It is very interesting to peruse Nate Silver’s post-mortem analysis of the pollsters. The thing that really stands out for me is that, among the most active pollsters, there was only one (Pharos Research Group) that showed a significant skew toward Obama, and that pollster had been flagged as unreliable by Mr. Silver and was not used in his model. The other three pollsters who showed any Democratic skew were basically accurate (RAND Corporation: +1.5%, Quinnipiac +0.3%, and We Ask America +0.1%). By contrast, all other frequent pollsters showed a Republican skew, and the most respected (Gallup) showed the worst skew of all (an astonishing +7.2%). American Research Group (+4.5%) and Rasmussen (+3.7) were also pathetic.
Nate has some observations about polling methodology that you may find interesting, but he doesn’t discuss the subject that interests me. What I want to know is what benefit Romney received from the systemic skew of the polls in his favor. If the polls had been accurate, the race would have been considered uncompetitive all throughout the fall, leading to much lower fundraising and enthusiasm on the Republican side. In short, the polls were bullshit, and they created a bullshit picture. No one wants to admit that at least some of the pollsters were intentionally wrong, but that is obviously the case with Rasmussen, ARG, and Gallup. It also appears to have been the case with Romney’s internal pollsters, at least to the degree that they released information to their donors.
This story line that Romney and his team were shell-shocked is a load of bullcrap. They may have underestimated Obama’s ground game, but not by seven points. They told this lie to keep the money flowing and to keep the media from calling the race early. And I think it enabled them to keep the House.
And I’m pissed.
The House wasn’t going to flip.
Look at the House popular vote margin. Look at the number of seats that did flip.
Now extrapolate. Even if you’re off by a factor of two on the high side, it’s still a narrow GOP hold.
I agree with you. The real scandal about the House isn’t in the polls, it’s the gerrymandering. Everybody knows the Democrat vote for the House was + 500,000 over the Republicans. (If you include late results from FL and AZ, I would think it’s higher than that.) Originally, Sam Wang thought the Republicans had about a 2.5 point advantage in the house. After the results were in, he had to revise that to approximately 5 points. That’s unconscionable.
This. I saw something that said the redistricting gerrymanders gave the Republicans a boost of 2.5%. That is enough or nearly enough to keep the House all by itself.
This is how you do it, using PA, and OH, two states Obama carried, as examples.
The House seats? 30 R, 9 D.
Cycle is pretty hard against us. 1992, 1996, 2006, 2008, 2012. We missed 1990, 2000 and 2010.
From my limited experience with election campaigns and my extensive experience with human greed, I make the following observation.
There is a set of professional campaign organizers/advisers, media consultants, pollsters and the like who make a considerable amount of money from campaigning including self-dealing on media buys and web site operation. These pros (like the pros you find in Times Square) have a vested interest in keeping campaigns going and at least coming in close. They are not actually interested in the candidate or any cause except their own careers and income. Telling the client to pack it in because it’s hopeless does nothing for their careers or bank balances. On the contrary, it is in their best interest to keep conning the candidate and the donors. When the inevitable defeat happens, one or more scapegoats including the candidate, his family, or his running mate can be blamed. In the case of Republicans, the Liberal media, and various paranoid conspiracies can be blamed as well. Anything or anyone except those harpies who fed on the hopes, dreams, fears, and prejudices of the candidate and his/her supporters. I wonder if they privately call volunteers “muppets”.
Can anyone tell me why the Chicago Federation of Labor is still making media buys instead of using the moneymore constructively? Or did I just answer my own question in the preceding paragraph?
As Chris Hayes pointed out, with a chuckle, there’s no doubt that poster boy Sheldon Adelson, who makes his own billions off of gambling, was taken in by his own methodology. Rove and the gang measured the gullibility greed factor of these SuperPac donors, fed it, then gleefully exploited pollsters who were paid to be a part of the game.
Newman and Redford’s characters in the Sting couldn’t have done it better.
As an aside, all this is making McCain look better and better and if he could have somehow gotten through the primaries intact, would have given Obama more of a race than Mitt?
