Let's get it straight: The polls were not wrong.

[milkyway]

They are merely samples of public opinion at a particular moment in time. They measure the present. They do not predict the future. Each poll has its own methodology and is accurate within a noted margin of error. Many polls of New Hampshire post-Iowa were taken, and every single one showed Obama with a comfortable lead. The polls obviously truly reflected the public's opinion at that time.

So what happened in the last 24 hours? A number of things, but I think primarily the "crying" incident as discussed here.
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=389&topic_id=2634674&mesg_id=2634674

btw, I'm an Obama supporter. Congrats to Clinton supporters. If she wins the nomination she will make a fine president, perhaps even a great one.

[JMDEM]

1. Oh right...

An 18 point swing in one day from a little crying jag.

Let's get it straight: you are deluded.

[Pavulon]

3. Response bias...

classic example.

[milkyway]

4. Wrong. Some polls had Obama up double-digits, but here's where the aggregate polls were at this

morning:

http://www.pollster.com/08-NH-Dem-Pres-Primary.php

And if it wasn't the "crying" incident and the media reaction to it, along with all the other seeming piling-on , how else do you explain such a substantial move in 24 hours?

[progressivebydesign]

7. ONE WORD: UNDECIDEDS. n/t

[jsamuel]

2. the polls were wrong because they overestimated the indie vote and underestimated the woman vote

[Old_Growth]

9. And Republican Crossovers n\t

[havocmom]

12. BINGO!

[milkyway]

10. A post at mydd.com speculates that the indies thought their vote for Obama was not needed,

so they voted for McCain instead. The polls had Obama with a more comfortable lead, seeming secure. This makes some sense to me, but I'm not sure if this was the reasoning of a meaningful number of indies.

http://mydd.com/story/2008/1/8/234223/0703

[KittyWampus]

11. if this is true, it may help Obama in the long run. Firming up his support.

[milkyway]

5. kick.

[progressivebydesign]

6. Thanks for a nice, reasonable, kind, post. It's a gem in sea of.. well..

a sea of something unpleasant tonight. And you're totally right about the polls. There were so many undecideds there, as well, that really sways things. Also, the whole cell phone thing regarding polls. If the proliferation of threads tonight about HRC stealing this primary are a taste of things to come for the next 48 states, I will be leaving DU after happily existing here for 7 years. :(

[crispini]

8. No, polls are always right, it must have been DIEBOLD CONSPIRACY

getting Shillary to start winning for her corporate masters!


:sarcasm:

I just wanted to get that in there first.

Myself, I'm a fan of Occam's Razor. I think you're right on the money.

[althecat]

14. Don't stumble Crispini you had it right the first time...

[althecat]

15. Three words "Reluctant Clinton Responders"

rCr

Clinton has the same problem Bush has. People don't want to admit to being her supporters...

[IDemo]

13. Regardless of methodology, the polls were wrong

Without a very much larger sample size than any polling organization is realistically able to poll, there is no reason to believe that they reflect a close enough resemblance to the actual voting block to make them scientifically viable. Polls are meaningless statistical exercises.

[Hutzpa]

16. What a load of Throllop!!

[Hutzpa]

17. What a load of Throllop!! eom

[bridgit]

18. Tracking polls are dicey, many here at DU are on the 'no call list'...

others only use cell phones, or are just plain gone or slam the phone down. Data becomes skewed by having the pool made thicker & smaller. Samples used to generate some kind of graph. Polling is nearly a reductive process anymore. I loved watching Matthews try to glean some info about education and coming up so lost as Olberman was all like, "Should we go to a commercial, Chris?"

And for that matter, so long as people like Matthews are allowed to pursue & interpret the data, there's no guarantee of valid conclusion.

[tandem5]

19. I think the term "likely voter" says it all...

I don't think the raw polling was wrong - its when the polling groups start predicting the percent contribution each raw category will play in the final result that things get magical. I've always been weary of "likely voter" polls. In a straight election I'm more inclined to believe a "registered voter" poll, but this was an open primary where two candidates on either side depended on a single independent group, voter turnout among democrats was underestimated, and there was a sudden and dramatic shift among women toward Clinton in the lead up to the election. I can't imagine coming up with an accurate "likely voter" model in a calm two-way race let alone this mess.