Take a look at the exit polls for NH when people are asked when they made up their minds:Something weird is going on here. If we assume that the exit polls are an accurate reflection of when people made up their minds and how they voted, there is a substantial shift to Obama post-Iowa, but only a modest shift back to Hillary in the final day.
If we pool the "In the last three days" (among whom Obama beat Clinton by 3 points) and the "Sometime last week" (among whom Obama beat Clinton by 15 points) categories, that accounts for 31% of the electorate, among whom Obama beat Clinton 39-32. So that would be Obama's post-Iowa momentum, which propelled him from a tie in the polls to an alleged double-digit advantage going into election day. And then Hillary totally wipes out this surge with her little 3-point advantage among the 17% of voters who decided on the last day? That's not nearly enough.
I'll buy that there was a late surge for Hillary, and that it was probably fueled by sympathy for her and for female candidates in general, and may have been crystallized when she cried in Portsmouth or during her impressive debate performance. (And it may have even been caused by Chris Matthews.) But the same-day surge toward Hillary reported in the exit polls is just not consistent with all the tracking polls we were seeing right up to election day. It seems to me that Obama's poll position in New Hampshire has been way overstated all along.
Another possibility is that independent voters, who were expected to show up for Obama, instead participated in the GOP primary and voted for McCain. But the exit polls don't really support that. 42% of the Democratic electorate was independent, compared with 34% of the Republican electorate. And the Democratic electorate was much larger than the Republican one (284,000 Dems compared to 233,000 Reps).
So what's going on here? Was it really a race effect?