Before everybody gets carried away trying to explain Barack Obama's loss to Hillary Clinton in New Hampshire, just remember, there's a rule in science that says the simplest explanation is probably the correct one. And the simplest explanation is this:
There was no Iowa bounce for Obama and Clinton never rallied. The polls were just wrong.
Forget the Wilder effect. Forget the Bradley effect. Forget trying to explain the results as a women's revolt after Chris Matthews' condescending comments. The polls taken in the lead-up to the New Hampshire vote were all over the place. Zogby had Obama at 42. Fox had him at 32. And 20 other polls filled every point in between.
People scratching their heads questioning how Hillary won are basically asserting that all of those polls were correct. That Obama literally had a double-digit post Iowa surge and then Clinton rallied from behind - all in 5 days. The only way to believe that argument is to believe that the polls were measuring dynamic changes in support over a very short time period. Despite the fact that many of the polls were taken concurrently over several days, yet still didn't agree with one another.
This was true in the lead-up to the Iowa vote as well, though nobody seemed to notice. On election day, Zogby predicted Obama would win Iowa by 7, while ARG predicted he'd lose to Clinton by 9.