I'm interest in the predictions vs the results.
Don't understand why that is a problem......
New Hampshire's primary represents some unique challenges to the eventual result of polling:
1) unlike most other states with primaries, the candidates tend to spend more time in NH per capita than any other place, and the state is small enough for the voters to make an INFORMED choice and change their minds due to repeated exposure (this more in cases of multi-candidate fields)
2) New Hampshire became MORE important with the creation of the Iowa caucuses in 1976 because basically it turned into an almost "you have to win at least one or go home." And because of the attention the winner of Iowa got - and Iowa is easy to pull an upset in since it puts a premium on passion and willingness to have others know for whom you voted - the polls rocket all over the place. (This got worse when the time frame moved from a month like in 1980 to less than a week in 2008).
3) New Hampshire - with the exception of the miniscule number of African-Americans (2%) - has one of the most representative cross-sections of voters in the entire US (it's 92.7% white and 50.7% female. And perhaps more so than any other state, you have voters from there who would vote for the President of the NRA on the issue of guns and with Bernie Sanders on universal healthcare - these positions in the EXACT SAME VOTER.
4) The Yankees of New Hampshire who aren't passionately liberal or conservative - and that's a chunk of them - refuse to talk to pollsters, which skews the results. I lived in Salem, NH last year, and to my utter shock there was a "Trump Superstore" there selling Trump merchandise. The willingness of the extremists to talk but the less passionate to be more reserved than the average voter skews it, too. (Note: I cannot explain the WHY, but I have observed this in both my assignments up here and one of my good friends is a lifelong New Hampshire native who concurred with my assessment but couldn't really explain the "why" either other than a lot of the state IS rural (but not rural Alabama).
===================
The example that has always stood out to me is the 1992 GOP primary result:
Bush 53%
Buchanan 38%
Now, I'm not going to blame this particular result on the media, but what happened in the early exit polls for most of the day was the press was getting the count around 52-48 in favor of Bush, borderline 1968 McCarthy win while losing territory. The press such as it was in 1992 reported that every 1/2 hour on CNN Headline News all day so that by the time the evening news hit, the President less than a year after winning that Gulf War and rising to 91% was being repudiated by his own party. But then the actual votes were counted and, well, you can usually figure 1/3 of a normal electorate is against an incumbent (well, in pre-Trump politics you could). So Buchanan getting 30-35% of the vote wasn't even a story in the real world. Richard Nixon got 17% of the GOP primary vote in 1964 WHEN HE WASN'T EVEN RUNNING!!! (15,000 people wrote him in - because you can do that there). But when the press was conditioned to believe, "Oh my, Pat Buchanan is within four points of knocking off Bush, who is from New England," it was because of the exit polls. The election post-mortem opted to say that Buchanan voters were more likely to admit they voted for him than Bush voters did.
You may also remember the 1996 Senate election in NH.
That's the night all the networks called the seat for Democrat Dick Swett (Yes, that's his name, folks!) over Republican Bob Smith. The Voter News Service made an honest mistake in how they did their caclculations, fixed it, and life went on.
But you're correct that polling prediction vs results in New Hampshire is.....well, it's like some of the elections in Florida have been.
