Obama polls: half full or half empty

Anti-Obama sites say that the country hates Obama and is sorry they elected him. They say that his approval rating is falling sharply. Birther Report cites a story from the Washington Examiner, and inserted one of its trademark bogus headlines:

Poll: 71% of Obama voters, 55% Democrats ‘regret’ voting for his re-election

Because of some errors in the original reporting of that story, Birther Report may not have intentionally misrepresented the data, but it’s misrepresented all the same. Actually 71% of the 10% who said they wouldn’t vote for Obama if they had it to do over said they regretted their vote, and that looks like 7.1%, not 71%. Romney voters are more confident about their 2012 choice, but then Romney hasn’t been president for them to judge.

Even with good data, the result can be spun half full/half empty. The Examiner story was titled: “Poll: Only 79% of Obama voters would vote for him again,” but an equally true headline would be: “Poll: Only 10% of Obama voters would change their votes now.” The difference is in where you put the 11% undecided.

The research was done by YouGov.com (a group I volunteer for), and it’s presented in the context of how well Romney might do in a 2016 run for President. Their conclusion is that Romney is no more popular than he was just before the election.

Believe it or not, there are still birther polls. YouGov surveyed 1000 US  adults between February 8-10. Given only the two choices, 62% say they believe it’s true that Obama was born in the US (and 38% say it’s false). Interestingly, 25% responded that people who say Obama was born outside the US don’t really believe it and are saying it just because they don’t like Obama. The hard core birthers, the ones that are sure Obama was not born in the US, were 15% of the sample. 6% said Obama’s Presidential Library should located be in Kenya.

What might be considered a striking success for birthers, though, is that a whopping 70% say that presidential candidates should be required to show their birth certificates to get on the ballot! Obama was way ahead of the curve on that one, releasing his in mid 2008.

As to Obama’s approval rating, a Washington PostABC poll before the State of the Union showed the President with a 44% approval rating (50% disapproval). That’s up 2 points since November.

About Dr. Conspiracy

I'm not a real doctor, but I have a master's degree.
This entry was posted in Birther Report, Polls and tagged , , . Bookmark the permalink.

39 Responses to Obama polls: half full or half empty

  1. Joey says:

    Gallup does daily tracking polling on Presidential Job Approval. Today Obama is at 42% approve and 52% disapprove.
    Scott Rasmussen of rasmussenreports.com also does daily tracking polls. Scott tends to over-sample conservatives (he’s a Fox News contributor). Today he has Obama at 48% approve and 51% disapprove.

  2. CarlOrcas says:

    Doc,

    How are participants selected and how do they make sure they’ve got a balanced representative sample?

    I visited the site and can’t quite figure out how it works.

  3. People volunteer to become members of the YouGov panel, then YouGov asks panel members a bunch of questions about their politics, religion, ethnicity, employment status, finances, and zipcode. When they do a poll, they randomly select people from the panel using a proprietary methodology to get a representative sample.

    The YouGov panel in the US is about 1.2 million people (2.4 million worldwide). See:

    http://research.yougov.com/about/our-panel/

    Somewhere there is a really big article comparing sampling methodologies between pollsters, including YouGov, but I don’t have it handy. See also:

    https://en.wikipedia.org/wiki/YouGov#Methodology

  4. alg says:

    Well, the polls aside, I will say that I am disappointed in Mr. Obama’s performance. That doesn’t mean I would have voted differently in 2008 and 2012. Mr. Obama was clearly the better choice in both elections. But I have always regarded Mr. Obama as a smart, yet inexperienced candidate. This inexperience has led to many mistakes in both domestic and foreign affairs.

    In 2008 I was stumping for Hillary. Frankly, I still think she would have made a better President.

  5. The section on the birther poll has been rewritten to include information from another page, a clarification, removal of the footnote and a link to the sources.

  6. I would point out that Hillary was Secretary of State.

    alg: This inexperience has led to many mistakes in both domestic and foreign affairs.

    In 2008 I was stumping for Hillary. Frankly, I still think she would have made a better President.

  7. CarlOrcas says:

    Dr. Conspiracy: Somewhere there is a really big article comparing sampling methodologies between pollsters, including YouGov, but I don’t have it handy. See also:

    In my past life I spent a lot of my time and other people’s money doing surveys and polls and was always warned about self-selecting samples.

    I’ll be interested in seeing the article.

  8. The Magic M says:

    I keep telling birthers and other right-wing fools that even a “36% approval rating” for Obama does not mean 64% will vote for “the other guy”, or for the GOP in 2014.

    (Especially in the context that Congress’ approval is way below Obama’s, and especially after the shutdown, most people identify Congress = GOP.)

    This becomes even dumber when you consider Obama isn’t running for a third term.

    RWNJ’s however delude themselves that a low approval rating means “guaranteed impeachment” or even “grounds for trying to impeach”. As if being unpopular was suddenly a “high crime or misdemeanour”.

