Today, using questions from the General Social Survey, Nate Silver tries to quantify the effect of Barack Obama's election on the racial attitudes of white Republicans and Democrats. On several of the most overtly racist questions ("blacks are lazy," "blacks are unintelligent") there's little evidence of change. But on two of the questions with more political salience, there's evidence of a pretty substantial effect.
The chart on the right illustrates this. The number of white Republicans who believe the government spends too much money on blacks had been trending slowly downward for years. Based on that trend, you'd expect the number today to be a bit above 20 percent.
Instead, it took a sharp upward jump in 2010 and again in 2012, ending up a bit over 30 percent. In other words, among white Republicans, it appears that the election of a black president has increased the belief that blacks get too many government bennies by about 10 percentage points. This belief is now at levels not seen since the anti-busing days of the 70s.
I'm not sure what conclusions to draw from this beyond the obvious ones. As you might suspect from some of my posts over the past few years, I basically blame Fox News and conservative talk radio for this state of affairs. Without Fox fanning the flames of racial animus over the past several years, I suspect we wouldn't see this effect—or at least, not as strongly. That's just a guess on my part, but I think it's a pretty good one.
UPDATE: Several commenters suggest a different explanation for this: white Republicans get more concerned with money spent on blacks under Democratic presidents and less concerned under Republican presidents. This doesn't fit the evidence perfectly, but it fits reasonably well. However, it doesn't explain Democratic attitudes, which you'd expect to be the exact opposite.
So....maybe. Overall though, I'd note that the trend lines are fairly similar for both Republicans and Democrats until about 2006-08. Then you get a significant divergence. The Obama/Fox News effect seems to explain this best.
Also, the subsample size for white Republicans and Democrats is fairly small (perhaps 300 or so), so there's a fair amount of noise in the trend lines.