All "Likely Voters" Are Not Alike
A look at the latest round of national poll results at Pollster.com tells us that all "likely voters" are evidently not alike. The "outlier" AP-GfK poll of likely voters reported yesterday shows a dead heat, with Obama leading by just one point! The GWU/Battleground poll also has McCain down by just a nose in the horserace (1 point). Meanwhile, at ABC/Washington Post, Obama's up by 11%; up by 4-6% in Gallup; and according to Zogby--who dissed the AP poll--Obama's up by 12% in the latest tracking poll and may be heading toward a blowout if present trends continue. A blowout?
Without access to the open secrets of the good Gallup Organization, as Mark Blumenthal tells it in his latest NationalJournal.com column, it's awfully difficult to find out how much of this variability in election projections is due to differences in how "likely voters" are determined by different polling organizations, how much of it is the result of sheer chance fluctuations--the only solid justification, in my mind, for all the questionable averaging in the "poll of polls" business--and how much of it is the consequence of different voter turnout model assumptions. Maybe Mark can help clear some of this up in his next set of postings at the National Journal, at Pollster, and at TPM. But, short of full disclosure from all the major pollsters regarding all the proprietary details (see Nancy's post the other day) of their likely voter/turnout models--and making available all their datasets for replication of their projections--we're not very likely to ever know the full "likely voter story".
As a little thought experiment...we should ask ourselves how likely would it be that a "likely voter" in Gallup's model (or dual model as it is now) would end up being classified, simultaneously, as a "likely voter" by Zogby, by ABC/POST, by AP-GfK, GWU/Battleground, SurveyUSA, etc.? I'd like to think there would be a lot of overlap, but I strongly suspect there would be some notable deviations. Alas, we're not very likely to find out in this election cycle.
Scott's posting yesterday tells us that figuring out who's a "likely voter" might be a hell of a lot harder than we thought because of the great new field position created by the DNC and Obama-Biden's ground game. While we're still worrying about how to determine who's likely to vote and who's not, we might also want to take a look at some of the ambiguous standard questions and response categories used by Gallup and others to decide who's a likely voter. Asking whether people will "will definitely vote, probably vote, probably not vote, or definitely not vote", or questions like: "How often would you say you vote, always, nearly always, part of the time or seldom..." would be prime candidates as questions for cognitive laboratories at Michigan and elsewhere. One respondent's interpretation of "definitely" could well be someone else's "probably". So one pollster's "likely voter" may not be another's. All "likely voters" are surely not created equal. Mike's right: we could use a lot more depth in this polling business without sacrificing clarity. Where do we begin?
Nate at fivethirtyeight.com addressed this yesterday:
http://www.fivethirtyeight.com/2008/10/some-likely-voter-models-are-suspect.html
We won't know the real answer until November 4.
October 23, 2008 5:02 PM | Reply | Permalink
From POLITICO.com:
Fake America located
In tonight's NBC News interview with Brian Williams, McCain explains that the "elites" are located in D.C. and New York City:
WILLIAMS: Who is a member of the elite?
PALIN: Oh, I guess just people who think that they're better than anyone else. And-- John McCain and I are so committed to serving every American. Hard-working, middle-class Americans who are so desiring of this economy getting put back on the right track. And winning these wars. And America's starting to reach her potential. And that is opportunity and hope provided everyone equally. So anyone who thinks that they are-- I guess-- better than anyone else, that's-- that's my definition of elitism.
WILLIAMS: So it's not education? It's not income-based? It's--
PALIN: Anyone who thinks that they're better than someone else.
WILLIAMS: --a state of mind? It's not geography?
PALIN: 'Course not.
WILLIAMS: Senator?
MCCAIN: I-- I know where a lot of 'em live. (LAUGH)
WILLIAMS: Where's that?
