Brendan Nyhan has written the best summary of all the regression posts that have been circulating.
Nyhan pulls at some of the conventional threads we've noticed, like Clinton's strength among regular Democrats and Obama's prowess in states where liberal Democrats have the time and income level to caucus.
But he notices others that we've missed, namely that Obama did better in core Democratic counties in Missouri and California than Hillary Clinton did (although Clinton won a majority of the counties, generally -- it's not clear whose support better exemplifies a distributed appeal.)
Also: Obama seems to do better among white voters in states with larger black populations. The data doesn't explain why, but I would suspect that extrinsic factors are responsible, including the degree to which Clinton contests or does not contest a state.

Hmm. Maybe. Or maybe a person's whiteness or blackness or relative browness determines their electoral decisions a lot less than you might think. If there were really reliable guide to how people vote, wouldn't we be doing a better job of predicting? I think pundits have tendency to overthink this stuff. But I forgive them, since they get paid for overthinking this stuff.
Posted by alex | February 12, 2008 9:50 AM