The N.H. polls - all that could have gone wrong

[Honeycombe8]

I see more and more pundits jumping on the racial bandwagon theory. It's possible that a few people changed their minds at the last minute because of race. It's also possible a few changed their minds about voting for an elderly man or a woman. But I've heard and thought of the following things that really make more sense:

1. Pundits are saying ALL the pre-election polls were WAY off. Fox News, actually, said that it's pre-election N.H. poll was pretty close. I don't recall the poll, so I can't swear that's true. But all the polls were clearly not showing a substantial win by Obama...I recall seeing one showing a modest win. And I read one article in the media by a columnist who thought Obama would win, but not by much.

2. Pundits are saying the exit polls were ALL WAY off. These weren't all published, so I don't know. BUT I DO RECALL THAT SOME (A LOT?) OF THE EXIT POLLS FOR THE 2004 KERRY RACE WERE WRONG. Remember? It doesn't only happen when one of the candidates is a non-white male.

3. One pundit reminded the host of a show that there was a record turnout in N.H., so it's possible that the poll models (both pre- and post-election) were not suited to the large voter turnout.

4. One pundit on NPR this a.m. pointed out that there was a large (15%) "undecided" vote. Those undecideds would have decided on the day of the election, and it is possible that the majority broke for Clinton, esp. after the "emotional moment" thing and the debate. I'm an Obama supporter, but I gotta say...I was feeling pretty sorry for the way some were piling on her, esp. for what was being described as a crying jag (I posted here a couple of times as a reminder that SHE DID NOT CRY). I can see where some who weren't steadfastly for one or the other might have cast their votes for her.

5. The pre-election polling was stopped too early. A lot happened after the last pre-election poll. The "emotional moment" incident, and the huge battering in the press she got over it, was one thing.

6. The Clinton Machine did a superb job at getting out their vote. A large number of over 45 women turned out. Clinton and Chelsea personally made rounds to the polls.

7. The weather was nice....a lot of older people went to vote.

8. A lot of the Independents that would've voted for Obama, broke for McCain. McCain absolutely needed to win N.H., so they made sure he got it.

9. Some people, having seen the exaggerated pre polls didn't go vote, possibly, thinking Obama had it in the bag. Esp. younger voters.

10. Can you say, "DEWEY DEFEATS TRUMAN!"? That is the infamous headline printed before election results were out, printing what was presumed to be a sure thing. Public opinion polls and other indicators said so. For those unfamiliar with this story...and I guess you can guess....TRUMAN WON. Turns out that the pollsters had stopped polling too soon...they were convinced of Dewey's win. And I wouldn't doubt that some Dewey supporters didn't vote, thinking it was a done deal. Surprise!

11. Finally, it could be that some women who were going to vote for Obama changed on that final day (not having been that strong in their support maybe) to Clinton because of the "emotional weeping" moment and because of her remembering to point out that electing a woman as President is "change" in and of itself.

This is an unusual year. I don't think there has to be a racial element to the polling being off. I think it is the fault of the pollsters, like what has happened a time or two in elections before.

I hope they study this and come to some conclusion about what went wrong, though, so they can fix it!

[TwilightZone]

1. In Dewey Defeats Truman, the pollsters stopped polling in October, not the night before.

There is a little difference. They stopped polling weeks before the election. NH pollsters were polling until the night before.

In fact, it was because of that race that polling was extended to the day before the election.

http://en.wikipedia.org/wiki/U.S._presidential_election,_1948

[Honeycombe8]

2. Actually, they stopped a few wks before. Exit polls here showed 50-50. Close.

Dewey was in 1948. Now it's 2008...this technological country moves a lot faster than 1948.

But if you look at the exit polls...at least some showed the voting was about 50-50. Not too far off, huh?

I think people chose to look at ONLY the polls that showed the big lead. Some journalists and some polls (pre and post election) did not show a huge lead.

[TwilightZone]

3. Re: stopped polling too soon - I'm not sure what you mean.

They polled until the night before the primary. I don't think that pollsters are allowed to poll the day of the primary, nor should they, so I'm not sure what you expected them to do.

[Honeycombe8]

5. Actually, some polls stopped on Saturday and other days. nt

[TwilightZone]

7. Doesn't look like it to me.

http://www.realclearpolitics.com/epolls/2008/president/nh/new_hampshire_democratic_primary-194.html

Suffolk, ARG, Zogby/Reuters, Rasmussen, and CNN/UNH all polled through Monday. The others listed polled through Sunday. None of the ones listed stopped on Saturday or "other days", so I'm not sure where you're getting your information.

[Honeycombe8]

8. Well, Sunday is two days before. Correct? Correct.

[TwilightZone]

10. Your assertion is that the pollsters stopped too early, and that skewed the results.

That assertion is not supported by the evidence. There is very little difference between the polls that ended Sunday and those that ended Monday. The polls that polled through Monday night had Obama in the lead by 5, 13, 7, and 9 points.

The problem is not that they stopped polling too early. The problem is that 18% (according to exit polls) of the voters didn't choose a candidate until primary day, and there was obviously no way for pollsters to have predicted where those 18% would go.

