Political Animal


May 15, 2012 10:55 AM Poll Quirks and Triumphalism

By Ed Kilgore

When I saw the headline for the new CBS/Times poll of the presidential race, I groaned aloud: not because “my” candidate was behind, or because a three-point Romney lead nearly six months from Election Day means anything, but because of the timing. I figured anyone with an interest in claiming that Obama’s expression of support for same-sex marriage was an act of political suicide would leap on the poll with all four sets of claws.

And sure enough, the sounds of extraordinarily early triumphalism are in the air amongst our little friends on the Right. Brett LoGiurato’s item on the poll for Business Insider bears the highly nuanced headline: “This May Be The Most Disastrous-Looking Poll For Obama Yet.” RedState’s Erick Erickson cites the poll as prime evidence for this modest headline: “May 6, 2012: The Day Obama Lost,” an allusion to the date of Joe Biden’s remarks on his comfort level with same-sex marriage. Katrina Trinko at National Review is equally certain the same-sex marriage issue has vaulted Mitt into the lead. You can only imagine what Dick Morris, who has recently been predicting an epic Republican landslide, will do with this data point.

Aside from the fact that Romney’s lead in the CBS/Times poll is within the MoE, there is this little matter of the particular poll’s quirky methodology. National Journal’s Steven Shepard explains:

The poll is actually a panel-back survey: Respondents to this poll were originally interviewed in mid-April for the previous CBS News/New York Times poll. In the mid-April survey, Obama and Romney ran dead even, with each candidate earning 46 percent of the vote.
Panel-back surveys are conducted for a number of reasons: Some pollsters occasionally consider it instructive to interview the same universe of people before and after a specific event. In other cases, it is simply cheaper to dial phone numbers behind which pollsters know are voters who are willing to answer public-opinion surveys; in doing so, pollsters avoid many of the nearly four-in-five adults who decline to participate in telephone polls.
But the practice also has drawbacks. Not every respondent to the original poll is available or willing to be interviewed again; indeed, of the 852 registered voters interviewed last month, just 562 were re-interviewed for this survey. Some analysts also believe that the experience of being interviewed previously affects how they view the race, potentially making them more likely to seek out information about politics and the presidential race, though the results of the poll are virtually identical to those of the previous survey.

Romney’s slight “surge,” and the finding that the pro-Obama gender gap has shrunk dramatically, aren’t the only notable things about this poll, BTW:

A CBS News analysis of the poll notes that, with the exception of a poll closely following the killing of Osama bin Laden by U.S. forces, its results represent Obama’s highest approval rating since April 2010. This conclusion, however, should carry an asterisk due to the poll’s unusual methodology.

So Obama’s approval rating is up while his share of the vote is down. Which finding should we be paying attention to? Hard to say, though in one of his classic “Chill, folks” posts about early polls, Nate Silver points to the president’s approval rating as the single most reliable indicator of where the race currently stands.

The safest approach is to pretty much ignore this one quirky poll and stay focused on poll averages, which currently give Obama a modest lead amidst a slight trend towards the tightening of the race in recent weeks. That’s highly predictable insofar as Romney’s nomination contest victory has boosted his support levels among the Republican voters who were backing his opponents. It’s possible that Obama’s same-sex marriage statement hastened this development a bit, but it was going to happen anyway.

If I were a conservative, I’d be telling my peers they need to get back on the meds and not begin the great snake-dance to victory every time a poll comes out showing something favorable to The Cause. And I’d be particularly scornful of those who are suggesting gay marriage is going to decide the election even as my candidate was ever-frantic to suggest that any talking point involving anything other than the economy was a fatal diversion.

Ed Kilgore is a contributing writer to the Washington Monthly. He is managing editor for The Democratic Strategist and a senior fellow at the Progressive Policy Institute. Find him on Twitter: @ed_kilgore.


  • c u n d gulag on May 15, 2012 11:16 AM:

    Well sure they have to cheer - no one on either side likes Romney - not even theirs!

