I think the most interesting news about the Des Moines Register poll is that it shows, again, how much more appeal Barack Obama has with swing voters than does Hillary Clinton. Indeed, all sides now seem to agree that the outcome of the caucus will depend on how many independent or Republican voters, who strongly favor Obama, will actually caucus. It's reminiscent of the 2000 Republican primary contest between John McCain, who had enormous crossover appeal, and George W. Bush, the favorite of party regulars. (I'm quite confident McCain would have won easily had he acptured the nomination.)
My colleague Jonathan Cohn, alas, has taken a rare leave from his senses to suggest that perhaps Clinton has more swing voter appeal than Obama. Jon's argument is that Clinton tends to win among less-educated voters, who are probably more similar to the profile of swing voters in a general election. To his credit, Jon only floats the possibility, but the possiiblity makes little sense to me.
First, a swing voter can be whoever you want him to be. Any voter you can take away from the other party is a swing voter. Rudy Giuliani would appeal to secular hawkish swing voters, Mike Huckabee would appeal to downscale socially conservative swing voters. Etc. It may be true that Obama's appeal is strongest among educated moderates, but that surely trumps Clinton, who seems to have no appeal whatsoever to swing voters.
Second, I strongly doubt that you can generalize from divisions within the Democratic base to the general electorate. Downscale Democrats have a strong loyalty to Clinton and are more reluctant to support a new name. That's hardly the case among downscale swing voters. I don't think there's anythig intrinsic about Hillary Clinton that gives her a natural appeal to blue-collar voters. Her support is a function of being the more familiar and established figure.