This article in The Gadflyer is a direct challenge to the findings of the Statesman series The Great Divide, which claims that America is becoming more segregated by politics. Author Philip Klinkner makes some interesting points, but I have to say I thought he relied quite a bit on numbers of unknown (to me, anyway) origin, and a nontrivial amount of handwaving. For example:
Other measures of recent election results show that while segregation along political lines has increased, the change is not remarkable. For example, in the 2000 election approximately two-thirds of voters lived in a county where the Democratic vote was within a range of 12 percentage points from the national average. Thus, most voters live in counties that are not all that different from the national results. Moreover, this range is not significantly different from that in past presidential elections.
Isn’t this basically saying that the standard deviation of Democratic voter percentage by county is twelve points? That actually sounds pretty high to me. Beyond that, though, give me the numbers. What’s their source? Can you show me a graph? And for crying out loud, can’t you tell me how this compares to, say, 1980 or 1960 or even 1992? I’d feel a lot more confident in this data if I didn’t feel like it were being pulled out of a hat.
I sent a note to Bill Bishop, coauthor of the Great Divide articles, to point this out to him. I’m interested to see how he responds to it.
UPDATE: I’ve received a reply from Bishop, along with an attached Word doc that goes into some detail to rebut Klinkner’s assertions. I’ll post it later – it appears to have been a Statesman article from June 13, but it’s not available online any more as far as I can tell. For now, I’m calling it for Bishop – if nothing else, he’s unafraid to show you his R^2 values, and I respect that.
In the meantime, check out this article on softening support for Bush in rural areas.