    Romney is their least likeable candidate since Nixon - but we all know how that turned out the 2nd and 3rd times he ran, don't we?

    As for Dick Morris, he'll get his next toxic column from the same place he got his others - from Otto Hizass.

  • stormskies on May 15, 2012 11:16 AM:

    This is total corporate manipulation equaling propaganda to try to make people 'believe' that Romney is neck and neck with Obama. If we were to actually 'believe' this propaganda it would mean that about half of our population want to return to the very economic policies that has nearly destroyed our country.

    It would mean that half of our country wants to eliminate medicare and medicaid. It would mean that half of our population wants to give even more tax breaks to the rich that not only would deepen our deficit, but would mean that all those that are not rich would be taxed more.

    It would mean that all of our social safety nets would be destroyed, and that millions of more Americans would be put into poverty while at the same time removing those safety nets.

    It would mean that half of our country wants the budget proposed by pig shit Ryan to be our reality.

    And so on, and so on .......

    It's complete, manipulated, propaganda .. by CBS of course .. the home of, among others, Chip "I am not a corporate cum slut" Reid ........

  • stinger on May 15, 2012 11:33 AM:

    So they had 280 people say they were for Obama, and 282 say they were for Romney. I have to question how scientifically those 562 people were selected. Are they even able to poll cell phone users yet? (I.e., anyone under the age of 60?)

  • ComradeAnon on May 15, 2012 11:38 AM:

    Let's ask Newt. He's good at predicting the future of elections.

  • stormskies on May 15, 2012 11:45 AM:

    CBS has simply employed a methodology in order to get the 'result's that they wanted in order to create this propaganda. I mean think about how the Romney campaign has been in reality: all the lies, all the duplicity, the sheer buffoon act called Romney, the clown car of it all. That is even how the corporate media has had to 'present' this buffoon because it's all that he is. And, yet, despite this, CBC now wants to pretend that he is in front of Obama.

    And, Chip "I am not a corporate cum slut" Reid is having one orgasm and the other because of it.

  • T2 on May 15, 2012 11:49 AM:

    the election will be close, but the gay thing won't be the decider. Turn out will, as usual.
    It's much easier to get people huddled under one tent to do something than to get people under a dozen tents to do somthing. The Hate Obama message is real clear, plain and easy to respond to and all the nuts are under the same tent.
    One need look no further than WIsconsin to see that the country is clearly divided. The Dems gather a huge amount of recall petition signatures, yet Walker is still leading handily in polls...
    The GOP has worked really hard to keep the economy sucking and they will get some votes against Obama for that. Romney will be kept under wraps the entire time and no one will know why they should vote for him, other than the fact he's not Obama. That will do for at least 48%. If that sounds scary, it should. Just try to have a rational conversation with a GOPer on politics...it is virtually impossible.

  • Mitt's Magic Underpants on May 15, 2012 12:04 PM:

    I've said all along the lazy corporate media will do everything to make this a horse race. It is easier for them, and that's all they care about. Look for more crap like this in the future, as well as building up Rmny.

  • Fess on May 15, 2012 12:06 PM:

    As one of the 4 in 5 adults who don't respond to telephone polls, I have to say there's a good reason for that. The last one I actually "participated" in turned out to be a push-poll where every question was along the lines of, "If you knew that (the Democratic candidate) was a secret satanist and vampire, would you be more or less likely to vote for him?" Several more recent ones lost me at "upon completion of this questionaire, you'll be eligible for a FREE 3-day cruise to the Bahamas!!!" Right. I live in CA, by the way.

    I suppose there are legitimate polls out there, but none of them connect with me for some reason. Just the vampire ones.

  • low-tech cyclist on May 15, 2012 12:06 PM:

    of the 852 registered voters interviewed last month, just 562 were re-interviewed for this survey.

