Yet there is an issue with the scale, which is quite distorted by single player/low population (<50k) county values. They should be filtered out for the definition of the coloring scale. They don't provide any good information since county population sizes vary by magnitudes. (I wouldn't even recommending filling the county with a color in these cases as simple dot at the county centroid should suffice)
Granularity matters, counties are a not a optimal unit for this, aggregating by congressional districts would better, yet still not optimal.
[added] Alternatively counties could be clustered to some minimum size. The US Census Bureau provides sets of counties with above 50000 total population as a XLS spreadsheet. http://www.census.gov/people/eeotabulation/data/eeoupcoming....
When that's done, you can see that the South is the clear leader.
Its not like that is the only thing to do in town either, they have cable TV, the internet, and bars just like the rest of America (well, except for a few dry counties).
I honestly have no idea what to make of it. Calling football in the south a religion is cliche, but as far as I can tell it is just an accurate way of describing it.
Unfortunately it then becomes clear that sample sizes are so small that we can't draw any real conclusions. You look at Garfield county in Montana and there are only 25 college-age males, so the 1 player anomaly makes it suddenly look like a big sports hub.
Cool concept and execution, but insufficient data.