I don’t agree that the pollsters were deliberately wrong.
Look at Gallup. First of all, that’s an organization that has a decades-long reputation that they don’t want to trash. Second, you don’t skew polls by 7 points. If you want to put out b.s. numbers, you’re a little smarter than that.
I think the pollsters and the Repug political pols made the same mistake: They couldn’t believe that Obama’s base would turn out in the numbers they did. This biased all the likely voter models they used, and probably influenced their GOTV efforts as well. Their mistake was probably partly prejudice — minorities and young people had been starry-eyed enough to fall for the Obama fad in 2008, but were too detached and lazy to turn out in 2012 — and partly their own epistemic closure — nobody they knew had anything but extreme hatred and disdain for Obama and his accomplishments. Add in overconfidence in voter suppression, and you have a recipe for self-delusion on a grand scale.
Really? So, why did Gallup “fix” their data one day before the election? From seven points up to one point up.
When did Gallup abandon standard polling procedure and begin making up models based not on data but their opinion of turnout?
Gallup’s polls were corrupt this year. About as blatant as can be.
As for Rasmussen, they also “fixed” their numbers for election day, moving it to a one point advantage. But their model is designed to skew Republican. And, yet, they still fixed it to look reasonable at the end.
Total corruption.
Like Rasputin, the provide early BS ‘leads’ to form a narrative and keep the GOP base engaged. They drop back these bogus leads right before election day so they don’t look like total hacks. I always discount Rasputin and Gallup 3 pts or so to get a little closer to reality.
“They told this lie to keep the money flowing and to keep the media from calling the race early. And I think it enabled them to keep the House.”
Yup.
Rasmussen always “fixes” their polls at the end. They want their final results to look respectable after all, even if the outfit is owned by GOPers(remember, Scotty used to work for W/Darth ’04).
So, why did Gallup “fix” their data one day before the election?
Gallup’s polls were characterized by a high level of volatility throughout the election. That eight point swing wasn’t at all unique for their polls this cycle. Go back and look at their releases over the past few months.
When did Gallup abandon standard polling procedure and begin making up models based not on data but their opinion of turnout?
Huh? Everyone’s polls were based on likely voter models, and those inherently involve assumptions about turnout, and different rates of turnout. That is standard polling procedure. That a particular pollster screwed up their turnout model might or not be evidence of a problem, but noting that a pollster used a likely voter model isn’t evidence of anything.
As for Rasmussen, they also “fixed” their numbers for election day, moving it to a one point advantage.
Rasmussen has been pulling that scam for years. Gallup hasn’t. I’m going to need more than a high level of volatility and the existence of an LV model before it makes sense to start shouting about “total corruption.”
Gallup is supposed to have a rigorous likely voter screen. The screen tells them who will vote and then they are supposed to discover the Dem/Rep differential from that rigorous screening. In other words, they don’t look at 2004 (D/R even) and 2008 (D+7) and then arbitrarily pick a number based on a hunch. They tried blaming the ground game and late registration drives. I think they ignored their own methodology and adjusted their polls to help Romney. The strongest evidence for this is the huge swing back to the middle they had on the last day. Other polls showed slight movement, but nothing dramatic that would translate to hundreds of thousands of voters, maybe millions, changing their mind in the last week.
The strongest evidence for this is the huge swing back to the middle they had on the last day.
But once again, their polls were swinging around like crazy throughout the campaign. If there really had been a Rasmussen-like, distinct departure at the end, that would be good evidence (as with Rass) of deliberate fudging.
I think it’s a little more complicated than that. First of all, Gallup has a history of skewing Republican, maybe not as much as this year, but definitely enough to raise eyebrows.
http://www.huffingtonpost.com/2012/06/17/gallup-poll-race-barack-obama_n_1589937.html?ncid=edlinkusa
olp00000008
The biggest problem for all pollsters lies in the “likely voter” parameter, because of the vastly different ways one may define who is likely to vote. There is also the problem of representative sampling, connected with such things as land lines/v. mobile phones, or minorities (certainly Latinos) that are just under-sampled.
