Every time you think you know how politics works....
Consider: Hillary Clinton's own internal tracking poll projected an 11 point loss.
So her victory is more than the sum of its parts.
Here are some of those parts
** Women came home for Hillary, for some reason
** the campaign's absentee ballot program worked well
** The campaign believes that working folks had trouble voting in the caucuses;
** Clinton has always had a much deeper base of support in New Hampshire
** New Hampshire voters have already taken their independence seriously
** Barack Obama spent a month in Iowa, making an argument tailored for Iowa; he did not have the time build up the momentum in New Hampshire; the bounce was ephemeral.
** She stepped out of the bubble: she took control at her events, forcing the fire marshal to let in more voters, tearing up, giving honest answers; she’s started to warm up to her traveling press corps a bit. (A bit.)
** She took control of her campaign, writing most of her stump speech, making most of the key decisions herself
** Barack Obama’s television ads portrayed him as a god; black and white photographs with white lettering; very thematic; very arrogant, in a way – many of them featured Obama talking to voters; Clinton’s ads were more conventionally inspiring, with lots of American flags, lots of American images; not as threatening or ponderous.
** She performed well at the debate
** The tears worked; the free media was influential
And there has to be more. Maybe there was a sublimated racial conflict -- that is something that even a Clinton adviser brought up in trying to explain the day's events. The Bradley effect may be back -- although white women would appear to be the culprit, not white men, who overwhelmingly chose Obama.
Of course, the fact that women chose Clinton instead of Obama has no racial significance whatsoever. Lying to pollsters is another phenomenon, but it is hard to account for its prevalence in New Hampshire and absence in Iowa, and it's hard to account for its absence in Obama's own internal tracking polls, which are always interpreted in light of Bradley effect.
The exit polls suggest a simpler explanation: there is no reason why the Bradley effect would be present among older white women (Clinton's base) and not older white men (who voted for Obama). There is no reason unrelated to gender that would cause a demographic split like the ones shown in the exit polls, which, incidentally, turned out to be fairly accurate.
Here is my best guess: in Iowa, working class women and older women found it more difficult to caucus for a variety of reasons, professional, modal or otherwise. Voting in primaries is so much easier; it is correspondingly easier for a campaign whose voters are tougher to turn out; we've seen that young voters are ready and willing to turn