Research

Abstract: Social media has been a core information distribution center in recent decades. With more and more people sharing information and attitudes on social media, people are exposed to other people's (even strangers') points of view much more readily than before. This paper develops a social learning model of incomplete information with bounded rational Bayesian agents that can observe only their private information and their predecessors' choices in a sequential network. Through the comparative analysis of the model, this article identifies how the diffusion dynamics of true news and misinformation differ and how the characteristics of news, information timeliness, and the informativeness of the private and public beliefs of truthfulness determine these differences. Using high-dimensional Twitter data related to COVID-19 misinformation, the empirical analysis confirms the theoretical findings and quantifies the impact of signal informativeness on diffusion patterns. Furthermore, the findings of the paper provide valuable insights into policy implications for mitigating the spread of misinformation and enhancing user welfare in social media environments.