Going Soft or Staying Soft: Have Identity Factors Become More Important Than Economic Rationale when Explaining Euroscepticism?

Abstract

The scholarly debate on explanatory factors of public opinion towards the EU has shifted from economic and utilitarian (hard) factors to also encompass identity/ affective (soft) factors. This study investigates the explanatory strengths of these two factors. It does so in the context of 12 long-standing EU member states at two time points (1994 and 2005), drawing on Eurobarometer data. Results from the multi-level analyses show that identity-based as well as utilitarian factors play a significant role in explaining Euroscepticism in both years. Furthermore, the explanatory power of hard factors is very stable across time. We conclude that, against expectations, soft factors did not explain more variance in 2005 than in 1994, but already played an important albeit neglected role in explaining Euroscepticism.

KeyWords:

• Euroscepticism
• economic
• identity
• soft and hard factors

Acknowledgements

We thank the Dutch Organization for Scientific Research (NWO) for funding this research as a part of their Conflict and Security theme. We acknowledge the support of the Amsterdam School of Communication Research (ASCoR). We would like to thank Rens Vliegenthart for contributing to this paper with his thoughts and criticism.

Notes

1. Our results are based on data from the 1994 and 2005 Eurobarometer. More information on the data selection can be found in the method section.

2. Whereas Euroscepticism is multidimensional (Boomgaarden et al. 2011), here we focus on a standard measure of EU support, namely perceived benefits of a country’s membership to the EU.

3. Data limitations led to a more diachronic approach than initially anticipated. Leaving out some of the EU countries or excluding some of the less important variables would not solve this problem. Hence, we base our conclusions on two time points (1994 and 2005). We argue that these time points are representative of a larger period, but we acknowledge that we cannot make claims about longitudinal developments on the basis of these data.

4. This decision substantially decreased our sample size; however, the variable we created gives a more exhaustive idea of real EU attitudes.

5. Scalar invariance tests were performed, and the measurements appeared equivalent over time. Greece and Ireland were outliers, but were kept in the analyses. Since we are interested in over time comparisons, this should not jeopardize the interpretation of these results.

6. Descriptive statistics are available from the author upon request.

7. For more information on GDP: http://epp.eurostat.ec.europa.eu/portal/page/portal/eurostat/home/

8. For more information on immigration figures: http://www.oecd.org/home/

9. Due to many missing values on this variable (14 per cent in 1994, 16 per cent in 2005 we used multiple regression imputation (with education level, age and gender as predictors), reducing the amount of ‘missings’ to 2 per cent in both years while maintaining the same effects in the analyses.

10. Age is included as age in years minus 16.

11. Education is measured in number of years of schooling.

12. The method we use does not control for the categorical nature of the dependent variable (five categories, not normally distributed). We chose the method because it provides variances on both levels, which is important to answer the main hypothesis. For reasons of construct validity, we performed a two-level multinomial logistic regression analysis. The interpretations of both methods yield the same conclusions (consult authors for exact results).

13. There were no problems of heteroskedasticity in the model.