MCCAIN: Well, in our nation's capital and New York City. I've seen it. I've lived there. I know the town. I know-- I know what a lot of these elitists are. The ones that she never went to a cocktail party with in Georgetown. I'll be very frank with you. Who think that they can dictate what they believe to America rather than let Americans decide for themselves.
UPDATE: Numerous readers in those cities note that they were the targets on 9/11.
October 23, 2008 5:14 PM | Reply | Permalink
McCain says "...I know what a lot of these elitists are. The ones that she [Sarah Palin] never went to a cocktail party with in Georgetown."
McCain gives himself away here as a member of the elite himself. He's been to those cocktail parties, has been going for decades. Shame Brian Williams didn't pick up on that--but then, he wouldn't...
i guess it's not polite to point out someone's hypocrisy on national TV. Also, just an aside, clearly Sarah Palin is not a member of the elite, at least according to McCain's reckoning.
October 23, 2008 11:24 PM | Reply | Permalink
Oh, i overlooked the obvious: Brian Williams no doubt attends those cocktail parties himself.
October 23, 2008 11:39 PM | Reply | Permalink
I think this answers most of your questions. If you look at the internals of the poll, you'll find that a FULL 44% of their "likely" sample were EVANGELICALS!!!
This is unconcionable cooking of books. Any other reasonable survey would consider that DOUBLE the maximum number of evangelicals in a likely voter sample. For shame, for shame!
Here's the link:
http://www.ap-gfkpoll.com/pdf/AP-GfK_Poll_3_Topline_FINAL.pdf
October 23, 2008 6:18 PM | Reply | Permalink
Yes...BLOWOUT...I have to sit here and listen to Rachel Maddow predicting how McCain will win this election because of her strong desire to be "right", refusing to be optimistic and it makes my stomach churn.
After this disaster of the last 8yrs and with McCain being such an erratic authoritarian hypocrite liar whose 'experience' is completely negated by his poor judgment (Palin), who bangs from wall to wall on economic plans, stands in the midst of his failing campaign and rabble rouses low information voters to see Obama as a terrorist etc scary negro, and then gets pundits like Maddow to go along with all of the "steal the campaign" explanations like the phony "Bradley effect" or Bogus polls showing the race neck and neck so stealing it won't look so obvious, I wonder how can irrationality (like Maddow quoting right wing prints to justify an Obama loss)look right past the obvious like 3/4ths of America wanting the republicans out, like Obama's superior manner over the sinister bitchy McCain and his utterly phony VP groupie, and believe that Americans want to continue this disaster which won't survive 4 more years of Republicans and not feel they have completely missed the point. The only way McCain wins is if it is stolen or the SC intervenes and appoints McCain president...but even a great number of republicans won't help that happen this time seeing McCain/Palin as dangerous for our country.
Look at yourself in the mirror and repeat over and over again until the collective unconscious screams it back at you...OBAMA WILL WIN IN A LANDSLIDE. HE IS TO BE OUR NEXT PRESIDENT. McCain/Palin are the end of a movement that nearly destroyed our democracy and are going out the way they came in...like snakes in the grass and rats in the gutters.
October 23, 2008 7:06 PM | Reply | Permalink
Re: This is unconcionable cooking of books.
I don't think that this was deliberate, but they should have noticed this anomaly and either discounted the results, or normed them to a better approximation of the general population. If one of these national polls discovered that 20% of their respondants were Black you can be sure they would do just that.
October 23, 2008 8:22 PM | Reply | Permalink
Just did my own informal poll - I went and sighted in my rifle for the deer season at a rifle range in Minnesota.
4 years ago every other vehicle had a Bush/Cheney bumper sticker and the gun range had signs for all the local republican candidates everywhere.
This year, my son and I did not see any bumper stickers and there was only a small NRA sponsored table with local republican candidate literature. The Reagan/NRA vote just doesn't care about the republican candidates this year and I don't think they are going to go out and vote.
October 25, 2008 2:05 PM | Reply | Permalink