Does Measurement Affect the Gender Gap in Political Partisanship?

Abstract

Abstract There is a partisan gender gap in American politics: women are more likely than men to identify as Democrats. Relying on an experiment that randomly assigns measures of partisanship, Burden (2008) finds that part of this gap can be explained by question-wording effects: when partisanship measures prime affect—rather than cognition—women are significantly more likely to identify as Republicans, significantly decreasing the partisan gender gap. Given recent changes in the salience of both gender and partisanship in American politics, we revisit the possibility that measurement can affect the size of the partisan gender gap. Using a pre-registered replication of Burden (2008), we do not find consistent evidence that changes in the wording of the partisanship question can significantly alter the size of the partisan gender gap. Contrary to Burden (2008), our data show that measures of partisanship that prime affect do not lead women to be more likely to identify as Republicans. Rather, these measures lead women of both parties to be more likely to identify as independent—a pattern that leaves the partisan gender gap in place, but suggests that both parties may be losing women.