[Honeycombe8]

13. Okay, so you don't buy that one. There are 10 other reasons with some evidence

behind them.

[TwilightZone]

14. Sure, what the heck:

1) The pundits were wrong. Zogby had it down to a 2-point race the day before the primary. There was a 15-point shift in the two days prior to the primary.

http://www.huffingtonpost.com/john-zogby/polling-the-new-hampshire_b_80657.html

Rasmussen and another pollster or two have come out with preliminary info that is similar, and I suspect that we'll see more of that in the next few days. It only looked like a big win for Obama because the polls were using three- and five-day rolling averages. Even with the averages, Suffolk had it at only five points. Not exactly a landslide. MSNBC said that one poll had shown it as a one-point race, but I haven't located that one.

As we've already discussed a bit, I think they just ran out of time. Their polls couldn't reflect the change that was occurring because they were three day or five day averages, and the change largely occurred in a day or day and a half.

2) I haven't heard any pundits indicate that the exit polls were off. The first exit polls I saw showed it as a 38/38 tie, and as more exit polls came in, it shifted slightly to 39/37 Clinton. I don't really see any reason to think that they were fabricated.

3) It's very possible that turnout was underestimated. I'm not sure that proves anything, though, other than that the polling might not have accounted for it and was, thus, inaccurate.

4) There were a lot of undecideds, and they were being largely ignored by the media and by most of us on DU. In addition, the margin of error in the polls was being largely ignored, as well. A 6-point lead doesn't really mean much if there's a 6-point MOE.

5) We covered late polling. ;)

6) Very possible. Turnout is very important.

7) True.

8) In addition, more Independents went to Hillary than many expected. Obama won Independents 41-31. Hillary won Democrats 45-34. Independents might have also thought Obama was going to win without their votes, so they felt free to vote for McCain in larger numbers.

9) Not sure I'd agree with that, considering the huge turnout and large Dem vs. Rep advantage.

10) Covered that

11) No one really knows, but it's certainly possible.

Personally, here's what I think are the most likely scenarios:

I think that the polls underestimated turnout. They almost certainly underestimated the gender gap (57% of Dem voters were women). They probably overestimated the Independent vote for Obama, combined with a better showing by McCain, which also would have drawn Independent support from Obama.

I also think that the Obama "surge" post-Iowa was probably overstated. He gained momentum, no doubt, but I think everyone overestimated it. The recovery by Hillary in the final couple days was largely hidden by the rolling averages that most of the polling companies used.

Lastly, I'd like to apologize for being kind of sarcastic in some of my prior responses. I've been responding in a few "it must be fraud" threads, and I think that carried over into this thread. Your summary is a good one - just don't let the pundits convince you of anything that may not really be there! :)

Edit: fixed link

[Honeycombe8]

4. I think the pundits and others are trying to blame the people instead of themselves. They are ...

embarrassed about how easily, and thoroughly, they jumped on the Obamarama Wagon and on the "Clinton is dead in the water" Wagon.

I'm an Obama supporter, and I was excited and thrilled, and all of that...and more. But I noted to a friend (and here on DU) that temperatures needed to come down a bit. It was only one state, historical though it was. That the Clintons were not done for, etc., etc.

But when there's a flurry around someone like that, they tend to disregard reality sometimes. It's easy to get caught up in the excitement. Obama IS pretty exciting!

[MethuenProgressive]

6. The polls said "40% Undecided" - that was ignored by the pundits.

Nothing wrong with the poll, just with the spinners.

[elocs]

9. I heard on public tv tonight that polling stopped on Sunday,

24-48 hours before the election during which time there was movement to Clinton. There are not that many actual votes in NH so that an 80000 vote difference translates into 3 percentage points. And I did also hear that there were women who were going to vote for Obama, when push came to shove they remembered their feminist roots and felt they had to vote for Clinton. The polls were wrong for a combination of many things. What should be really held up for scrutiny are all the commentators who went on to parrot the polls that turned out wrong.

[Zynx]

11. I fully support what you say here.

Too few days of polling, a lot of events not being taken into account, high turnout, etc. Polls were garbage.

[Honeycombe8]

12. Here's a FOX poll that showed 50-50 tie between Obama & Clinton Sat. before...

http://news.yahoo.com/s/ap/20080105/ap_on_el_pr/democrats_rdp

This is a link to a Fox news poll that was done on the Saturday before the Tuesday election. It showed a tie between Obama and Clinton at 33% each, with Edwards at 20%.

This might be the poll I heard Fox refer to...saying that their poll was closer than anyone else's. And it wasn't far off the mark.

Most of the polls showed Obama winning, but it's NOT TRUE that ALL the polls showed a win by him...OR that he had a huge lead. Those were only SOME polls.

[TwilightZone]

15. That same poll showed Obama up 9% on Sunday

Link: http://www.unh.edu/survey-center/news/pdf/primary2008_demprim10708.pdf

That's a huge one-day swing, especially if it's a rolling-average poll. Sounds like Fox took the second-to-last poll to bolster their claim instead of the final one. Not surprising!