Overreporting

In many surveys, respondents tend to report more socially desired behaviors than they actually performed. In addition to this type of misreporting—called “overreporting”—respondents are also inclined to understate that they have engaged in socially undesirable behaviors, which is called “underreporting.” Similar to underreporting, overreporting is assumed to be connected to social desirability bias and thus occurs on the cognitive editing stage of the question-answer process.

Among other topics, overreporting of voting and being registered to vote has been in the focus of methodological research for decades. Since respondents in national- and state-level and local election polls tend to overly state that they have voted in the election, voter turnout has traditionally been overestimated. Usually, overreporting is identified applying post-survey validations using record checks (like in the National Election Study).

Since not every survey can afford a cost-intensive validation study, several attempts have been made in order to reduce vote overreporting, either by softening the question wording so that respondents will not feel embarrassed to admit that they have not voted or by a set of preceding questions on voting behavior in other, prior elections. It was assumed that respondents [Page 561]would be more willing to admit that they have not voted in the most recent past election if they were able to report voting in previous elections. However, neither tactic succeeded—the proportion of vote over-reporting remained unaffected.

Overreporting is associated with respondent characteristics. Respondents who hold strong positive opinions on a particular behavior are more likely to falsely report this behavior in a survey.