Measuring Voter Registration and Turnout in Surveys: Do Official Government Records Yield More Accurate Assessments?
50 Pages Posted: 1 Aug 2011 Last revised: 21 Aug 2011
Date Written: 2011
In post-election surveys, the proportion of respondents who claim to have voted is routinely greater than officially reported turnout rates, which has often been attributed to respondent lying. Such concerns have led researchers to prefer gauging respondents’ turnout by using government records of their behavior (called “validated turnout measures”). Using data from an online survey of a nationally representative sample of American adults, we found that this latter method did produce lower apparent rates of registration and turnout than did selfreports. However, the assessments based on government records appear to be less accurate in describing each respondent than their self–reports. Results based on government records appear be distorted by two countervailing biases: (1) a “downward bias” resulting from failures to roperly match some survey respondents to their government records, and (2) an “upward bias” that occurs because survey respondents turn out to vote at a higher rate than non-respondents and report their behaviors accurately. The downward bias is greater than the upward bias, which causes the illusion that aggregate turnout rates become more accurate when generated using government records. Thus, the use of government records appears to reduce the accuracy of individual-level registration and turnout measurements rather than improving validity.
Keywords: voting, turnout, elections, surveys, lying
JEL Classification: C80, C42
Suggested Citation: Suggested Citation