Hillygus, D SunshineLaChapelle, Christina2023-03-282022https://hdl.handle.net/10161/26846<p>In the study of politics and campaigns, scholars have focused on television as the primary medium for advertising. But recent years have seen political candidates turn to internet and social media platforms, which offer different opportunities and constraints for their campaigns. How do candidates communicate with voters in social media ads? In this dissertation, I explore the nature and dynamics of digital campaign rhetoric. I construct a large-scale dataset of all Congressional campaign ads run on Facebook during two recent U.S. elections. Using computational text analysis, I show that ads distributed by candidates are frequently polarizing — they attack members of the opposing party and convey loyalty to their own party using identity-driven rhetoric. Yet candidates are strategic in their use of this rhetoric. They avoid using polarizing language when targeting voter networks but deploy it at high rates when targeting partisan donor networks outside their constituency. I argue that social media platforms incentivize candidates to adopt such a strategy by making it easy to narrowcast polarizing messages only to audiences most likely to be responsive. The result, however, is candidates displaying different “faces” to different groups of the American public. Overall, my findings have important implications for the asymmetric distribution of polarizing speech around virtual spaces. With this dissertation, I offer a framework for understanding how Congressional candidates, internet technology, and the quest for political power come together, contributing to broader trends of polarization in the U.S.</p>Political scienceCommunicationCampaignscandidate strategypolarizing rhetoricpolitical advertisingSocial mediaDissertation