This thesis investigates political microtargeting during the 2024 U.S. presidential election by analyzing over 120,000 unique Meta (formerly Facebook) ads from the official Trump and Harris campaigns, as well as from outside groups. Using sentiment analysis powered by OpenAI's GPT-4o, I assess how messaging in ad captions—categorized by central theme, emotional tone, and candidate alignment—varied by the age range, gender, and state of targeted Meta users. My findings reveal robust, statistically significant variation in political ad content across demographic traits, demonstrating how tone, theme, and candidate alignment shifted markedly depending on the age, gender, and location of intended audiences. These messaging differences shed a light on how official campaigns and outside groups leverage Meta's microtargeting technology to deliver distinct narratives to specific population segments, potentially shaping tailored perceptions of political candidates and issues. Through a comprehensive literature review demystifying political microtargeting, a novel case study of political microtargeting at-scale, and feasibility-minded policy recommendations, this thesis argues that political microtargeting erodes democratic legitimacy and demands urgent regulatory scrutiny. To promote transparency, replicability, and further applications of this work, all results from this thesis will be hosted in the University of Chicago Library's Institutional Repository under a DOI.
Kate Davis (Sun,) studied this question.