New CUNY scholarship on matching voters with their preferred parties – the future of democracy?

Could this be the future of democracy? Imagine deciding for which candidate or political party you should vote for in an election – what is the basis of your decision? Voting decisions are no an easy tasks, and the data show it. For example, in the 2016 presidential election a full 13 percent of voters made up their mind only in the last week prior to the election.

In recent years, so called Voting Advice Applications (VAAs) have been developed to better match voters with the party or candidate that best conforms to the policy positions a voter cares for the most. This is by no means trivial, since it requires a measurement of both voters’ and parties’ positions, as well as a way to weigh the relative importance of the issues.

A new paper, co-authored by CUNY scholar Javier Padilla and published in the Journal of Elections, Public Opinion and Parties proposes a novel method to achieve these tasks. The authors apply an adaptive algorithm that learns the configuration of the conceptual space from users’ answers. In other words, in answering questions about policy issues the algorithm is able to estimate the dimensionality of the policy space for a specific voter. Distances to parties and candidates as well as the relative importance of these distances can then be assessed on an individual bases rather than assumed from the outset. The algorithm adjusts itself only based on the answers of those users who claim that they are “very interested in politics”. “That way, the algorithm only learns from those users who are educated, politically informed and vote because of ideological considerations (issue voters).” (p.9).

Can such an algorithm be a valuable tool in a democracy? Previous research has shown that VAAs have an actual effect in real-world politics. Of course, their use presupposes a certain degree of political interest and political as well as technical sophistication on the part of voters. In a highly polarized environment, it might be questionable whether most voters actually cast their votes based on issues or rather because of certain traits of the candidates. It is, therefore, no co-incidence that VAAs have been mainly used in European countries that are generally characterized by elections in a less polarized environment and with more parties competing than in the US. Nevertheless, developing and providing tools that incite voters to think about issues rather than candidates may be in an important step towards more rationally grounded elections.