Centralized and Decentralized Personalization: Development of Indicators for Assessing the Concentration of Personal Votes in PR List Systems (Belgium 1995-2014)

Elections are discussed mostly in partisan terms. Yet, electoral reforms around the world have been increasingly giving weight to personal forms of voting underlying the intra-party dimension of elections (i.e. the internal competition between candidates on the same list). Therefore, the study of ‘the election within the election’ has become a growing research area in the broader context of personalization of politics. Scholars study the factors explaining why some candidates are electorally more successfully than others. They identified structural factors (e.g. party and district magnitudes or the list position) as well as factors related to the human and political capitals of candidates (e.g. political experience, gender, and background of immigrant origin). Recent research established a distinction between centralized personalization (a few political leaders concentrate most preferential votes on the list) and decentralized personalization (a larger number of candidates manage to attract personal votes). The implications of these various forms of personalization are critical for the functioning of representative democracies. Centralized personalization can serve political parties’ internal recruitment goals (e.g. greater diversity in terms of gender or immigrant background) thanks to powerful list pushers in a given district. On the opposite, decentralized personalization constrains political parties in their selection of candidates (e.g. profiles of candidates in line with the district’s median voter). The goal of this contribution is twofold: (1) developing indicators describing the concentration of preferential votes on a list; (2) assessing the factors explaining the (de)centralization of personal votes. We develop an empirical analysis based on 21.275 Belgian candidates for the 1999-2014 period at regional and federal elections (Flanders, Wallonia and Brussels). Our preliminary results validate the critical importance of factors already acknowledged in the literature (e.g. presence of ‘top-incumbents’, party and district magnitudes) but furthermore reveal the key effect of less-studied factors such as the districts’ geographic dispersion.