Everyone should read John Judis' analysis of the exit polls, but I want to quibble with a point he makes about race. Here's John:
The exit polls ask voters whether the "race of the candidates" was "important" in deciding their vote. If one looks at the percentage of Clinton (and earlier Edwards) voters who said it was "important," that is a fair estimate of the overall percentage of primary voters who were not inclined to vote for Obama because he was black.
The exit poll asked voters whether the race of the candidate was the "most important" factor, "one of several" factors, or not a factor at all in deciding who they voted for. 79% of people said it was not a factor at all. 6% said it was the most important factor, but we don't have any numbers on how those people voted. And 14% said it was one of several factors. Of those 14%, not quite six-in-ten went for Clinton. That adds up to approximately 8 percent of the electorate. (John calculates it at 11.4%, which I do not think is right. He is including the 6% who said it was the most important factor, but the exit polls included that number even though we do not know how they voted).
Anyway, eight percent is a somewhat large number, but the wording of the question still makes me extremely suspicious. Pretend you are a Democratic voter. You have read or seen numerous stories on the first election between a woman and a black man. You'd like to see a woman president and a black president. And you have decided to vote for Hillary Clinton. An exit pollster asks whether race "was one of several factors" in deciding your vote. It seems very likely that you may say yes, even if you chose to vote against Obama.
I am willing to believe there are people out there who will not vote for Obama because of his race, but the exit poll is not convincing on this measure.