Semi-automatic Selection of Questions (Statements) in Voting Advice Applications

The paper discuses how to automate (or partially automate) final selection of questions (statements) for a Voting Advice Application (VAA) out of all questions available to authors of the VAA (e.g., the authors want to use 35 questions and have available answers by the candidates to 80 questions). It is still, however, dependent on the input by the makers of the VAA. The proposed method is based on optimizing a) the differences (discrimination) between candidates (parties), b) the salience of the questions in the elections, and c) coverage of the space of the elections issues by the VAA. It is based on the method described in Škop (2015) and it was used for two national VAAs in Europe in 2016. The process of selection of the final statements for a VAA is one of the key issues in its preparation (as demonstrated for example by Walgrawe et al., 2009, or studied further by Lefevere and Walgrave, 2014). Van Camp et al. (2014) cover in their study nine countries and 26 different VAAs and summarize how the VAAs are made, how the questions in VAAs are constructed and selected. They come with several “rules” generally shared by the authors of the VAAs (these are no definitive obligatory rules for the authors of VAAs, also the authors of the article called them “rules of thumbs”). First, the VAA questions cover a large amount of issues. Second rule is that the questions are selected on the basis of their salience in the political debate. The questions in a VAA should be relevant, they should cover current affairs. Third, the questions should be related to an important political cleavage (i.e., economic left vs. right or conservative vs. progressive cleavage). Fourth rule states that the VAAs should look forward to the future rather than backward to the past. The VAAs questions should be thus more prospective than retrospective. Fifth rule is that the questions should discriminate among candidates or parties, some of the candidates or parties should have a positive answer to the question while others have a negative one. If we take the above “rules”, the proposed optimization method optimizes the application for the second, third and fifth rule given the first rule. Literature: Škop, Michal. 2015. “Measures of Differentiation among Candidates as an Indicator of the Quality of Voting Advice Applications.” KohoVolit.eu working paper. https://docs.google.com/document/d/1KVTso9cdwvCxhPC3bEp8rAEY_7v3gWvCzHNYVOjcMR8/pub Van Camp, Kirsten, Jonas Lefevere, and Stefaan Walgrave. "The content and formulation of statements in Voting Advice Applications." (2014): 11-32. Walgrave, Stefaan, Michiel Nuytemans, and Koen Pepermans. "Voting aid applications and the effect of statement selection." West European Politics 32, no. 6 (2009): 1161-1180.