What would be an equally interesting map is a heat map of correlation coefficients between income and obesity. I'll bet it would look almost exactly like the map of obesity, but there might be some interesting pockets that could show where rich people might be fat or poor people might be fit. We could learn something from those data, I think. veen why don't you build that? :) The data must be out there in a relatively usable form somewhere.
That would be interesting, and I think I could totally build that (the data used to make the map is downloadable from the site, and I can get my hands on economic data) - if I didn't have licensing issues with the ArcGIS software package. I'm waiting for support staff to get back to me.
It's insane how enormous those clusters are, too. I found this map in a mediocre article on Atlanta. I mean, look at this. This is an insane amount of segregation: For the unaware, cluster analysis produces hotspots -the red zones- where the zone represents an area in which obesity statistically significantly high (according to the Getis-Ord Gi* statistic). The barrier here is 95% for orange and 99% certainty for red. The cold spots (blue zones) are the opposite. They represent the areas where obesity is significantly low. Grey means that there is that that area is not part of a cluster (i.e. the chance that it is, is lower than the required treshold).
Expected Appalachia to be generally lit up, but population density being comparatively low, there are just points here and there. And likely less complete data for communities under a certain size.