    In birtherstan, where there is only black and white, or “100% for or 100% against”, “disapproving with Obama” translates to “considering Obama a usurper”, so that’s another delusion to bolster their “any day now” dreams.

  9. Andrew Vrba, PmG says:

    My problems with polls using such a small percentage of the population, is that they can’t really be trusted, especially when the results are present as a representation of a much larger group of people. I mean, if I poll 500 people about their opinion of meat, and 300 of those I poll just happen to be vegan, and I present the poll as “Over half of Americans dislike eating meat!”, that’s extremely misleading.

  10. Bob says:

    From a blog called Kung Fu Monkey October 7, 2005:

    John: Hey, Bush is now at 37% approval. I feel much less like Kevin McCarthy screaming in traffic. But I wonder what his base is —

    Tyrone: 27%.

    John: … you said that immediately, and with some authority.

    Tyrone: Obama vs. Alan Keyes. Keyes was from out of state, so you can eliminate any established political base; both candidates were black, so you can factor out racism; and Keyes was plainly, obviously, completely crazy. Batshit crazy. Head-trauma crazy. But 27% of the population of Illinois voted for him. They put party identification, personal prejudice, whatever ahead of rational judgement. Hell, even like 5% of Democrats voted for him. That’s crazy behaviour. I think you have to assume a 27% Crazification Factor in any population.

    ☞ LINK

  11. roxy7655 says:

    If the poll had compiled a total of how many approve of Obama’s performance PLUS how many disapprove because he “isn’t liberal enough,” I believe THAT figure would be well over 50%.

    “Disapprove” covers a lot of ground. That’s an important distinction, rarely made.

  12. aarrgghh says:

    Andrew Vrba, PmG: My problems with polls using such a small percentage of the population, is that they can’t really be trusted, especially when the results are present as a representation of a much larger group of people. I mean, if I poll 500 people about their opinion of meat, and 300 of those I poll just happen to be vegan, and I present the poll as “Over half of Americans dislike eating meat!”, that’s extremely misleading.

    error due to sampling size is a factor that can and should always be transparently accounted for with percentage points representing the level of confidence. presuming your hypothetical represents a gross error i’d suspect your sampling process was not truly random (or as random as practically possible), especially if your 500 “just happen” to be whole foods shoppers.

  13. sfjeff says:

    The Obama haters love that Obama’s approval rating is low now……but were not impressed when his approval rating was higher.

    I voted for Obama, but I have been disappointed by many things done under his name- among them the rollout of the ACA website- which was in my opinion a sign of poor management.

    I personally think its a good thing when people question their candidates performance- Birthers like to imagine that everyone who doesn’t buy into house of bovine excrement cards marches lockstep behind their imaginary Obama Fuehrer- but us anti-Birthers have shown ourselves to be far more willing to question Obama’s record than Birthers have been willing to question their Birther leaders.

  14. OllieOxenFree says:

    roxy7655:
    If the poll had compiled a total of how many approve of Obama’s performance PLUS how many disapprove because he “isn’t liberal enough,” I believe THAT figure would be well over 50%.

    “Disapprove” covers a lot of ground.That’s an important distinction, rarely made.

    I pointed this out during the ACA debates. When single payer was still on the table, the approval for Healthcare reform was very high. Nearly seventy percent. Then it was removed as a discussion, and approval dropped, but was still well over 50 percent approval when a “public option,” was still on the table. When the public option was removed in the final drafting of the law, it was only then that approval began to drop below 50 percent, and then you saw every Republican/Conservative screaming that it was now unsupported by the American people. They used the polls to claim that “a majority” of Americans did not want healthcare reform. Truth was, those who no longer supported the law, did so because it did not go far enough to reform healthcare, especially with mandatory coverage.

  15. CCB says:

    roxy7655 and OllieOxenFree, very good points. While I had been aware that polls which now show a majority disapproving of the ACA include in that majority the “socialists” who wanted a single payer system, I had not thought about presidential polls in the same light. It’s tempting to believe that Obama’s unfavorables include not only those who think he should be Cruz’s sock puppet but also those who believe that he is Cruz’s sock puppet because I voted for Obama twice.

    Does anyone know anyway to tease out of the polls the number of those who disapprove of Obama because he has not, for example, nationalized the Koch brothers’ corporations, etc.? That would be one way to test the proposition that some disapprove of Obama because he is too, gasp, conservative.

  16. Joey says:

    Andrew Vrba, PmG:
    My problems with polls using such a small percentage of the population, is that they can’t really be trusted, especially when the results are present as a representation of a much larger group of people. I mean, if I poll 500 people about their opinion of meat, and 300 of those I poll just happen to be vegan, and I present the poll as “Over half of Americans dislike eating meat!”, that’s extremely misleading.

