Virtuous villages and sinful cities? A spatial analysis into the effect of community characteristics on abstention and invalid voting in local elections in Belgium 2006-2012
Faculty of Social Sciences. Political Sciences
Elections, Public Opinion & Parties conference, Lancaster (UK), 13-15 September 2013
The influence of the size of a community on politic al participation has only scarcely been researched since the famous work by Dahl and Tufte on the subject in 1973. In most cases a smaller community seems to be more propitious to hi gher levels of participation. The few researches that have been performed on the subject point at a stronger influence of the size of the place of residence compared to some of the much investigated personal characteristics and this warrants more attention to this subject. Using spatial analysis techniques, in this paper this question is investigated for electoral turnout at the local level using aggregate data from all Belgian municipalities in the 2006 and 2012 loc al elections. The paper looks at both the absolute level of the depend variables as their cha nge between the last two elections. The large dataset allows us to investigate the effects separately for turnout and invalid voting, as literature suggests these are not caused by exactly the same mechanisms. The results show that the size of a community measured by its popula tion has a clear effect on turnout and blank and invalid voting. The actual size of a comm unity in surface matters far less and this effect is even in the inverse direction as theory w ould predict. The results furthermore demonstrate that compulsory voting in Belgium seems to be able to keep young people showing up at the voting booth, but they have a hig her likelihood of spoiling that vote. The technical issue as to whether a municipality uses p aper and pencil ballots or computer voting (in a booth), seems to be one of the more influenti al community variables related both to turnout and blank and invalid voting. Finally a str ong regional effect is present for both dependent variables, with clear differences between the language communities in Belgium even after controlling for a host of other communit y variables.