Virtually every major poll leading up to the NH primary had Obama ahead, some by as much as 13%, leaving many pollsters and voters wondering about the disparity. Mark Blumenthal has an excellent round up of possible explanations here.

He covers all possibilities, from the Bradley effect to voters changing their minds at the last minute to error on the part of pollsters. And yes, he also considers the fraud argument that has been floating around based on Hillary's significant lead in areas with Diebold voting machines compared to paper ballots.

Personally, I think The Cry changed a lot of voters' minds, particularly older women who turned out for Hillary. The polls were basically right about Obama--he received close to the 38% they predicted--they just underestimated Hillary's support.

Combine Hillary's last minute successful campaigning with a favorable ballot placement, and I think the shift starts to make sense. In the past, candidates' names were rotated on ballots from precinct to precinct, but this year the ballots were the same throughout the state. Based a random drawing, Hillary was the first major candidate near the top of the ballot. Studies estimate that favorable placement can lead to a 3% increase in votes, which is roughly the margin of her victory.

Regardless, she won, and I recommend reading Blumenthal's analysis before jumping to any conclusions. Let's just hope polls leading up to the remaining primaries and caucuses are more accurate.