Jason links to an Economist blog post which says, of me, "Mr Zengerle's colleague, Jon Chait, contends that few of Mrs Clinton's blue-collar supporters would run over to John McCain's side if Mr Obama wins the nomination." That's not what I said. I wrote, "Sure, some of the Clinton voters will go for McCain in the fall, but you can't assume all (or even very many) of them will." I was referring to the total Clinton vote, a subset of which consists of blue-collar whites. My point was that it's erroneous to assume that the entire Clinton vote is an anti-Obama vote. Most Clinton voters will back Obama if November. If Clinton had prevailed, most Obama voters would back her, too.
Clearly, a significant number of white blue-collar Democrats will defect to McCain in November. (A significant number would have defected if Clinton won, but the number will probably be larger for Obama.) My point was that you can't just look at one or two voter blocs in isolation, which is what reporters and pundits all seem to be doing now, the Economist blog included:
Mr Zengerle's colleague, Jon Chait, contends that few of Mrs Clinton's blue-collar supporters would run over to John McCain's side if Mr Obama wins the nomination. Similarly, Mr Obama's supporters in the black community and in urban areas are unlikely to shun Mrs Clinton if she were to become the nominee. But which scenario is more likely. In other words, whose support of the party is more fleeting? The answer is obvious: Clinton's blue-collar supporters are the more politically tetchy of the two groups—the more likely to change sides. And that matters a great deal in close, must-win states like Pennsylvania and Ohio. Advantage Clinton.
Look, as I wrote, Obama is currently running slightly ahead of Clinton in trial heats against McCain. Therefore, as of now, the number of potential Democratic votes that Obama can get but Clinton can't must be at least as large as the number that Clinton can get but Obama can't. That's simply a mathematical truism.
Obama runs way ahead of Clinton among independents, and currently that advantage more than makes up for his deficiency among white, non-college educated Democrats. You can make the case that Obama's independent support is soft, or that he'll get into debates with McCain and start crying like a baby, or whatever you want. But reducing the electability question to which candidate can win blue-collar whites, or whether blue-collar whites are more loyal than African-Americans, is an exercise in stacking the deck. You have to look at the whole swing vote, not just the portion that's friendlier to Clinton.