An r^2 of .81 — the dependent variable is log(number of cases), with log(population) on the right hand side. https://twitter.com/richard_florida/status/1246827366854336515

Key table below, from https://marroninstitute.nyu.edu/uploads/content/The_Coronavirus_and_the_Cities%2C_27_March_data%2C_final_draft_31_March_2020_VersApril3.pdf

Looking at per-capita rates, rather than the level, means a lower r^2. But there's still plenty of factors that help explain local covid19 differences: like density, demographics, and climate.

(table below from this thread: https://twitter.com/JedKolko/status/1246155182464839680)