So which pollster was closest to reality in New Hampshire?

[Ichingcarpenter]

Many were way off the mark only one came close


• FEB. 4-7 Emerson College 783 LV
0.90
Sanders +12
54%
42%


• FEB. 4-6 Monmouth University 502 LV
0.81
Sanders +10
52%
42%


• FEB. 7-8 American Research Group 409 LV
0.79
Sanders +9
53%
44%


• FEB. 4-8 University of New Hampshire 363 LV
0.73
Sanders +26
61%
35%


• FEB. 5-7 UMass Lowell 407 LV
0.59
Sanders +16
56%
40%


• FEB. 2-3 Marist College 567 LV
0.43
Sanders +20
58%
38%


• FEB. 2-4 Suffolk University 500 LV
0.42
Sanders +9
50%
41%


• FEB. 2-4 MassINC Polling Group 393 LV
0.35
Sanders +15
54%
39%


• FEB. 2-6 Franklin Pierce University 407 LV
0.32
Sanders +7
51%
44%


• FEB. 2-4 Gravis Marketing 702 LV
0.32
Sanders +16
58%
42%


The final results were

Live results: New Hampshire
DEMOCRATS 89% reporting



Bernie Sanders 132,410 votes (--) 60.6%



Hillary Clinton 83,368 votes (-49,042) 38.2%

[mhatrw]

1. According to Huffington Post's Pollster, 12 of the last 13 polls underestimated Sanders' victory.

Only this one came close, but it was an overestimate.

This poll underestimated Sanders' margin by a whopping 15%.

[Android3.14]

2. So WTF is going on with these polls?

Can landlines vs. cell phones really cause that sizable of a skew?
Are the campaigns and PACs paying for specific results in hopes of influencing the outcome?
Are the exit polls accurate?

It's obvious that Sanders and Trump overwhelmed any attempt at shenanigans by the established politicians, but it looks like we are unable to trust the results in other elections when the pollsters cannot even come close.

[Ichingcarpenter]

3. I had two semesters on polling and statistics

with my pol/sci degree and from what I've seen
these days many of the polls are crap and have political agendas, with loaded questions, bad
poll samples, bad data collection etc..

I wouldn't and don't take polls serious anymore and yes even on line polls are included in that summation.
Only when they are over 12% differences you might see a trend.

Anyway exit polls, depending on the company, can reveal better on what is really happening. But as we saw with Kerry in Ohio the vote can be stolen too

[Stevepol]

7. All the polls but one and THERE'S A REASON!

Ever since electronic voting machines have started to count the results of elections, the results have consistently tilted red in comparison with the exit polls. In other words, in all likelihood, the "real" vote was maybe 5-10% more Dem than the electronic vote tally would indicate.

After the "red shift" became common place, the polling companies began to think there must be some explanation for it and they came up with a real doozie: Republicans are more close-mouthed than Dems when approached by pollsters. Evidence for this? None whatever but this is the only thing they came up with or could come up with because obviously IT COULDN'T BE THE VOTING MACHINES! That would just be whining. Everybody knows voting machines can't be maliciously programmed. Look at how shiny and new they look, look at the bells and whistles, etc., etc. They would never do this! Never mind that, as one expert has said, it's "trivially easy" to rig an election using voting machines, never mind common sense. IT COULDN'T BE. WE'RE THE GREATEST DEMOCRACY IN THE WORLD! Don't you dare tell us that the machine manufacturers (extremely right wing in most cases) would do this or local ideologues or rabidly anti-abortionists. How would they do it? Never happened.

So the polling companies, not wanting to go out of business because of what seemed to be lousy predictions began to build into their predictions the "red shift." So when they discovered in their polling that Bernie had a 22% advantage on Hillary, they fudged it to make sure they didn't get exposed as chumps. Trouble is New Hampshire isn't Ohio or Florida. New Hampshire has a pretty good oversight over the votes evidently, and from the looks of it they keep a tight control over the tallies in each precinct so it would be harder to cheat without getting caught. Hence, the polling results. The New Hampshire poll apparently didn't bother to use the "red shift" to come up with their pre-count polling numbers.

[Godhumor]

10. The most likely answer is the simplest one. A large % of people made up their minds in the last day

It was seen in both states. Polling that stopped more than 24 hours before the election started can be quite wrong as a result (In NH I don't believe anyone polled within 24 hours of the primary.).

[RunInCircles]

4. Please Redo Your Math

Marist was the closest to getting it right not the poll you are crowing about.

[Ichingcarpenter]

6. Please Redo Your math education.

Which number is closer to 60?
58 or 61?


• FEB. 2-3 Marist College 567 LV
0.43
Sanders +20
58%
38%


FEB. 4-8 University of New Hampshire 363 LV
0.73
Sanders +26
61%
35%

Do you need me to post a number line?

[A Simple Game]

9. It is the difference that is the important number not the high and the low.

Marist pretty much nailed the difference, your poll was off by 3 or 4 points. Why do you think Bernie's number being close is important but not Hillary's number?

[debunction.junction]

5. Thank you for posting.

As far as predicting Bernie voters, 61 versus 60 is amazing. With 10.9% of precincts left it could get even closer.

[Enthusiast]

8. Kicked and recommended!

[Orsino]

11. I see at least one of them was close than UNH's.

Marist's looks like the best prediction.