ABSTRACT
This article investigates the link between attitude formation at the national and the supranational level of the European Union (EU). While the existing studies have provided strong evidence that attitudes towards national institutions fundamentally condition attitudes towards the EU, the mechanisms through which these spillovers occur are not clearly spelled out. Our main contribution is to theorize the complex ways in which the national politicization of the European integration process affects support for the EU by focusing on critical moments in the EU integration process and the electoral fortunes of the political parties doing the cuing. To test our theoretical claims, we employ multilevel models using six rounds of the European Social Survey combined with party-level data from Chapel Hill Expert Survey, and various country-level data. The analyses show that spillover effects are crucially conditioned by the level of politicization of European integration at the national level.
Acknowledgements
The authors would like to thanks the editors and the three anonymous JEPP reviewers for their useful comments and suggestions on earlier versions of this article.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes on contributors
Macarena Ares is a PhD researcher at the European University Institute.
Besir Ceka is an assistant professor of political science at Davidson College, USA.
Hanspeter Kriesi is the holder of the Stein-Rokkan Chair at the European University Institute.
Notes
1. For evidence suggesting that diffuse support for EU institutions has proven highly stable in the context of the Great Recession, see Ringlerova (Citation2015).
2. Because the CHES is conducted every four years, the three rounds of the CHES were matched to two rounds of the ESS each (CHES 2002 for ESS 2002 and 2004; CHES 2006 for ESS 2006 and 2008; and CHES 2010 for ESS 2010 and 2012).
3. There are some instances in which in the CHES two parties are coded as a single unit but the corresponding parties are treated separately in the ESS (and vice versa). In the first case, voters of the two parties were assigned the same CHES code. In the second, we assigned voters the different parties in the coalition based on the party to which they felt closest. Those who did not specify a party they felt close to were assigned the major party in the coalition.
4. There is no trend in this respect across the six rounds. The countries and rounds included in the analyses are: Austria (rounds 1 to 3); Belgium (rounds 1 to 6); Bulgaria (rounds 4 to 6); Czech Republic (rounds 4 to 6); Germany (rounds 1 to 6); Denmark (rounds 1 to 6); Estonia (rounds 3 to 6); Spain (rounds 1 to 6); Finland (rounds 1 to 6); France (rounds 1 to 6); United Kingdom (rounds 1 to 6); Greece (rounds 1 to 2 and 4 to 5); Hungary (rounds 2 to 6); Ireland (rounds 2 to 6); Italy (round 1); Lithuania (rounds 5 to 6); Latvia (round 4); the Netherlands (rounds 1 to 6); Poland (rounds 2 to 6); Portugal (rounds 1 to 6); Romania (round 4); Sweden (rounds 1 to 6); Slovenia (rounds 2 to 6); and Slovakia (rounds 2 to 6).
5. The trust index was constructed as respondents’ average trust on: the legal system; the police; politicians; and the United Nations. The satisfaction index was constructed as respondents’ average satisfaction with: his/her life as a whole; the state of education in his/her country; and the state of health services in his/her country. A factor analyses was conducted separately for each set of items, indicating that for each set all items strongly load on a single dimension, with only one eigenvalue above one. Results of the factor analyses are presented in Tables A.1 and A.2 in the Online Appendix. Cronbach’s α is of 0.804 for the trust syndrome and 0.643 for the satisfaction syndrome. Additional factor analyses were conducted separately by country and ESS round (not shown), they all return a one-factor solution.
6. The third aspect – the expansion of the actors to non-executive actors – does not make sense here, since we are only dealing with one type of actors, i.e., political parties.
7. A detailed description of how variables are coded is available in the Online Appendix.
8. The Czech Republic, which was the only other state to follow the British veto, later adopted the ‘Treaty on stability, co-ordination and governance’ which covers the Fiscal Compact.
9. Although the assumption of uncorrelated residuals is unlikely to hold – there probably are exogenous variables jointly influencing our endogenous variables – this procedure still allows us to get an approximate estimate of the indirect (and total) effects.