How Second Order Are Local Elections? Voting Motives and Party Preferences in Belgian Municipal Elections

Abstract

A defining characteristic of second-order elections is that voters base their decision on considerations that were developed for a different policy level. Therefore, this kind of elections does not contribute to the quality of democratic representation. Municipal elections are often considered as second-order elections. In this article, we use data from an exit poll (n = 4,591) held during the 2012 municipal elections in Belgium. Results suggest that although voters predominantly invoke local aspects as determining their vote choice, still three-quarters votes for the same party locally as for federal elections. Among voters who deviate from their federal party preference, knowing local candidates and concern about local policy issues are the main sources of deviation. The conclusion therefore is that local candidates do make a difference and contribute strongly to the salience of electoral decisions on the local level.

Key Words:

• Second-order election
• local elections
• Belgium
• exit poll
• vote choice heuristics

Notes

1. Interviewers were at their assigned polling station for the full period polling stations were opened. This implies they were interviewing from 8 am to 1 pm where paper ballots were used, to 3 pm where votes were cast electronically and to 4 pm in municipalities in the Brussels capital region.

2. Question wording for the federal vote intention was the following: ‘What party would you vote for if today federal elections were being held?’

3. Voters who indicated to have voted for a joint list of socialists and greens, for example, are only coded as deviating from their federal preference if they indicated neither the socialist nor the green party for their federal vote intention.

4. This rank order can be found in the Flemish and Walloon region and is in line with previous studies (Deschouwer 2012).

5. We compared the fit of the intercept-only model to the fit of an ordinary logistic regression that does not take into account random variation at the level of municipalities. This likelihood ratio statistic is 19.63 and has a p-value (for one degree of freedom) of 0.000. Consequently, we have to reject the null hypothesis of no between-municipality variation in deviant voting and therefore we use multilevel modelling techniques.

Additional information

Notes on contributors

Sofie Marien

Sofie Marien is an assistant professor at the University of Amsterdam, the Netherlands. Her research and teaching focus on political trust, political participation, political communication and research methods. Previously, her work has been published in the European Journal of Political Research, European Sociological Review, Electoral Studies and Intelligence.

Ruth Dassonneville

Ruth Dassonneville is a PhD student at the Research Foundation Flanders (FWO) of the University of Leuven, Belgium, where she has written a PhD dissertation on electoral volatility. Her main research interests are voting behaviour, economic voting, election forecasting and longitudinal analyses. Her work has been published in Electoral Studies, European Journal of Political Research, Party Politics, Political Studies and Political Science Research and Methods.

Marc Hooghe

Marc Hooghe is professor of political science at the University of Leuven, Belgium. He has published on social capital, political participation and social cohesion and currently has an ERC Advanced Grant to investigate the democratic linkage between citizens and the state in Western Europe. His work has been published in the British Journal of Political Science, Electoral Studies, Party Politics, International Political Science Review, European Journal of Political Research and Political Behavior.