Mapping the Census

Because I always love a county-by-county map of the U.S., Tuesday's AP map in the Star Tribune couldn't help but grab my attention. (See the story here, but they didn't put the map on the website.)

It shows participation in the 2000 Census, with black areas reflecting 75%+ participation and the lightest gray areas indicating an under 50% participation rate.



The pattern seemed pretty similar to what I recalled of the food stamp map I wrote about in an earlier post. Here's that map again: In this map, blue areas, and particularly dark blue areas, indicate higher levels of food support use.



So under-participation in the Census correlates with poverty. I suppose that's not a big surprise.

But it's interesting to compare another map, which was prepared by the Census Bureau to predict participation in the 2010 census. (Map from USA Today, March 3.) They based their predictions on 12 factors, including poverty, ability to speak English, and home ownership.



While there are a lot of counties on this map whose participation matches their 2000 participation, there are also many that do not. Check out Wyoming, for instance. It had pretty darn good 2000 participation, but is predicted to be unlikely to participate heavily in 2010. Or the southern half of California -- its 2000 participation was not bad, relative to Texas or Louisiana, but the prediction is that it will be dismal.

I wonder if something has changed in these places, or if the predictive model is just wrong?