Oklahoma's results are interesting because they seem to be so far outside of the overall pattern. Obama only did this badly in one other state - Arkansas (69-27). And this was the only state where Edwards picked up significant support. Democrats out-polled Republicans 401,000 to 330,000
Clinton won the state overwhelmingly, with 55% of the votes cast. Obama picked up 31%, and John Edwards got 10% of the vote. Considering that Edwards was polling ahead of Obama when he dropped out of the race, the residual support for Edwards, although high, probably isn't as surprising as it looks.
As usual, Oklahoma primary voting can be misleading. Despite being solidly Republican at the national level, there are about twice as many registered Democrats in Oklahoma as Republicans. In general, Oklahoma Democrats are on the conservative end of the spectrum.
The first thing that jumps out at me about the Oklahoma results is the lack of young people - only 9% of the people captured in the Democratic exit polls were under 30. Among Oklahoma Republicans younger voters made up 14% of those polled. I'm a little bit disappointed that they asked different questions of the Democrats and Republicans - for example, they asked Democrats how often they attended church, but they also split Republicans into born-again or not born-again. In a place like Oklahoma that's likely to be an important criterion among Democrats as well.
Contrary to what I found in the other polls, Clinton did better than Obama among Liberals in Oklahoma. She also did better than Obama among people who thought the Kennedy endorsement was very important - 53% of those people voted for Clinton, while 43% of them went for Obama. Even in Arkansas, where Clinton's margin of victory was larger than in Oklahoma (42 points) people who thought the Kennedy endorsement was very important went for Obama. Not here.
John Edwards picked up what amounted to either a protest vote or an ignorance vote - it's likely that at least some of the people who voted for him didn't know he had dropped out of the race, but 10% of all Dems? (More really, because a fair proportion of the people who didn't know he dropped out would have still voted for Clinton, since she was the front-runner). Edwards did best among people who called themselves conservatives - despite having the most populist rhetoric. Of course, so did the populist Huckabee. But was this conservative vote for Edwards "identity" politics? Was it the people who couldn't bring themselves to vote for a black man or a woman?
It's interesting to look at Oklahoma on a county-by-county basis. The only county that appeared to mirror the state's average was Osage County, which went 52-34-10. Elsewhere, there were some interesting patterns. Obama outperformed his average in two counties - Oklahoma County (where he actually beat Clinton) and Tulsa County. He also did relatively well in the counties around OKC and in Osage County (which is adjacent to Tulsa). Elsewhere in the state Obama did less well. In eastern and southern Oklahoma, and much of western Oklahoma, Clinton and Edwards did better than average, at Obama's expense. Interestingly, Edwards did best in the far western counties - and he did so at Clinton's expense. In eastern Oklahoma, Edwards tended to do well at Obama's expense.
Western Oklahoma is overwhelmingly white. Eastern Oklahoma has a large Native American population - 10-20% in most counties, over 40% in Adair county. The African American population is concentrated in the cities - Oklahoma City Metro, Tulsa - and in the southwest of the state. While the exit polls weren't able to say anything about minority voters, it would appear that Obama did best in counties with a large proportion of African Americans. Clinton did best in counties with a large proportion of Native Americans (although she did well everywhere). Clinton did worst in far western Oklahoma (where the population is overwhelmingly white) and in Oklahoma County (which has the largest African American population).