    Just for starters, I'd like to know how the 562 respondents to this second wave of interviewing responded to the first wave. Were they evenly split between the two candidates back in mid-April? Or did they lean towards one candidate or the other?

    I'd also like to know about demographic shifts in the sample as a result of the 290 dropouts between the two waves. Is the Wave 2 sample more white, more minority, more male, more female, higher or lower income, more or less educated, than the Wave 1 sample? And if there were such changes, did they do any weighting adjustments to compensate for them?

    But having said all that, we're talking about a pretty minute change in the polling results between Wave 1 and Wave 2. It's gone from dead even to a 3-point Romney lead. Neither the difference between the candidates, nor the difference between the results of the two waves, is outside the margin of error. This is a real nothingburger of a poll result.

    And as Ed said, it's just one poll, and you should be looking at polling averages anyway. It doesn't matter whether you get your averages from Nate Silver or RealClearPolitics or where - follow the averages. And if you're looking at the results of just one poll, compare it also with other recent instances where the same pollster has asked the same questions. How does this month's PPP poll compare with last month's? How does this month's WaPo poll compare with last month's? And so forth.

  • POed Lib on May 15, 2012 12:16 PM:

    I am for gay marriage. I am also for democrats to STFU about it. I don't think it helps our side in the slightest. I think that Obama's foolish statement about it will do nothing but lose Hispanics, and as such, it made no sense.

    I do hope fervently that gays now shut up, pay up, and work like dogs to get Obama re-elected. This happened to Clinton. He was elected, he listened to the gay activists, co-operated with them, and lost all of his political capital. I hope fervently that this will not happen with Obama. Gays are at most an optomistic 5% of the population. Blacks are a solid 12%, hispanics 14%. If he gained a little in the gays, I hope he did not lose a boatload in the hispanics who are culturally conservatie.

  • Equal Opportunity Cynic on May 15, 2012 12:26 PM:


    Although i am not a professional in political polling, my vague understanding is that any poll with any pretension to seriousness nowadays uses cell phone numbers in addition to land lines.

  • Peter C on May 15, 2012 12:31 PM:

    When 4 out of 5 adults decline to participate in polls, they cannot be used to predict an election. They *might* be able to explain 20% of electoral behavior, but not more.

    Polls like this are designed to tell (sell) a STORY. The 1% need a STORY to explain the results of an election after they've fixed it.

    Watch out for STORIES; disbelieve them loudly and publicly. Make sure your friends VOTE.

  • bob atkinson on May 15, 2012 12:38 PM:

    If I'm not mistaken this poll showed a one or two point preference amongst women for Romney. Now that should be quite an indication that the breakdown of the 562 people was anything but a balanced sample.

  • stormskies on May 15, 2012 12:42 PM:

    You are so right Peter C. They need to create these lies .. stories .. so as to condition the population to be able to accept the rigged election ...

  • Sgt. Gym Bunny on May 15, 2012 12:47 PM:

    Huh? I've never been asked to do a telephone poll. Wait a minute... I don't have a landline, and if a mystery number pops up on my mobile's caller ID, I don't bother answering. Nevermind...

  • June on May 15, 2012 12:51 PM:

    This CBS poll is the kind of "jounalistic" nonsense that earns itself a speedy trip to my "ignore" button.

  • gdb on May 15, 2012 1:06 PM:

    Best bet is that the gay marriage issue is a wash, politically. The biggest factor affecting the Nov 2012 vote is almost certainly "the Economy, stupid"--- or, more accurately, the "Direction of the Economy". If employment rates and the GNP are obviously improving, bet on BHO --- if obviously heading south, bet on Mittens. His weirdness with dogs and other personality traits and his anti-Keynesian policies to really destroy the economy will matter little. BHO will be not undeservably blamed by voters for the downturn who will vote ABO-- anybody but Obama.

  • Reynardine on May 15, 2012 1:59 PM:

    Don't rule out that a number of "poll results" favoring Romney may be put out there just to cover for massive voter suppression and Diebolding of the election, or even to justify (God, if such there be, forbid) faithless electors.

