Answer Scales in Voting Advice Applications

This paper focuses on one particular element of the design of voting advice applications (VAAs), which has received little attention, namely the answer scale that is adopted to measure users’ preferences and to position parties or candidates. In practice VAAs adopt a variety of formats, including a simple for/against dichotomy, a five-point agree/disagree scale, or a slider that enables one to distinguish between as many as one hundred scale positions. The key question that we address is an empirical one: does it make any difference what type of answer scale one adopts? Because many VAAs present their advice in a two-dimensional spatial model, we analyse how suitable different answer scales are for constructing the dimensions in such models. Furthermore, we run simulations to analyse if the use of a different answer scale would have resulted in a different advice. We end the paper with concrete advice for makers of VAAs.