Models of party competition often predict a convergence of party positions towards the mean or the median of the electorate’s distribution of policy preferences. However, this result contradicts most empirical findings of party configurations, in which two or more parties seldom present identical positions. Several attempts have been made to develop models that yield to a more realistic prediction of party positions. Two strands of research can be distinguished: The tradition of behavioral research relying on non-policy factors, and the tradition of spatial modeling, defining policy issues to be crucial for individual vote decisions. The aim of the dissertation project is to combine both strands of research within one game theoretic model of party competition. This will be achieved by considering group-specific weighting of policy dimensions, where the group membership is defined by behaviorist factors. This model will be applied to German federal elections covering the time span from 1980 to 2009 in order to analyze the co-evolution of the German electorate’s social structural composition and the policy offer of parties.
