07 February, 2008

Obama's Multiple Equilibria

Something I noticed while browsing the exit polls on Tuesday is that in more liberal states, like Massachusetts and California, Obama tends to win voters who never attend church but he loses those who attend regularly. On the other hand, in more conservative states, like Alabama, he tends to lose voters who never attend church and does very well among those who attend church regularly.

He has a weird dual attraction--the most (socially) liberal of the liberal love him, but so do the most (socially) conservative of the conservative. That's a pretty hard combination to pull off--could it be driven simply by the fact that African-Americans like him for other reasons and there are lots of African-Americans in conservative states that attend church regularly?

An Andrew Sullivan reader notes a similar pattern for race:

In states where the black population is less than 5%, Obama has a record of 7 wins, 2 losses and 1 undecided (NM). In states where the black population is 20% or higher, he is undefeated at 4-0. However, in states that are between 5-20% black, his record is a fairly dismal 4-10 (with one of those victories being Illinois).

The theory here is that Obama does well where the black population is so low that identity politics isn’t an issue. And, he does well where the black share of Democratic primary voters is so high that he needs few white voters to carry the day.

This pattern--winning the extremes, losing the middle--reminds me a whole lot of models in economics in which increasing returns to scale leads to multiple equilibria.

No comments: