It's Not Only About Lists: Explaining Preference Voting in Belgium

Abstract

This article analyses the decision of Belgian voters to cast a preference vote on the occasion of the 2009 regional elections. And what appears is that preference votes could be given three meanings. First, preference voting appears to be a sophisticated voting behaviour more accessible to politically interested and involved voters. Less politically active voters more often limit themselves to marking their ballot on the top of the list without differentiating their support among candidates. Second, preference voting is very much a token of voter–candidate proximity. Voters are more likely to support candidates when they know one or several specific candidates directly or via the media. Finally, preference voting is also very much dependent on the structure of institutional incentives. The more influence a preference vote has on the process of intra-party seat allocation, the more likely voters are to make the effort. All in all, this article shows the diversity of motivations behind preference voting, and more importantly the different meanings it could take in elections.

Acknowledgements

The authors would like to thank the three anonymous reviewers as well as Mark Franklin for their comments and suggestions. They have helped to improve significantly the quality of this article. We also thank members of the partirep project for comments on earlier drafts, as well as for the commitment in the voter survey. Audrey André is FWO research fellow. She would like to thank the FWO for its support. The authors would like to thank the partirep consortium and the Belgian federal science policy (BelSpo) for their support.

Notes

The same reasoning holds for ethnic minorities. Citizens belong to an ethnic minority lack conventional political resources, but they are more likely to identify with and vote for candidates with a foreign origin (Berg & Bjorklund, 2011; Teney et al., 2010). However, ethnic minority voters are excluded from the analytical framework because their strong underrepresentation in our sample does not allow reliable statistical analyses.

The region of Brussels has been excluded from the partirep Election Study due to the cost of bilingual interviewers and the difficulty of conducting survey research in urban areas. The number of eligible voters in Brussels only amounts to 7% of the country's total, moreover. Detailed information on the survey can be found at www.partirep.eu.

Before the introduction of multiple preference votes in 1995, voters could only support a single candidate within the list they voted for.

In Flanders, the D'Hondt divisor is applied in each of the six districts for lists having passed the 5% threshold. Wallonia uses a more complex two-tiered system. Direct seats are first allocated in 13 lower-tier districts. Remainder seats are then allocated in five upper-tier districts to lists that have won at least two-thirds of the Hare quota in one of the lower-tier districts.

To be comprehensive, the list votes transferred to the highest-ranked effective candidates also comprise (half of) the ballots cast for substitute candidates only.

An analysis over time for the regional assemblies does not make sense as they are only directly elected since the 1993 state reform.

No multi-collinearity among the set of predictors can be observed: the mean variance inflation factor (VIF) of the resource model is as low as 1.14.

The party dummies capture the difference between large and small parties and as a consequence the effect of party magnitude logically disappears.

The only minor differences when adding district-specific fixed effects is the relatively lower impact of the degree of urbanization. The effect of the degree of urbanization in the fixed effects model still has the expected sign but no longer reaches statistical significance because some districts are more rural than others.