Are Voting Advice Applications Useful for Electoral Studies?

The aim of this article is to study whether Voting Advice Applications (VAAs) can become a reliable research tool to address two major objectives of electoral studies, namely (a) documenting the distribution of political variables of interest, and (b) establishing relationships between these variables and electoral choice. To answer this question, the authors examine, in the context of the Québec provincial election of 2012, the differences and similarities between a random-sample telephone survey, a non-random Internet panel and a VAA (the Vote Compass). The main lesson is that relationships between electoral choice and its determinants observed when using VAA data are very close to those observed when using Internet panel data. Generally, the same explanatory factors emerge from multivariate models where the effects are in the same direction and of the same magnitude. This result is all the more interesting and intriguing given that 1) the demographic profile of VAA respondents was less representative; 2) data from the VAA differ on most political attitudes examined; and 3) the electoral preferences of VAA users do not match the official results of the election. Moreover, sociodemographic weighting mitigates some of these biases, but does not make them disappear.