At this point, I think we've all stopped hearing arguments of the form "Obama lost Massachusetts in the primary so he'll lose it in a general election" or "Obama won South Carolina in the primary so he'll win it in a general election" but there's a frustrating persistence of the idea that performance in a primary campaign in a swing state might be a good indication of general election strength there. In reality, there's just very little reason to believe that. I would very strongly prefer Obama over Clinton, but that doesn't stop me from very strongly preferring Clinton over McCain. All this throat-clearing by way of introducing a quote from this post from Noam Scheiber, commenting on some new Pennsylvania polling data:
A poll showing that Obama can get blown out in the Pennsylvania primary and still hold his own there against McCain suggests working-class white Democrats simply prefer Hillary, not that they find something inherently objectionable about Obama, whom they're apparently happy to support in the general.
Right. The poll indicates that Clinton will do much better than Obama
in the Democratic primary but Obama will do slightly better than Clinton in a general election. There's nothing paradoxical or even counterintuitive about that, but somehow we've gotten twisted around in knots over this sort of thing.