There is ample opportunity for error here, and not much incentive to change, since all pollsters concerned about reputation can readjust their polls towards correctness in the last few days of the campaign.
And it isn’t easy to say how much of it is deliberate deception and how much self-deception – how much is laziness and how much cost-cutting.
That’s why poll aggregation, when done well (as Nate Silver and Sam Wang do) are so much more accurate than individual pollsters. In the aggregate, pollsters cancel out one another’s biases.
Actually, the non-cellphone pollsters where quite accurate this year.
The aggregate does the best job, provided you account for bias, as Nate does. Without that adjustment, the aggregate was way off. And it’s easy to see why when you notice that none of the polls (save one shady one that Nate excluded entirely) skewed more than 1.5 points to Obama, while all the other polls skewed at least somewhat to Romney. Basically, there were no skewed polls for Obama but they were at least 3 egregiously skewed polls for Romney.
Sam Wang did about as well as Nate, but he adjusts for bias in a simpler way — by taking the median rather than the average. This seems to work quite well.
I would like to hear from both Rasmussen and Gallup what their explanations are for their terrible track records. And it better be good.
That’s a deeply irresponsible allegation, Boo, without at least some additional track record that the Romney campaign was predisposed to lie to the public or its donors to advance its own cause.
(Can you hand me the pliers? I put my tongue in my cheek so hard that now it’s stuck…)
LOL. You’re going to need a jackhammer for that job.
Humor often provides the most enlightening takes on matters such as these and Garrison Keillor’s PHC yesterday live from Chicago had a great couple of skits along the lines of this same theme.
Can’t figure out from the allowed HTML how to insert a hyperlink. However you can check the show out at http://www.prairie.publicradio.org
[Title Goes Here url/goes/here]
Or you can just put a regular html link.
Given our TradMed, do you really think they were ever going to allow things to be portrayed as an Obama blowout? Never!! Their easy ads dollars would dry up. Look at right here in PA. How much did it cause Casey to spend? And all of Smith’s millions did what to the race? Made him less of an embarrassment compared to Man-on-Dog? But who did all those millions benefit? The TV stations!!!!!
I just don’t believe Willard gave 2 shyts about helping the GOP. it was all about him
NPR uses the Gallup pro-GOP pollsters to comment on polling, This is another example of NPR’s current Repukeliscum tilt. Go to their website
http://www.npr.org
and tell them to FIRE FUCKING GALLUP.
Just a reminder about what credible outlets do when they realize their pollsters are fudging things:
http://www.dailykos.com/story/2011/05/30/980544/-Markos-has-settled-the-Research-2000-Lawsuit
NPR’s been tilting Repug since at least 2004, when it became very marked during the Swiftboat Liars brouhaha. Not sure an angry note will make much difference, as it is apparently now NPR’s policy to tilt that way. As long as they have their audience and their contributors, why change?
NPR has leaned Right ever since the Republican House threatened to take away their funding. I wonder when we will start seeing specials doubting Climate Change.
I really think it started during the Bush Admin. There were any number of stories both then and after Bush left office about how the Repug at the head of the Corporation for Public Broadcasting was throwing his muscle around in a political way. Of course, as far as I can tell — mostly from the frequency of complaints I see on sites like Daily Kos — they’ve stuck with it, firmly solidifying their rep as the Nice Polite Republicans. It’s not really surprising, since they seem to be depend increasingly on business “underwriters”(aka sponsors) as well as the wishy-washy totebagger set.
Be Pissed, Boo.
We’ll always win the mid-terms…your constituency watches American Idol…out constituency still studies the Constitution…
We kicked your ass in 2010, and we’ll kick your ass in 2014, when the teeny-boppers don’t have a celebrity to vote for…your ideology depends on twitter and Facebook…
Bookmark it, libs!
Hey, our troll is back! And gee, somehow he has nothing to say about the ass-whuppin’ his side just took….
We’ll always win the mid-terms
You mean like in 2006?