Priming Self-Reported Partisanship: Implications for Survey Design and Analysis

• In Event: Theoretical Implications of Empirical Models

Abstract

In many survey settings, researchers seek to estimate an average outcome conditional on membership in a particular subpopulation. We conduct two studies to examine how survey design features affect these conditional outcomes. In our first study, we implement a preregistered survey experiment to quantify how question ordering affects respondents’ willingness to identify as a Republican partisan and presidential approval conditional on self-identifying as a Republican. While our estimates are generally imprecise, the overall pattern of results is consistent with the hypotheses that the treatment increases the probability that a respondent identifies as a Republican and decreases the probability that self-reported Republicans approve of the president. In our second study, we show that an informational intervention reduces the probability that a respondent identifies as an in- dependent. Our analysis shows that seemingly minor features of survey design can affect self-reported subpopulation membership and average outcomes for these subpopulations.

Authors

• Brendan Pablo Montagnes, Emory University

• Zachary F. Peskowitz, Emory University

• Kaylyn Jackson Schiff, Emory University