Much ado about Nothing? Sources and Consequences of Vote Overreporting

Based on three sets of validated turnout data collected in the canton of Geneva, Switzerland, the purpose of our paper is threefold. First, we investigate into the sociodemographic and political sources of overreporting, and we take at a closer look at the role played by the timing of the interview. Second, we test whether vote overreporting differs across survey modes. Third, we raise the question whether and to what extent overreporting affects the determinants of turnout and, therefore, bias electoral participation research? Our validated turnout data was collected in three post-election surveys in 2011, 2012 and 2015, and it is highly reliable: In each case the initial sample was drawn from the official vote register, and vote validation occurred through an electronic reading of the voting card that voters had to sign and send back by postal mail or put in the ballot box. The three surveys differ in terms of survey mode, which will enable us to check for survey mode effects: The first was a single-mode telephone survey, the second was a mix of mainly telephone survey complemented by an internet survey, and the third was mainly an internet survey complemented by a telephone survey.