Is Gallup’s place atop poll world slipping?
Source: Omaha World Herald
By Matthew Hansen
On Oct. 21, Gallup released a national poll that dropped like a neutron bomb: Mitt Romney had a whopping 7-point lead over President Barack Obama, according to the much-respected polling company whose operational headquarters is in Omaha.
Sixteen days later on Election Night that presidential bombshell looked very much like a dud. Obama coasted to a second term and ended up winning by more than three percentage points.
Now Gallup is taking incoming fire from both independent experts and the Obama campaign for what they see as the companys errant presidential polling. The criticism comes during a postelection reckoning inside Gallup, as its East Coast pollsters and Omaha statisticians grapple with a crucial question.
What, if anything, went wrong?
FULL story at link.
Read more: http://www.omaha.com/article/20121124/NEWS/711249922
thucythucy
(8,043 posts)about this?
Andy Stanton
(264 posts)Buying into the right wing fictionalized universe, no doubt. A strong majority of the poll responders couldn't possibly be self-identified Democrats could they? Time to "unskew" the results.
The Wizard
(12,541 posts)the language used in the questions asked.
PatrynXX
(5,668 posts)He's smarter than anyone on this site. I'm not someone who did well in math. I'd assume Pollsters would be good at math but looks like their margin of error is +- 20 points
Mass
(27,315 posts)started to notice, but it is not new.
R. Daneel Olivaw
(12,606 posts)If they could be so off on the Presidential election then what other races and topics could they be wrong about as well?
ProudProgressiveNow
(6,129 posts)Third Doctor
(1,574 posts)I wondered why the news shows stubbornly kept using them. Money changed hands no doubt.
Chemisse
(30,806 posts)with their polls, the same thing will happen to them.
heaven05
(18,124 posts)done slipped, badly. may they never enjoy the trust and prestige from the american populace, EVER, again! Rethug/tea party suckup poll, brown nosers ect ect ect I really enjoy the fact that a majority of americans did show that we are not as dumbed down as they thought we were. Fight the corporate, fascist world that is destroying our planet, MOTHER EARTH!
Walk away
(9,494 posts)If a news agency uses Gallop's polling data in the future it should be with a caveat.
Iliyah
(25,111 posts)I read the Pew poll when it came out with Mittens trouncing Obama and either tying with women or women surpassed themselves and was voting for Mittens, et al., What a total bullshit polling. Gallup was a joke and got crap poured all over them. Corporate media polling at its finest. But the kicker, Mittens and the gop establishment truly believed in their alternate reality really thought they were going to win. They ignored the minorities and was truly hoping the voter suppression, purging and trying to stop early voting would allow their idiot candidate to waltz in. I believe the goppers also had their secretary of states on notice of basically stealing the swing states if it were very close and thats why Pres O's campaign needed to win fairly high %.
It was a damn shame that these GOPpers always talk about patriotism, freedom and being American and they did the total opposite.
Great summary!
movonne
(9,623 posts)mrdmk
(2,943 posts)Gallup be your friend. They are a business leader in that field.
Just like the CDO's of Wall Street, how much B.S. is Gallup willing to fan to make American Business look good?
Now there collective creditability is in question! Too Bad, sucks to be them...
aaaaaa5a
(4,667 posts)Going to the Gallup and increasingly PEW site is a waste to time.
Downtown Hound
(12,618 posts)Which was almost right on.
ProudProgressiveNow
(6,129 posts)anokaflash
(53 posts)As said on Saturday Night Live, all the Gallup poll showed was Mitt had the Old White People with Landlines Demo all wrapped up! LOL
melody
(12,365 posts)Sad, but true.
BumRushDaShow
(128,712 posts)That's what Gallup's buddy Rash Limpballs wished for regarding the President. Karma's a bitch.
aaaaaa5a
(4,667 posts)I'm an average Joe. I along with many people on this board were touting how far off Gallup's tracking and approval ratings were back in May and June!
And no, not because we were Obama supporters, but because it was very apparent their demographics and crosstabs didn't add up. In fact many said their registered voter model was a far omore accurate than their ridiculous likely voter model. And like almost every other piece of analysis here and at DK, that turned out to be accurate too!
Why are some people on this board so much smarter than the folks who are suppose to be experts and who actually do this for a living?
Its actually scary that the "so called experts" are so bad at their jobs and can be routinely schooled by political internet forum nerds this often.
Downtown Hound
(12,618 posts)Moveon actually ran a piece back then in I think The New York Times showing how Gallup consistently oversampled Republicans in their polls.
