Rachel Porter, Sarah Treul and Maura McDonald (University of North Carolina)
Abstract: The record-high number of women who ran for the U.S. Congress during the midterm elections led many journalists to proclaim 2018 as another "Year of the Woman." Although not every female candidate was successful, this large number of women running for office provides the opportunity to advance our understanding of how candidate gender influences elections. Pairing an original dataset of text from over 1,500 congressional campaign websites with a structural topic model, we demonstrate that prior political experience, more so than gender, is the primary driver influencing how female candidates present themselves. We further show that when a female Democratic candidate runs in a primary election, male candidates in that same race are 30% more likely to take up women's issues in their own campaigns. Using a machine-learning model, we demonstrate that this issue uptake isn't simply "cheap talk" -- when men talk about women's issues their language is indistinguishable from that of female candidates. To tease out the causal mechanism between female candidate emergence and men's issue uptake, we implement a difference-in-difference approach.