Modeling Media Influence on the Formation of Polarized, Misinformed Attitudes

Abstract

PhD Defense:

Misinformation has become a widely studied topic that spans a variety of disciplines. For example, while computer science may study information spread through social networks, media studies may analyze why individuals choose certain media sources, and psychology may focus on factors correlated with belief in misinformation. All these aspects are equally important and contributive, but quantitative models of how populations form attitudes when exposed to misinformation rarely incorporate these interdisciplinary concerns. The work of this PhD thesis takes steps towards addressing this gap by motivating an agent-based model that incorporates some of these interdisciplinary dynamics on top of existing opinion formation models. Throughout the work, we focus on a motivating case study: misinformation and belief formation surrounding Covid-19 mask-wearing. Using our model, we are able to show both theoretical and empirical results about how information may be shared and adopted in complex media ecosystems; ultimately using empirical news media and opinion polling data to validate the model and reason about what dynamics may have led to widespread belief in Covid-19 mask-wearing misinformation. But more than their immediate results, the model and validation method present a new environment for media researchers to continue quantitatively studying media effects and misinformation.