Once again, this is old news...to everybody except the lame brains in the corporate media.
final registered model, actually showed Obama winning by 3 percentage points. The Rasmussen Poll shouldn't be left out either. When all the Polls changed to likely voter models, they were over doing white voters and older voters. A lot of these came from the South too, but if you look at population, the Northeast and West have higher populations.
The midterm elections are not a good parameter to base your model on either because they tend to be low turnouts. The model should have been the last general election which was 2008. Demographics were also very important, because it has changed since 2008. There was also a lot of new Pollsters, mostly Republican, popping up in this cycle, that were anti Obama. Remember the one Republican Pollster, that claimed North Carolina was a done deal for Romney. They claimed the same about Virginia and Florida. That was never the case. Pew was also pressured by the Republicans to alter their methods, more favorable to Romney.
I personally never thought it was close and also thought Obama had a chance to sweep all the swing states, including North Carolina. It is still a mystery to me though why he only got 2,178,000 votes out of 2,872,000 Democrats on North Carolina's rolls? I would like to know why so many didn't vote, just like 2008. If the Democrats could research that problem, they would have a much easier time winning the state. Some of those counties are not turning out or the registration numbers are way off.
ProudProgressiveNow
(6,129 posts)Wolf Frankula
(3,600 posts)All polls and surveys, except for the simple A,B,C, preference poll are written to get a desired result. Every goddamn one, without exception. And even preference polls have a built in bias, the order of the options.
Also refusal and non-contact rates have soared. A lot of firms have gone out of business, because their traditional method is failing. A sizable number of people don't have land lines, use caller identification to screen out these type of calls, or are simply not available in the 3-9pm traditional calling time.
Wolf
Blanks
(4,835 posts)On their website. Every presidential election had been correctly predicted by a periodical (I forget it's name).
They predicted FDRs loss at one point and as a result of small representative sampling; Gallup was able to accurately predict the winner.
The same thing has happened to them now. Their sample is not a representative cross-section of all voters because so many have cell phones. If all you sample are land lines; in this day and age you won't get a representative sample.
unc70
(6,110 posts)The "random" sample of telephone numbers for most polls come from a company with close ties to Romney and Bain.
http://www.democraticunderground.com/10021686160
I made this discovery myself before the election, but few people noticed. It needs a good DU investigation.
closeupready
(29,503 posts)Obviously, right-wing journalists in their right-wing echo chamber were shocked to see their results proved to be so wrong.
People who live in the real world and work for a living, however, saw this as par for the course compared to other recent poll results published.
Kahuna
(27,311 posts)in favor of their registered voter results which were more in line with other polling.
sarcasmo
(23,968 posts)SCVDem
(5,103 posts)Didn't they go away with Blockbuster?
Yeah, I watch South Park.
hrmjustin
(71,265 posts)Iggy
(1,418 posts)n/t
Midwestern Democrat
(806 posts)reached a point where they're basically on probation - in the next election cycle nobody will believe their poll results without confirmation from another respected pollster.
David__77
(23,367 posts)You have a confidence interval at a given level of statistical significance, like +/-6% at a 90% level of confidence. Here, ten percent of the time, you will be further than 6% in error. Maybe they should obtain much larger samples to whittle away the sampling error. It can't be THAT expensive to do so. I think 10,000+ would be required in a country like ours.
on point
(2,506 posts)Ken Burch
(50,254 posts)Last edited Sun Nov 25, 2012, 12:51 AM - Edit history (1)
n/t.
defacto7
(13,485 posts)Everything that we've become accustomed to over the years, every sacred cow, the trusted services and institutions are coming down. Times are changing and we need to be watchful for reality. The Republicans have been wearing blinders for too long. We as left wingers have been lucky enough to be aware and didn't flinch when we were laughed at for keeping out eyes on reality, not to be swayed by the cloud of myth surrounding politics or world affairs.
Trust no institution beyond what can be verified. Trust no great power or figurehead that does not balance with reality or the will of the people. Question everything!!!
tclambert
(11,085 posts)Is it an online poll? Do you poll using landline phones or cell phones or both? How do you find peoples' phone numbers? do you poll registered voters or only likely voters? How do you define likely voters? Do they identify themselves as one or the other? Do you multiply raw results by any factors to unskew the selection bias in your sample of respondents? And what about all those people (like me) who refuse to waste time talking to pollsters?
Man, it's a miracle they ever get close to the right results.