Fig. 2.

Incremental $R^2$ from Adding Connections to Individual U.S. Counties.

Note: Panels show results from regressions to predict COVID-19 cases per 10k people by county on March 30, 2020. The incremental $R^2$ is the increase in $R^2$ from adding $log\left(SCI\right)$ and $log{\left(SCI\right)}^2$ to a particular U.S. county, over and above a set of baseline control variables: 100 dummies for percentiles of distance to the county under investigation; population density; median household income; and dummies for the six National Center of Health Statistics Urban-Rural county classifications. The graphs show the distributions over the incremental $R^2$s for adding social connectedness to each county with a population over 50,000 in turn. Each regression in panels (a) and (b) excludes counties within 50 and 150 miles of the county of interest, respectively. In each panel the 10 largest incremental $R^2$ are labeled.