Comparing Candidate Survey Experiments with Real-World Voting Behavior

• In Event: Exploring the Roots of Political Gender Inequality in South Korea and Japan

Abstract

Political scientists have examined characteristics of electoral candidates that are preferred by voters. A common practice is to use conjoint survey designs in which survey respondents are asked to choose one of a pair of hypothetical candidate profiles that consist of several characteristics such as age, gender, partisanship, policy preferences, and so on. In these experimental studies, survey respondents tend to prefer young and female hypothetical candidates. However, young and female candidates are substantially less likely to hold political offices. In this study, we focus on the case of Japan, where gender and age disparities among politicians are large, to disentangle the reasons for the discrepancy between experimental evidence and real-world elections. We conduct a conjoint survey experiment about candidates right after the July 2022 Upper House elections and compare responses to the experiment with the self-reported voting behavior in the election.

Authors

• Charles T. McClean, Yale University

• Yuki Shiraito, University of Michigan, Ann Arbor