Empirical Construction of the Political Spectrum of Germany

Wednesday, July 12, 2017
Melville Room (University of Glasgow)
Andranik Melik-Tangyan , Institute of Economic and Social Research (WSI) and Karlsruhe Institute of Technology

The data from the German voting advice application (VAA), the Wahl-O-Mat, are used to empirically construct and visualize the political spectrum of Germany. For this purpose, we consider the positions of 28 German parties on 38 policy issues declared before the 2013 federal election and associate the parties with the 38-dimensional vectors of their policy profiles. These vectors are used to define the party proximity in the party space. Principal component analysis (PCA) of the proximity matrix reveals that the parties constitute a thin ellipsoid whose two longest diameters cover 83.4% of the total variance. Reducing the model to just these two dimensions, a one-dimensional contiguous party ordering is found which is exactly the left–right axis rolled into a circumference, reflecting that the far-left and far-right ends of the political spectrum approach each other, though remaining somewhat  distant, resulting in the spectrum’s horseshoe-like shape. For comparison, contiguous party orderings are constructed using four other models, and the results are discussed. The 2013 parties’ policy representation capacity along the party orderings reveal latent trends that three years later (in 2016) start to manifest themselves in real politics.   

.   

Paper
  • Tangian 2015 WP_66.pdf (2.6 MB)