Political Matchmakers – How do the decision rules employed by Vote Advice Application Sites influence their advice?

Voter Advice Application (VAA) websites are an increasingly important feature of electoral campaigns, especially in established multi-party democracies. VAA’s seek to enhance issue voting by showing each user how ‘close’ or ‘similar’ individuals users’ stated personal issue positions with the issue positions of the parties. This paper highlights three aspects of VAA’s. First, the visual presentation or advice that VAA’s present to their users is critically dependent on the precise built-in Multi-Attribute-Utility-Decision (MAUD) rule to conceptualize issue voting that underlies a VAAs advices. This paper employs the logfiles of over 600,000 users of the Kieskompas.nl site during the 2010 election campaign in the Netherlands to simulate the advice that the site would have generated under alternative MAUD rules. The results show that different MAUDs result in strikingly different vote advices. Next we employ a representative multiwave panel survey of IntomartGfK among show that the simulation results hold for VAA users. Advices generated by VAAs based on different decision rules generate strikingly different advices. These advices have a huge impact on the vote.