COVID-19 Twitter discussions in social media: disinformation, topical complexity, and health impacts

The COVID-19 pandemic has led to widespread disinformation about the virus and its treatments on social media. This study analyzes the factors behind this phenomenon and its effects on public health using U.S.-based COVID-19-specific datasets, a validation dataset from Georgia State University, and three additional public health datasets. The study answers three research questions: factors behind social media disinformation, effects of disinformation on public health, and the role of bots versus human accounts in spreading disinformation. The methods used involve topic modeling of social media discourse, a technique for identifying disinformation, and a structural model to test the relationship between disinformation and public health during the pandemic. It finds that spreading disinformation takes a growing toll on public health. However, the public has become more resilient and saturated with vaccine-related misinformation. The study also estimates the prevalence of bots and their role in generating and spreading pandemic disinformation compared to human accounts. The study contributes to the literature by identifying COVID-19-related discussion topics, a validated method for identifying social media disinformation, insights on the dual effect of disinformation on public health, and estimates of the prevalence of bots and their role in spreading disinformation.