Complexity in European Political Spaces: Exploring Dimensionality with Users

I estimate and describe the political spaces of 27 European countries in a Bayesian multidimensional item response theory (IRT) set-up drawing on data generated by the EUvox VAA prior to the European elections of 2014. Spatial descriptions of political actors are commonplace in political science and are the cornerstone of most matching algorithms used in VAA. Yet questions remain regarding the ideological structure of the belief systems of the mass public - is one dimension sufficient to describe this structure, do we need more dimensions, and what defines them? This paper attempts to address this question by estimating one-, two-, and three-dimensional IRT models by country. The results are illustrative in highlighting the heterogeneity in political spaces across Europe for a sub-group of the population. They are also of relevance for the design of VAA as the estimated item parameters allow us to distinguish which items aid in the ideological placement of users and which don't. These findings can be helpful in an endeavour to design shorter surveys without losing precision.