Education and voter turnout revisited: Evidence from a Swedish twin sample with validated turnout data

Abstract

The association between education and voter turnout is well-established in almost a century of research. The causal status of this correlation, however, is still subject to debate. Results in the previous literature differ substantially, and this may reflect both methodological differences and heterogeneous effects across populations or types of elections. This study addresses the question using a discordant twin design and variance decomposition methods with validated turnout data for both first- and second-order elections in a large sample of Swedish twins, paired with population-wide sibling data. Results show that education does not have an effect on national electoral turnout, but does have an effect on turnout in the European elections. Furthermore, the association between education and turnout is shown to be affected by substantial genetic confounding, which leaves a non-trivial amount of bias even in sibling based designs. This underscores the importance of taking genetic confounding seriously in observational research.