Abstract: | Socially desirable responding may cause differences between preelection polls and election outcomes. Bias-free preelection polls are, however, important because they influence election outcomes. Benefiting from a unique opportunity, we experimentally controlled social desirability using the Crosswise Model in a poll prior to the German federal election in 2013. We found that due to strategic voting, outcome expectations were related to voting intentions; the election outcome was therefore likely distorted by social desirability bias in preelection polls. We recommend using indirect questioning techniques whenever a vote for at least one of the competing parties may be perceived as socially undesirable. |