Experiments on the effects of opinion polls and implications for laws banning pre-election polling

Abstract

Many nations ban the release of pre-election poll results based on the assumption that voters will be adversely influenced by poll information. The AAPOR notes that there is no scientific evidence that voter decisions are influenced by media polls. This study uses survey experiments to assess if respondents might be influenced by a hypothetical candidate’s poll standing. It advances our understanding of poll effects by testing which type of people might be most responsive to information about poll standing. Results are consistent with a theory proposing that voters with weaker political preferences (those less politically engaged) may be more likely to support candidates who are leading in media polls. Although the experimental effects are substantial for some of the less politically engaged respondents, these people may be least likely to become aware of media poll information in a real world setting. The effect of poll information on candidate choices is likely to be limited for the electorate overall.