Misinformation During a Pandemic

We study the effects of news coverage of the novel coronavirus by the two most widely-viewed cable news shows in the United States — Hannity and Tucker Carlson Tonight, both on Fox News — on viewers’ behavior and downstream health outcomes. Carlson warned viewers about the threat posed by the coronavirus from early February, while Hannity originally dismissed the risks associated with the
virus before gradually adjusting his position starting late February. We first validate these differences in content with independent coding of show transcripts. In line with the differences in content, we present novel survey evidence that Hannity’s viewers changed behavior in response to the virus later than other
Fox News viewers, while Carlson’s viewers changed behavior earlier. We then turn to the effects on the pandemic itself, examining health outcomes across counties. First, we document that greater viewership of Hannity relative to Tucker Carlson Tonight is strongly associated with a greater number of COVID-19 cases and deaths in the early stages of the pandemic. The relationship is stable across an expansive set
of robustness tests. To better identify the effect of differential viewership of the two shows, we employ a
novel instrumental variable strategy exploiting variation in when shows are broadcast in relation to local
sunset times. These estimates also show that greater exposure to Hannity relative to Tucker Carlson
Tonight is associated with a greater number of county-level cases and deaths. Furthermore, the results
suggest that in mid-March, after Hannity’s shift in tone, the diverging trajectories on COVID-19 cases
begin to revert. We provide additional evidence consistent with misinformation being an important
mechanism driving the effects in the data. While our findings cannot yet speak to long-term effects,
they indicate that provision of misinformation in the early stages of a pandemic can have important
consequences for how a disease ultimately affects the population.