Winning at the polls and in the polls: the incumbency advantage in surveys of U.S. house voters

The success of House incumbents at the polls is well known and has been studied extensively. This paper focuses on the incumbents' success in the polls: the support bestowed upon incumbents by survey respondents is substantially higher than that received from the voters. The incumbency advantage at the polls, estimated at about 10% in the most recent elections, is almost doubled when measured in the polls. The data, drawn from the 1982–1996 National Election Studies, show that respondents do not reward all winners; candidates elected to open seats have not benefitted from the kind of bounce consistently enjoyed by winning incumbents. In addition, the pattern of respondents' misreports appears to be inconsistent with earlier explanations based on instrument effects. Respondent bias should be accounted for in order to reach correct estimates of the incumbency advantage in individual-level data.