The Viability of Voting Advice Application Data for the Analysis of Coalition Formation

Modern coalition theories consider both the office and the policy motivation of political parties. As a consequence, data on parties’ strengths as well as on their positions is needed, when it comes to the application of these theories. While the first – the seat distribution – is known as soon as the electoral result is announced, the estimation of party positions is a more ambiguous and difficult undertaking. Different standard approaches like expert surveys or analyses of party manifestos are confronted with various problems: The question of how manifestos can be aggregated to party positions, for example, is highly disputed, and expert surveys simply are not available for many elections. Recently, Voting Advice Applications (VAAs) have been provided for many elections. While VAAs have originally been developed to compare parties’ and voters’ congruence, they also allow for analyses of disagreement between parties. Since VAA data can broadly be used and is easy to collect, it might be a good and efficient alternative to the above mentioned data. However, there is also some doubt if VAA data adequately reflects proximities between parties. Are the questions asked in VAAs those who matter for parties in coalition formation processes? Do the questions allow for sufficient depth to identify substantial differences? Are very concrete questions adequate to reflect general ideological conflicts? With regard to the potential but also to the doubt of the viability of VAA data for the analysis of coalition building, we contrast results of coalition formation models based on VAA data with expert survey based and manifesto based models. Our results show that, with few reservations, VAA data is appropriate to estimate the policy component of coalition theories.