Validating and Improving Voting Advice Applications: Estimating Party Positions Using Candidate Surveys

Locating political parties correctly regarding different policy issues is not just crucial for research on parties, party competition, and many similar fields but also for the electorate. For the latter, it has become more and more important as the relevance of voting advice applications (VAA) has increased and as their main usage is to compare citizens’ policy preferences to the offer of political parties. However, if party positions are not adequately assigned, citizens are provided with suboptimal information which decreases the citizens’ capacities to make rational electoral decision. VAA designers follow different approaches to determining party positions. In this paper, we look beyond most common sources like electoral manifestos and expert judgments by using surveys of electoral candidates to validate and improve VAAs. We argue that by using positions derived from candidate surveys we get the information by the source itself, but at the same time we overcome most of the disadvantages of the other methods. Using data for the 2014 European Parliament election both in Greece and Germany, we show that while positions taken from the VAAs and from the candidate surveys do match more often than not, we also find substantive differences and even opposing positions. Moreover, these occasional differences have already rather severe consequences looking at calculated overlaps between citizens and parties as well as representations of the political competition space and party system polarization. These differences seem to be more pronounced in Greece. We conclude that candidate surveys are indeed a valid additional source to validate and improve VAAs.