Residents of U.S. counties with more connections to China or Italy were more likely to follow early pandemic restrictions

Residents of U.S. counties with more social connections (measured as Facebook friends) to China or Italy – the first countries to report major COVID-19 outbreaks – were more likely to adhere to social distancing restrictions at the onset of the pandemic, according to a new study. The findings suggest that social networks supplied pandemic-related information that significantly influenced individual behavior. However, social ties also could have negative effects on pandemic-related behavior. For instance, areas within the U.S. with more connections to counties with low education levels, high proportions of 2016 Trump voters, and high fractions of people who deny the existence of climate change experienced less adherence to mobility restrictions during the early pandemic. Previous research suggests online social connections may act as warning systems during natural disasters and can impact personal health decisions by spreading both facts and misinformation. To explore the role of these connections in the context of COVID-19-era social distancing, Ben Charoenwong and colleagues measured social connectedness between U.S. counties and foreign countries using aggregated, anonymized data from Facebook’s Social Connectedness Index. The researchers combined this data with anonymized, county-level mobile phone location data that served as a proxy for adherence to social distancing measures between February 1 and March 30, 2020. They found that a one-standard-deviation increase in social connections with China or Italy correlated with a nearly 50% increase in the effectiveness of mobility restrictions. The authors also conclude that differences in how effectively Democrats and Republicans adhere to social distancing arise from discrepancies in the information they receive (facilitated by their social connections) rather than due to biases in how they interpret that information. “This finding has important policy implications, as it suggests changes in the information environment can boost the effectiveness of non-pharmaceutical interventions,” the authors write.

###