Study: Information sources drive COVID-19 fear more than demographics

A new study published on arXiv investigates the drivers of self-reported fear during the COVID-19 pandemic, going beyond traditional demographic and epidemiological factors. Using data from the Delphi US CTIS survey, the researchers introduced a Probabilistic Causal Model to estimate causal relationship strengths. The model allowed them to identify which variables most strongly influence fear levels, accounting for confounding factors through causal adjustment and Average Treatment Effect calculations. The findings reveal that the information ecosystem plays a surprisingly dominant role—exposure to different sources explained more variance in fear than variables like age, income, or local infection rates.

Specifically, institutional and expert-driven sources (e.g., health authorities, scientists) were associated with increased fear levels, likely due to their emphasis on risks and precautions. Conversely, politicians, religious leaders, and alternative information channels correlated with reduced fear, possibly by downplaying threats or providing reassurance. These results underscore how information consumption patterns can shape public emotional responses during health crises, with potential implications for risk communication strategies. The study also demonstrates the value of causal inference methods for analyzing behavioral responses to pandemics, offering a framework that could be applied to future public health emergencies.