As we start the the New Hampshire postmortems, there are two distinct sets of analysis to do. The first is what I've seen on most of the TV commentary. You take the exit poll data from yesterday and compare them to the Iowa exit polls to see what changed in the support for Obama and Hillary. (Relatedly, you can closely analyze what New Hampshire voters said most affected their voting decisions.) It's a useful exercise, and I'm as interested as anyone in what made last night's results different from Iowa's.

But for my money the second set of analysis is far more interesting: comparing the exit poll data from last night to the final rounds of polling before the primary. Why were the results so divergent from the late polling which showed Obama with an overwhelming lead? What happened exactly that the polls didn't show?

Most of the coverage I've seen has conflated the two, explaining what happened in New Hampshire by contrasting it to Iowa and then using that contrast to explain the deviation between the New Hampshire polls and the actual election results. Those are two separate issues, and I'm not sure the analysis for the former holds up for the latter.