The Effect of VAAs on Electoral Decision-Making: Evidence from a Large-Scale Experiment

Despite the steadily increasing number of VAA related studies in recent years, surprisingly little is known about how their use is affecting electoral behavior. To what extent and in which direction does the information displayed to VAA users affect their voting decision? Are there differences in the effects across different user groups? Do VAAs lead to more swing-voting or does it lead to strengthen already existing party ties? Does the use of VAAs lead to more moderate or more polarized voting decisions? For all of these crucial questions VAA research has been only able to provide inconclusive answers or mixed results and has often faced methodological problems. This research paper contributes to this line of research and aims to overcome methodological problems by using data from a large-scale experimental study conducted in Switzerland in 2019. The participants were drawn from a representative sample of 60,0000 randomly selected voters from the canton of Zurich, who participated in a pre- and a post-electoral survey. These participants were divided in a treatment and a control group. The treatment was that these participants were asked to additionally use a clone of the Swiss VAA smartvote before the election. This cloned version of smartvote provided all the information of the original website but allowed for individual and complete tracking of all participants (answers to the questionnaire and the received voting recommendations) and to link this data to the survey data. By combining an experimental design with the recruitment of participants out of large representative sample of voters our study allows for the first time for an unbiased analysis of a VAA’s causal effects on the electoral behavior of real voters. Based on this data our paper will address the following research questions: 1. How do VAAs affect the political congruence between voters and their vote choice? 2. Are there subgroup-specific differences regarding the VAA effects on vote choices? 3. Does the use of VAAs lead to more swing-voting or does it strengthen already existing party ties? 4. Does the use of VAAs lead to more moderate or more polarized votes? Answers to these questions are not only relevant for VAA research but also with regard to the millions of voters using VAAs every year for the over-all quality of electoral and democratic processes.