John Judis has an eminently reasonable post responding to my item about Obama and the Philly suburbs, in which he points out that Obama did much worse among voters with college and graduate degrees than he did in a number of earlier states.
I agree, but I don't think the evidence of slippage here is dispositive (and, in fairness, John doesn't claim it is). My argument for why Obama fared poorly in the 'burbs was twofold: 1.) They're not as demographically favorable to Obama as we'd assumed going in. 2.) Demographics aside, Clinton benefited from the support of Governor Ed Rendell, a popular former mayor of Philadelphia, who swept that city's sprawling metro area in his 2002 primary with Bob Casey. John demonstrates that 1.) wasn't the entire story. But, then, I didn't think it was the entire story either.
Having said that, it's certainly possible that Obama did worse among college-educated voters because of the gaffes and unseemly associations that have plagued him lately, as John alleges. I don't have any hard evidence to refute it. But it's also possible that Rendell, who's popular among theses same voters, proved to be the difference here. I cite several pieces of circumstantial evidence for that in my previous post. I'd just add one more: Montgomery County, a fairly affluent, educated area that John expected Obama to carry but actually split down the middle between him and Hillary, was Rendell's geographic base in that 2002 primary. Rendell did better there--beating Casey 88-12--than any other county in the state, including Philadelphia.
Anyway, John and I have agreed that we can't shed much more light on this question till the night of May 6. So stay tuned for Act II of our exciting debate: "Charlotte, NC and the Research Triangle."