Data Analytic and Visualisation Tools for Voting Advice Applications

Parliamentary elections are one of the most important procedures in parliamentary democracies since they legitimate the government’s power by transferring the people’s sovereignty to the representatives in parliaments. The way a voter comes to a decision which party to vote for is therefore of exceeding importance for democracies. Voting advice applications (VAA) constitute an auspicious chance for eligible voters to check which political parties fit most to their own policy preferences. Having originally been developed for comparing a user’s policy positions with those of the parties, VAAs often indicate ranked lists of parties that represent congruence. Thereby, the user can reason that the party with the most similarities should be in his or her best interest. However, parliamentary elections often result in the formation of governmental coalitions since single-party majorities are of rareness. For this reason, voters have to consider not only the policy positions of single parties. They have to take supposable coalition policies into account – which parties are likely to move into parliament, which parties are likely to form a government, and, finally, which policies are likely to be implemented by the various coalitions. However, prior work on rational voting in multi-party systems has shown that voters can be overstrained when they have to consider expected policies of coalition governments. Consequently, they are potentially not able to identify which voting decision actually is in their best interest in spite of using a VAA, since coalition policies differ from the parties’ original positions given in the VAA. For this reason, a user’s rational choice which party to vote for does not necessarily correspond to the first-ranked party which ought to be preferred in terms of the VAA’s original aim, but rather corresponds to the party which is expected to minimize the distance between the user’s preferences and the expected coalition policies within the policy space. Multi-issue data embedded in VAAs constitute a new source that allows estimating which parties are close or distant to each other, and can be used for estimating coalition policies since it is generated close to elections and is proclaimed to include all relevant topics of an election. As the original aim of a VAA is to provide users with “voting advice” or “voting aid” but, at the same time, voters can be overstrained by taking possible coalition policies into account, we argue that there is need for data analytic and visualization tools for VAAs that support VAA users to find out their most preferred coalitions and to identify which choices are actually in their best interest in this regard. Thus, our contribution enables VAA users to apply the rational voter theorem to parliamentary elections. In our paper, we i) sketch the spatial modeling and the formal methods on which our data analytic and visualization tools are based, ii) describe the functionality of our tools and iii) illustrate their implications under various conditions at the example of the “Wahl-O-Mat”, Germany’s official and most widely used VAA provided for the German federal elections in 2017.