  • mudwall jackson on May 15, 2012 2:37 PM:

    "CBS has simply employed a methodology in order to get the 'result's that they wanted in order to create this propaganda."

    stormskies, i think you and dick morris have something in common: the same source of information. it's a shoddy poll, no doubt. but bad methodology is just that. bad methology. if you have evidence of that the poll was deliberately stacked against obama, please share it. considering the election is six months out why would they bother?

    just for good measure, please note that it is the new york times/cbs poll. i can guarantee you the new york times will never endorse the likes of mitt romney.

  • rrk1 on May 15, 2012 2:58 PM:

    In 2008, despite the tactics of the usual suspects in places like Ohio, Florida and elsewhere, the tide for Obama was so strong the election couldn't be stolen. In 2012 the story is likely to be very different. A close election is predicted. That's the first piece of propaganda. Outrageous voter suppression measures have been put in place by Rethugs all over the country, and we have yet to see what shameless tactics will be used on election day to keep certain people from voting (that's you Democrats). Then there are the computer hackers who use software backdoors to manipulate the count, A close election predicted by faulty polls is readymade to be stolen. We're being set up.

  • Robert Waldmann on May 15, 2012 8:10 PM:

    I read the CBS write up of the poll including the click for full reasults) and didn't find the key datum - which candidate did the May respondents support in April. The re-interview appears to show a very smallmshift to Romnay from tied in the fist interview in April to up 3% now (already so tiny that the news is "almost no news here). But the May sample is, of course, smaller than the original April sample, as some April respondents could't be contacted and some who were contacted refused to participate a second time. It should be very easy to look at what those who participated again in May said in April. Is the shift to Romney ispite of more attrtion of Romney supporters ? Partly due to more attrition of Obama supporters ? More than all due to greater attrition of Obama supporters ? I sure would like to know (although 3% is still just 3% -- the April sample was clearly unusually good for Romney).

    I might have missed the number I want in the write up, of course.

    My guess is that the number of people who say they changed there mind is tiny and the shift of 3% is almost exactly all due to not 100% successful resampling (that is panel attrition). This is embarrassing to pollsters for at least one of two reasons. First if responses correspond to actual voting, it suggests that there is little news in new polls (rather for normal polls new noise froindependent sampling). This would suggest an avrage of recent polls is better than the latest poll, as it is. Not good that the news is just anoher number to average. But the alternative is much worse. Few people saying they change their mind is a typical pattern called anchoring. If responses are not like the actual voting of the huge majority which is not polled, then pollsters are in deep trouble. If the interaction of respondent and canvasser matters ( as it does) polls can be completely misleading. A low number of reported revisions of intentions would draw attention to two big problems for pollsters and show that at least one is very big. I am suspicious enough to suspect that the omission of the key info (again if I didn't miss it) is related to pollster self interest.

  • The Fool on May 16, 2012 10:58 AM:

    "Panel-back surveys are conducted for a number of reasons: Some pollsters occasionally consider it instructive to interview the same universe of people before and after a specific event. In other cases, it is simply cheaper"

    This is a startlingly ignorant thing for a national Journal reporter to say. Pollsters do not do panel surveys to save money. In fact, such surveys are MORE expensive than regular ones.

    Panel studies are actually methodologically far superior when you want to measure the impact that events had on changing people's minds. You get to see all the change that happened. Let's say you have a situation where In January 60% voted for Candidate X and 40% voted for Y. Then in February one-third of Candidate X's supporters switch to support Y and one-half of Y's supporters switch to support X. If you conducts two regular non-panel polls it look slike no change at all. You still have the same proportions on each side. But in reality half the electorate changed their view. A panel study can detect that. Regular tracking polls can't.

    Cateris paribus, panel studies are methodologically superior. They are also more expensive. The National Journal reporter is an idiot.