Understanding Online Human Behavior in Sociotechnical Systems That Undergo Change

Abstract

Understanding human behavior in digital environments requires examining both what people "say'' and "do'', and how these actions both shape and are shaped by the systems they inhabit. This dissertation draws on Giddens’ Structuration Theory to investigate the interplay between individual agency and structural conditions in online behavior across two modalities: social media and virtual environments.

Part 1 focuses on explicit expression through large-scale analyses of Twitter (now X) during the COVID-19 pandemic and 2020 U.S. presidential election. Using publicly available datasets that I collected, I examine how misinformation spreads, how political polarization shapes networks, and how echo chambers reinforce affective and intraparty polarization. These crises amplify the entrenchment of misinformation, especially in ideologically similar communities. The resulting network structures illustrate the recursive relationship between individual engagement and platform design.

Part 2 shifts focus from expression to behavior, analyzing decision-making in two online games: League of Legends and Teamfight Tactics. Despite each game competitively rewarding different strategies -- specialization versus flexibility -- players' behavioral tendencies remain consistent across both platforms. These results highlight the persistent role of agency, even in highly structured competitive environments.

Together, these studies demonstrate the value of combining content-based analysis with behavioral modeling to capture both consciously curated discourse and underlying action. This dissertation contributes to computational social science by emphasizing the importance of modality, context, and structure in shaping online behavior. It also provides practical insights for platform governance and the design of sociotechnical systems that better reflect how people engage, express, and adapt online.