If Election were Held Today. The temporal proximity-effect and the polls.

Abstract

This thesis tests Franklin & Weber's finding that national elections have a temporal proximity-effect. They find that when a first-order election is closer in time, voters in second-order elections behave more similar to voters in a first-order election. By investigating respondent behaviour in polls - which, for voters, are similar to second-order elections - this thesis tests Franklin and Weber's novel findings. By using Swedish poll-results over a long time the thesis investigates whether a shorter temporal distance to a national election make respondents more likely to respond to the polls with instrumental rather than sincere concerns. The thesis finds that vote sincerity (the tendency to vote for ones favourite party) is larger in the beginning and middle of an electoral cycle, and that that it drops closer to the next election. Contrary to Franklin & Weber's findings this temporal proximity-effect is only found in the time leading up to the election, but after the election it seems to disappear. Besides testing Franklin & Weber's findings on voter behaviour, the thesis also shines some light on respondent behaviour in polls. It shows that temporal proximity is an important factor that determines how respondents answers in opinion-polls, implying that polls far from an election will be systematically worse at predicting election results.