The online voting advice application (VAA) smartvote has been a constant companion of Swiss elections at various levels ever since its inception in 2003. Promoting political participation has been cited as a primary motivation for the development of the smartvote application. We undertake the first systematic assessment of smartvote’s effect on electoral turnout hoped-for by its developers. We overcome problems associated with the use of non-representative VAA user samples in earlier research (Fivaz & Nadig, 2010) in that we draw on data from the 2007 Swiss post-election studies. A fundamental problem in establishing the causal effect of VAA usage on electoral participation is non-random treatment assignment. To counter self-selection into the treatment, we use entropy matching (Hainmueller, 2012) so as to directly model the self-selection process. We consistently find a substantively strong and statistically significant effect on electoral participation. The effect withstands several robustness tests, including a placebo test.
