Pandemic Politics: Timing State-Level Social Distancing Responses to COVID-19

Christopher Adolph; Kenya Amano; Bree Bang-Jensen; Nancy Fullman; John Wilkerson

Confirmed state-level caseload had only a small effect on social distancing timing. Controlling for partisanship, the number of confirmed COVID-19 cases in a state had only a small effect on the implementation of social distancing measures. (We emphasize this measure captures state-level cases: the total national cases are captured non-parametrically by the baseline hazard.) To show just how small this effect is, we compare a state with just four confirmed cases (the median among all state-days in our study period) to a state with 426 cases (the 95th percentile). The state with fewer cases was just 12.2% (95% CI: 6.5 to 17.8) less likely to implement social distancing on a given day, for a cumulative delay of just 0.38 days (95% CI: 0.36 to 0.41). While many of the early epicenters were Democratic-leaning states, it does not appear that the more aggressive action of Democratic states is a simple function of caseload. 5 This is, from a public health perspective, a fairly reasonable result: early implementation of social distancing is likely to be far more effective, so pro-active governors may have realized there was little sense in waiting for a certain threshold of cases to act, especially given delays and failures in testing.