First off, all the usual caveats: polls are unreliable; to the extent they're accurate, they're only accurate as snapshots in time, not as predictors; still lots of undecideds or people who could change their minds; yada yada yada.

I went to pollster.com and collected the date for all the post-NH polls. There were 7. I divided them up into 2 categories: polls of likely primary voters, and polls of registered voters. I also discarded one poll that listed it's findings to be of "primary voters" without explaining what that term meant: registered dems likely or somewhat likely to vote? registered dems + dem leaning indies very likely to vote? etc etc. So that's out.

The findings were kinda interesting.

For Registered voters, we had the following:

...............Jan 10-13...Jan 9-13...Jan 9-10...My guess

Clinton 45 (+12) | 46 (+1) | 49 (+9) | 47 (rising)

Obama 33 (0) | 26 (+3) | 36 (+6) | 32 (slight rise/neutral)

Edwards 13 (-7) | 14 (+2) | 12 (-2) | 13 (slight rise/neutral)

============================================

Even with the wildly disparate swings (two polls have Clinton gaining ~10 points, one shows here steady), the findings are consistent: Clinton with a big lead and neither of her rivals doing much to close the gap. If anything, she's widening the gap.

The likely voter polls tell an entirely different story.

............Jan 10-12...Jan 9-12...Jan 10-11...My guess

Clinton 38 (+3) | 42 (-5) | 39 (-2) | 40 (neutral)

Obama 35 (+5) | 37 (+14) | 38 (+6) | 37 (rising)

Edwards 13 (-1) | 11 (+1) | 9 (-4) | 11 (neutral)

=========================================

Edwards is, interestingly enough, surprisingly similar in both types of polls: ~12 points, staying neutral. One would think, given the hubbub about the media ignoring him, that the more-engaged a voter is, the stronger his/her feelings on Edwards would be, but at least by this crude analysis, that's not the case.

However, what is striking is the difference between likely voters and simply registered voters on Obama and Clinton. Perhaps this is just random error, but there seems to be a clear trend that Obama does better among likely voters, while Clinton does better among registered voters. What's more, these polls suggest Obama did not suffer a significant set-back in NH; while he's rising in all 3, she's basically steady. That's the exact opposite of the polls of registered voters. Perhaps this is an artifact of name-ID effects for the politically less-engaged.

Does this mean anything? Probably not, but to the extent it does mean something, I'd venture this: it means the race is much closer than I think most people realize, that perhaps the "Iowa bounce" for Obama was less about momentum than it was about legitimizing him as a political force, which would "allow" voters to get on board who previously wanted to vote for him but were maybe reluctant to "throw it away" in the face of the menacing Clinton machine. That was a lot of scare quotes in one sentece; apologies.

That is, if you're convinced the guy you like has little/no shot, there's not much incentive to back him or to get out and vote for him. But by beating Hillary, Obama is able to consolidate that support

Also worth noting that it's not obvious these polls reflect what, if any, changes in voter preference took place in the wake of the whole "race-baiting" flap. That certainly could've changed the fundamental dynamic of the race, and I'm guessing we'll probably see the release of a slew of national polls today tomorrow and again at the end of next week that could tell us more.

Interestingly enough, the analysis of likely voters jives quite well with the regression trends on pollster.com, though the more-recent polls show the race even closer than the ever-narrowing margins pollster shows. We could be looking at a protracted nomination battle. Kinda exciting.