Ruy Teixeria and John Halpin are out with a report-cum-victory lap on the 2012 election's demographics. Basically, they predicted the whole election when they predicted what colors of voters would come out, and how many of them Obama might win.
There's plenty to study here, but my favorite obvious-in-retrospect data comes from Pennsylvania. Recall that the Romney campaign made a late push there, sending the candidates and some TV money, in part because of Obama's alleged collapse with white voters and the gettability of suburbanites. There was no great movement of suburbanites.
President Obama did very well in the Philadelphia metropolitan area, carrying the region 65 percent to 34 percent, a decline of only 3 points from his vote margin in 2008. He also just about broke even in the Pittsburgh metro area (49 percent to 50 percent), a 4-point decline from his 2008 margin. Both of these margin decreases were less than his statewide decline.
This was what confused me about the Romney strategy in the rust belt/Appalachian states. Romney ran hard in coal country, and his final menu of Pennsylvania ads warned voters that Obama would keep fighting a war on coal. There aren't enough voters who care about coal! Romney and Super PACs never figured out how to appeal to moderate suburbanites, people who voted for Bush 41 then broke Democratic for 20 years.