indicators is likely to be included in the final models.
Hence, a parsimonious model is used as the reference group, which has the statistical advantage of a limited degree of complexity. This makes it more likely to be validated with data from other countries.
A total χ2-value of 711.71 (df = 81) is highly significant and proves that factor correlations, factor loadings and measurement errors vary across countries. A RMSEA of 0.10 and a GFI of 0.91 also indicate a rather poor fit of the multi-group comparison model. We find highly significant co-efficients, which indicate the variability of meanings of identification with Europe. That is, comparing countries concludes that the indicators are differently related to the latent political and cultural dimensions of identification.