This thesis explores how cultural discourse varies in both structure and content across political party lines on TikTok. To answer this question, I take presidential candidate accounts as partisan representations and examine the TikTok videos reposted by followers of candidate accounts. To analyze the TikTok repost data, I employ frequent itemset mining, social network analysis, and BERTopic modeling. I find that the semantic communities in Trump followers' reposts are more overlapping and less distinct than those found in Harris followers' reposts. I find that Harris followers' reposts cover a wider breadth of topics. I analyze several emergent themes in the content of the reposts, finding connections to gender, sexuality, and violence. These findings corroborate current literature surrounding American partisanship, culture, and social media. I discuss how TikTok mediates and produces culture, indicating that future research is needed to understand the political relevance and impact of the social media platform.
Anita Sun (Fri,) studied this question.