Explanations

The pollsters came up with a rational excuse for the consensus mistake on the New Hampshire Democratic primary (a mistake of 8-10%, which is more than the 3-5% range of error--at a 95% percent level of certainty--which the polls typically would accept). Fact is that the pollsters stopped polling the last two days, during which--it appears--Clinton picked up about 15% of women voters. Her "weepy moment" (Gwen Ifill) played well with those voters, according to the rationale.

The search for blame has now shifted to the thousands of media hounds, the pack which descended on the state ostensibly to cover what was happening, who failed to detect any sign of the mini-quake in HRC's favor over the last two days. Some are blaming white voters for saying they would vote for Obama, then not doing so (is that some sort of crime--like perjury, or a hate crime?)

I have another explanation, which goes to the heart of the problem with these opinion polls. David Brooks on the Lehrer Report pointed to the problem, though he provided no solution. Fact is that the exit polls asked voters when they had decided for whom they would vote, and those that said they decided in the last two days split evenly between Obama and Clinton.

My hypothesis--for which I have no proof and which would be very difficult to prove in any case--is that the shift did occur, but the women voters were somewhat inaccurate about when they decided for whom they would vote. They'd been thinking about Clinton vs. Obama (or Clinton vs. Edwards, especially, I'd say) or Clinton vs. Romney or whatever. The "weemo" pushed them over the edge in their decision, but they didn't acknowledge it as such in the polls (the rationale following after the action).

There's no deceit involved because they were different people taking the polls before and afterward. If they had been the same people, they might have recognized what answer they had given before and acknowledged the change in their view. The conclusions are:
1) Exit polling on the post hoc rationale for actions is likely to come up with self-deceiving answers. There's no way to check on this, and it is simply a product of human nature to invent a rationale afterward which does not accurately portray their semi-rational choice of a candidate in the voting booth.
2) New Hampshire voters insist on defying expectations--we've seen it time and again. It's the "live free or die" thing; in this case they gave Hillary the opportunity to live free and they "died"--sacrificed their interest.

But no harm done, really; Hillary was always going to stay in the race (contrary to what John Edwards dreamed after Iowa), and I would have expected her to rally herself at some point; Obama had to taste defeat somewhere along the way, and better to get it out of the way early.

It's still down to California, as I've always thought. A highly unpredictable result--I've been watching the odds move dramatically on Rasmussen, though so far it has attracted little betting interest.

In some ways, and on a smaller scale, but with more media candlepower per square voter, Nevada might provide some preview of its big neighbor. South Carolina will be the acid test of whether Obama can reach the desired level of Afro-American support, so both the states--which look to be likely Obama wins--will help set the stage for the big drama. I'm trying to say that the two Dem primaries in the intervening weeks are sideshows, not meaningless, but not worth the attention they will get from the media, who are desperate to redeem themselves and hoping for more good story.

On the Republican side, Michigan does indeed look to be the Romney-McCain showdown in which Cmdr. John will get his knockout punch. Giuliani will eagerly scoop up Romney's leavings.
As I've said before, I love the way this informal UNP system is working--I just think it should be happening in June!