Evaluation of survey effects in pre-election polls

Pre-election polls can suffer from survey effects. For instance, individuals taking part in the poll may become more aware of the upcoming election so that they become more inclined to vote. Such effects cause biases in forecasted outcomes of elections. We propose a simple methodology that takes such survey effects explicitly into account when translating poll results into election outcomes. By collecting data both before and after the election, the survey effects can be estimated and used as correction factors in later polls. We illustrate our method by means of a field study with data collected before and after the 2007 regional elections (for `Provincial States') in the Netherlands. Our study provides empirical evidence of significant positive survey effects with respect to voter participation, and this effect is the largest for left-wing voters. That is, surveys seem to motivate left-wing people who otherwise would not have participated in the elections. This means that both the voter turnout and the number of seats going to left-wing parties may be overestimated by pre-election polls that do not correct for survey effects.