Very interesting analysis! You say that "I’ve also excluded counties with fewer than 10,000 people, because the data for those counties is *extremely* noisy." I gather that you are calculating the average Covid rate across counties in each of the 5 category. It might be preferable to calculate the average Covid rate in each category by summing Covid cases in each category and dividing by population in each category. This gives different rates because it gives greater weight to populous counties. This would minimize the effect of small noisy counties, avoid the need to drop small counties (and the associated explanations), and be a better measure of the Covid rate in each category.

The craziest part are the regional trends. Wyoming and Colorado have never been more divergent in COVID, and from COVID numbers alone you’d think that Wyoming was near Tennessee.

Very interesting analysis! You say that "I’ve also excluded counties with fewer than 10,000 people, because the data for those counties is *extremely* noisy." I gather that you are calculating the average Covid rate across counties in each of the 5 category. It might be preferable to calculate the average Covid rate in each category by summing Covid cases in each category and dividing by population in each category. This gives different rates because it gives greater weight to populous counties. This would minimize the effect of small noisy counties, avoid the need to drop small counties (and the associated explanations), and be a better measure of the Covid rate in each category.

The craziest part are the regional trends. Wyoming and Colorado have never been more divergent in COVID, and from COVID numbers alone you’d think that Wyoming was near Tennessee.