Giving More Weight to Preferential Votes: Welcome or Superfluous Reform? The Case of the Local Elections in Flanders (Belgium)

Abstract

In response to a crisis of representative democracy in many Western countries, (local) governments have introduced instruments to circumvent political parties in order to establish more direct links between citizens and governments. One of these instruments is rendering electoral systems more personal, that is by giving more weight to preferential voting. Preferential voting is important since it constitutes a major element of the personal vote and it determines whether parties or voters are the main decision-makers in designating representatives. We have investigated, in relation to the local elections in Flanders (Belgium), in what kind of municipalities voters are most likely to cast a preferential vote, whether the electoral reform granting voters more power has had an effect, and if it has had an effect, in what kind of municipalities. We have put forward five groups of explanatory variables: socio-demographic, political, social capital, geographic and ballot form variables. Our analysis shows that variables from each group correlate significantly with the percentage of preferential votes, with population density and electronic voting as most important variables. A comparison between the 1994 and 2006 elections often yields the reverse picture: characteristics of municipalities that have a positive effect on the percentage of preferential votes cast have a negative impact on the evolution of preferential voting and vice versa (electronic voting being an exception). This results in the only obvious effects of the electoral reform being seen in urban municipalities, because elsewhere local politics was already to a large extent personalised by politicians being locally known. We could conclude that in these rural municipalities the electoral reform was superfluous.

Key Words:

• Elections
• political parties
• personal vote
• participation
• Belgium

Acknowledgements

An earlier version of this paper was presented at the ECPR General Conference in Potsdam (Germany), 10–12 September 2009, in a panel dealing with changes in participation and representation. We would like to thank the participants of this conference, as well as the three anonymous referees of this journal for their helpful comments.

Notes

In fact, the number of list votes per party is multiplied by the number of seats per party (as was done before) and then divided by three.

We would like to thank Guido Decoster and Tom Doesselaere for providing these data.

In our regression analyses, we have always checked for the degree of multicollinearity. This is the degree of covariation between the independent variables. As criterion, we have used the variance inflation factor (VIF). The values on this criterion are always considerable less than 10, which allows us to conclude that there are no problems concerning multicollinearity for our analyses.