Anatomy of an Election from a Gender Perspective

Abstract

This paper investigates how gender shapes online political communication during an electoral campaign by analyzing the 2025 Canadian federal election. Drawing on a novel dataset of 193,620 social media posts from 921 candidates, we employ transformer-based topic modeling and a linguistic gender axis to quantify communicative patterns. Our analysis reveals three key findings. First, a statistically robust stylistic difference exists between the discourse of male and female candidates. Second, this gendered linguistic gap is not merely an artifact of ideology; it persists systematically across all major political parties, demonstrating its robustness across partisan contexts. Third, thematic analysis shows that women more frequently employ community-action framing across a range of policy-related discussions, while men place greater emphasis on economic and industrial issues during the electoral period. While the magnitudes of these effects are shaped by Canada's electoral environment, the mechanisms we identify offer a transferable framework for measuring gendered discourse in other political contexts. By documenting these patterns at scale, this study provides a high-resolution portrait of gender's persistent role in contemporary political communication and a replicable methodological toolkit for its study.