The Design Effects of Voting Advice Applications: Comparing Methods of Calculating Results

In election times more and more voters have turned for advice to so-called Voting Advice Applications (VAAs), such as Wahl-O-Mat (Germany), VoteSmart (Switzerland), or StemWijzer (The Netherlands) (Cedroni & Garzia, 2010). The advice given to voters appears to have a significant impact on their vote choice, especially among floating voters (Ladner et al., 2008; Walgrave et al., 2008; Ruusuvirta & Rosema, 2009). Hence it is important to study the effects of their design (Wagner & Ruusuvirta, 2009; Walgrave et al. 2009). This paper focuses on the spatial framework adopted to calculate the (mis-)match between voters and parties. In its simplest form this would mean that statements are used to position parties and voters on a single dimension of left-right and then calculate the match to reach an advice. In practice, VAAs are more complicated, as they adopt a two-dimensional (EU Profiler) or multi-dimensional space (SmartVote). Even VAAs that do not present their results in terms of a spatial framework (StemWijzer) in practice also adopt a particular spatial model for calculating voter-party distances underlying the advice. In this paper we examine to what extent the spatial framework adopted by VAAs affects the advice given to voters. We use the actual answers given by millions of users of one of the most popular VAAs in Europe, StemWijzer. We determine the advice that would result from adopting different spatial frameworks. Results indicate that the framework matters and a substantial number of voters would have been given another advice, had another framework been adopted.