A perfect match? The impact of statement selection on voting advice applications' ability to match voters and parties

This study examines how statement selection systematically affects the output of voting advice applications (VAAs). Does the statement selection influence how often voters are matched with parties that 'should be' close to them? Our benchmark is a classic account of issue voting, the proximity left-right model. We analyze the Belgian VAA Do the Vote Test and find that the output resembles the left-right model. When more left-right statements are included, more left-wing voters get the advice to vote for left-wing parties and the same is true on the right, while simultaneously advantaging parties with more extreme positions on this dimension. We also analyze issue saliency and find that parties are disadvantaged when more statements about salient issues are included. These findings imply tough choices for VAA builders.