One of the most heavily criticized aspects of VAA development is the method of selecting relevant issues to be included. I propose a methodology for issue selection based on the spatial theory of political preferences which operationalizes divisive political issues as vectors (normal lines) separating political entities with opposing views. Given its statistical nature it is essentially content-blind and can therefore be applied to virtually any political setting. I apply it to evaluate four VAAs using parliamentary votes in Poland, Czech Republic, Slovakia and Serbia. I measure how closely vectors of the selected votes fit a distribution of all the votes taken in the parliaments and therefore how well the VAA reflects the given political setting. In case of the Czech VAA, I suggest specific improvements in selection of political issues.
