Expressive Responding and Citizen Competence

• In Event: Advances in Public Opinion Research Methods

Abstract

Survey respondents sometimes engage in partisan expressive responding: they express support for their party by choosing responses that are more partisan than their underlying beliefs. Does this mean that Americans know more about politics than they let on? This paper examines the relationship between expressive responding and citizen competence through a replication and meta-analysis of twelve previously published studies. In the average study's control condition, response accuracy was 0.70 on a 0 to 1 scale, and the average partisan difference was 0.20. In the average study's treatment condition the average partisan difference was 0.15, a decline of -0.05 (-25 percent). Accuracy increased to 0.72, a difference of 0.02 (3 percent). This means that especially in relative terms, expressive responding's effect on accuracy is smaller than its effect on partisan differences. This is true for two reasons. First, depending on how the question aligns with the respondent's partisanship, expressive responding can either increase or decrease response accuracy. Second, there are structural reasons. Expressing the two estimands in terms of one another shows that even if expressive responding only hid knowledge of uncongenial facts, the absolute effect on the proportion of correct responses would be half the size of the absolute effect on partisan differences. This means that for practical purposes, expressive responding has little effect on measured citizen competence.

Author

• Matthew H Graham, Temple University