For those who spent the past 36 hours celebrating (and recovering) rather than obsessing over presidential politics, the big news over New Year's was the final Des Moines Register poll.  Released just hours before the end of 2007, it showed Barack Obama increasing his lead in Iowa over fellow Democrats Hillary Clinton and John Edwards.

But even before the Register's morning edition began hitting doorsteps, both the Clinton and Edwards campaigns had put out their counter-spin. And high on their list of complaints was the poll's unusually large sample of independent voters, whose overwhelming support for Obama was largely responsible for his expanded lead of seven points.

While independents can participate the caucus, as long as they register as Democrats for the day, it seems they've never shown up in the sorts of large numbers the Register's poll anticipated. If the poll had assumed turnout more along historical lines, the results apparently would have been different. And, as if to prove that very point, later in the day CNN and Zogby published polls based on more traditional turnout models. In both cases, Clinton finished first (although, for the record, the margins were reasonably close in all three surveys).

Over at the Atlantic, Marc Ambinder notes that while the Register's sampling of independents does seem rather large, the Register's pollster, Ann Selzer, “is considered to be the best in the business and has earned the benefit of the doubt.”  And over at the Politico, Ben Smith says it will all come down to turnout. So if you were looking for a definitive prediction of what will happen on Thursday, you're basically out of luck.

But if, like me, you're less interested in the Iowa results per se and more interested in what this process reveals about the candidates' relative appeal, then the Register poll does offer some insights.  Obama has always claimed he's the candidate in the best position to draw support from swing voters and, therefore, to win the election in November. The fact that he's winning so handily among independents in Iowa would seem to strengthen that claim.

Or would it? If you look closely at the way the Register's numbers break down (scroll down to the bottom of this page) another interesting tidbit emerges: the huge class divide among the candidates.

Obama was the clear favorite among respondents with a college degree: Thirty-nine percent of them told the Register they supported Obama, while just 20 percent backed Clinton and 22 percent backed Edwards. But among respondents without a college degree, it was Clinton who finished first, with 32 percent. Obama finished second, with 26 percent, just ahead of Edwards at 25. (Among respondents with some college, the candidates basically tied.)

So the question is which group is actually more representative of the swing voters a Democratic presidential candidate would need in November: people who don't align with a party (i.e., independents) or people lower on the income scale (i.e., people without a college degree).

Alas, I have no particularly informed opinion on this, either. But maybe somebody with more expertise in these matters, like my colleague John Judis or one of his friends, can help sort this out. (Apologies for turning this blog item into a bleg.)

--Jonathan Cohn