We asked our participants to predict the numbers of infections in 2020 six times from February to December and to predict the number of deaths in 2020 five times from March to December.
Numeric estimates of COVID-19 infections and deaths increased dramatically after March 2020. In March itself, however, when negative emotions were highest, our participants’ predictions of deaths and infections were quite low. Participants underestimated both deaths and infections, but their estimates increased as infections increased fairly well. Whereas the median estimate of infections was 13,500,000 by the end of the year, the actual number of infected in the U.S. in 2020 was over 20 million and the number of deaths exceeded 350,000 by December 31st.
But estimates of cases and deaths didn’t correspond to people’s perceptions of their own likelihood of catching COVID-19
However, even as cases and deaths rose in April and later, participants’ estimates of their own likelihood of getting coronavirus remained steady. Logically, as cases rose, people should have thought that their likelihood of getting it was higher.
These are the percentage who thought they were “somewhat likely” to “completely certain” to get coronavirus.
Instead of relating to the numbers of cases and infections, the perceived likelihood of getting coronavirus tracked emotional reactions instead (see below). When people judged their own likelihood of COVID-19, they may have used their feelings to assess their risk more than they used the statistical rates of cases around them.
These findings are consistent with the idea that knowledge of the numbers does not necessarily lead people to understand what those numbers mean for them. Indeed, as we’ve discussed in the New York Times, people who spent more time looking at statistics were more fearful and seemed to focus particularly on negative news, such as death rates vs. survival rates.