Do polls reflect opinions or do opinions reflect polls? The impact of political polling on voters' expectations, preferences, and behavior

Can political polls alter the choices voters make on election day? Prior research on cognitive consistency suggests they can. This article develops a set of hypotheses based on cognitive dissonance theory concerning the effects of exposure to the results of political polls on voters' expectations about the outcome of the election, attitudes toward the candidates, voting intentions, and choice. These hypotheses were tested during experiments conducted during the 1992 U.S. presidential election and the 1993 New York City mayoral election. The results demonstrate that political polls do alter voting behavior. Voters use political polls as a way to maintain or move to a state of cognitive consistency. Depending on which candidate voters expect to win as well as the candidate for whom they intend to vote, polls can have no effect, lead voters to change their expectations about who will win, or lead voters to actually change their preferences and their voting behavior. The results have important implications for public policy and for survey methodology.