No participation without representation

ABSTRACT

How does the degree to which European citizens see themselves represented by political parties relate to their willingness to participate in an election? We argue that the closer citizens are to the parties running in an election in terms of the socio-economic left-right conflict and the European integration policy dimension, the more likely they participate in elections to the European Parliament. The analysis, which is based on the 2009 and 2014 European Election Studies datasets, indicates that the ideological left-right distance between citizens and parties mattered for participation in both EP elections under study. Moreover, the European integration dimension has become relevant for citizens’ turnout decision over the course of the European economic and sovereign debt crisis, but only in countries that are part of the Eurozone. These findings have implications for the nature of European elections and representative democracy on the European level, and indicate that missing policy responsiveness can lead to lower voter turnout.

KEYWORDS:

• Abstention
• European economic and sovereign debt crisis
• European Parliament elections
• policy distance
• political participation

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Constantin Schäfer is a PhD student at the Graduate School of Economic and Social Sciences (GESS) Mannheim and research associate at the Chair of Comparative Government at the University of Mannheim, Germany.

Marc Debus is a Professor of Comparative Government at the University of Mannheim, Germany.

ORCID

Constantin Schäfer http://orcid.org/0000-0002-0693-0888

Marc Debus http://orcid.org/0000-0002-7151-7942

Notes

1 Due to the EU accession of Croatia in 2013, the 2009 EES dataset only contains respondents from 27 EU member states. Since a robustness check has shown that the results do not substantially change once we exclude the Croatian respondents of the 2014 EES dataset (n = 1,074), we decided to keep them in the combined dataset (see also Table A4 in the online appendix).

2 Since we are not interested in country-level determinants of turnout and because of the ongoing debate on how many observations on the upper level are necessary to perform multi-level models (e.g., Stegmueller 2013), we refrain from evaluating the hypotheses on the basis of multi-level models. Estimating such models deliver, however, very similar results (see Table A4 in the online appendix).

3 The number of coded parties per country in the EES datasets ranges between 5 and 14 in 2009 and 5 to 9 in 2014, respectively.

4 We follow Merrill (1995) and Krämer and Rattinger (1997) who argue and empirically demonstrate that the individually perceived distances between the respondent and the available parties are the best predictor for proximity-based decision-making.

5 This notion refers to the average position of all respondents from a given country.

6 The ‘First Post-Election Survey’ of the EES 2014 does not include a variable that provides information on the perceived position of parties by respondents on the European integration dimension. However, what is available from another dataset created by the EES study group (the ‘Second Post-Election Survey’) is the average position of relevant parties on the European integration dimension as perceived by the respondents. We combine both datasets to estimate the distance between respondents and the parties closest to them on the EU integration policy dimension.

7 A detailed overview on the operationalization of all variables is provided in Table A2 in the online appendix.

8 Tests for multicollinearity, e.g. calculating variance inflation factors and checking correlations between the independent variables, do not show problems when estimating the models. Note also that we cannot use the weighting factors provided by the original datasets because we merged the 2009 and 2014 EES survey datasets that have been accompanied with very different types of sample weights.

Additional information

Funding

This work was supported by Deutsche Forschungsgemeinschaft [grant number DE 1667/4-1].