Revisiting Spatial Voting Models with Big Data

Example VAA output

Abstract

Issue preferences constitute the leading explanation for vote choice. However, while there is general agreement among election scholars that issues matter, how issues matter remains disputed. This project revisits the debate between the two most prominent spatial voting theories – proximity and directional voting theory – using a novel and innovative data source: Voting Advice Applications (VAAs). A core limitation of prior studies of spatial voting has been reliance on highly limited and biased measures of issue linkage between voters and parties. VAAs – a new kind of online voter information tool that enables citizens to learn about their ideological congruence with competing political parties or candidates – offer vast improvements over previously used data sources. First, VAAs allow to measure mass-elite issue linkage across large numbers of policy issues, and thus to eschew strong a priori assumptions about political dimensionality, such as that a single left-right dimension structures the political space. Second, VAAs enable direct comparisons of policy preferences between voters and elites on the same scale. Third, party positions are objectively coded in VAAs and therefore exogenous to voter preferences. And, finally, VAAs query preferences regarding concrete political issues rather than broad dimensions, thus increasing the probability that issues are understood the same way. This project aims to harness the potential of VAA-generated data to improve empirical tests of proximity and directional voting theory.