    And yet public opinion polling of elections can be pretty darn accurate in predicting actual results. In the 2012 election cycle, more than 20 different national election polling firms conducting polls of 1000 participants with a plus/minus 3 percentage points statistical margin of error correctly predicted the results of the election.
    Here’s a link to a review of the outcomes of 2012 election polling:
    http://fivethirtyeight.blogs.nytimes.com/2012/11/10/which-polls-fared-best-and-worst-in-the-2012-presidential-race/?_php=true&_type=blogs&_r=0
    A poll that was within 3 points of the 51% Obama/47% Romney result is deemed to be “accurate.”

  17. Benji Franklin says:

    Andrew Vrba, PmG: My problems with polls using such a small percentage of the population, is that they can’t really be trusted, especially when the results are present as a representation of a much larger group of people.

    In the past, a sophisticated Pollster sincerely seeking valid results representative of the entire population can fine-tune the sampling screening so that obvious related-issue-atypical individuals do not constitute a disproportionate percentage of those sampled.

    But the problem with political polling, is that for too many election cycles now, poll results have been widely publicized along with increasing emphasis on the suggestion that the poll results from these relatively small samples, proceed to significantly effect the outcome of related elections.

    The reliability of all kinds of polling has always been subject to manipulation by the sampled respondents, if they perceived in the offered responses, a path to satisfying their own self interests. Sophisticated means of detecting such schemes and countering that effort with more screening questions, generally relies, for efficacy, on a lack of sophistication on the part of the poll respondent.

    Unfortunately, results from political polling emerge as easily discerned (by the poll respondent) positive or negative statements bearing on the candidate’s or party’s character or viability as a candidate or party.

    By this theory, Rush Limbaugh’s audience would assume that to hurt Obama, they should respond to every poll involving Obama in a way that will, (stopping shy of declaring him worse than Hitler), substantially on balance, reflect poorly on him, whether or not they personally regard those responses as accurate.

    Ditto for persons who are strongly pro-life and see Obama as promoting the murder of unborn children.

    If this theory is true, the small samples could be much less reliable than advertised, because for most public issues, the polling sophistication of poll-results-focused partisan respondents could presumably easily deceive the polling sophistication of the pollsters attempt to record only honest responses to shaded questions on the issue.

    Questions like:”Do you think President Obama really cares about Senior Citizens?”

  18. And by extension, they would say that the economy is terrible, it’s a bad time to buy a major appliance, that they think they will be worse off in a year, etc. But I think somebody who listens to LImbaugh probably thinks that anyway.

    YouGov tries to gauge political bias by asking questions, for example, of Democrats, “are you a strong Democrat.”

    By the way, I am paid by YouGov (after laboriously earning 30,000 points, I got a $25 gift card).

    Benji Franklin: By this theory, Rush Limbaugh’s audience would assume that to hurt Obama, they should respond to every poll involving Obama in a way that will, (stopping shy of declaring him worse than Hitler), substantially on balance, reflect poorly on him, whether or not they personally regard those responses as accurate.

  19. Keith says:

    aarrgghh: error due to sampling size is a factor that can and should always be transparently accounted for with percentage points representing the level of confidence. presuming your hypothetical represents a gross error i’d suspect your sampling process was not truly random (or as random as practically possible), especially if your 500 “just happen” to be whole foods shoppers.

    Dewey Defeats Truman

  20. aarrgghh says:

    Keith: Dewey Defeats Truman

    this might come as a shock but nothing in life is certain. some things however, like statistical analysis, are pretty reliable, even after 65 years, as the “unskewed polls” guy and the romney campaign painfully found out.

  21. Judge Mental says:

    Several article authors who jumped on this particular disingenuous “71%” bandwagon have now issued amendents/updates to their articles reflecting this inaccuracy in description of the subject 71%. The blame is being placed on the pollsters for the allegedly confusing way in which they originally presented the data (which seems to perhaps imply that maybe the pollsters too have at some stage slightly updated or amended the way they presented the data, but I don’t know).

  22. JPotter says:

    roxy7655: If the poll had compiled a total of how many approve of Obama’s performance PLUS how many disapprove because he “isn’t liberal enough,” I believe THAT figure would be well over 50%.

    “Disapprove” covers a lot of ground. That’s an important distinction, rarely made.

    An excellent point Roxy! What an insightful machine you are … slicing documents (and polls) into fine layers with the wittiest scan line around.

    A lot of polls provide finer detail, more comprehensive analysis, disambiguation if you will, but it’s on the back pages of the report, and that’s not what gets reported. Only the highest-level, most attention-grabbing cherry gets picked.

    Invariably, full poll breakdowns are hard to find, when made available at all. This YouGov’s full results are here … 2 links removed from the YouGov story—and it’s their own poll!
    http://d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/cbs3h7scpm/econTabReport.pdf

    You can find the approval breakdown by ideology on p. 63. I don’t see an overall “approve/disapprove because…” question tho.

    I did like the “Other – Insult” response on location of Presidential Library. It got 4%.

  23. Crustacean says:

    JPotter: I did like the “Other – Insult” response on location of Presidential Library. It got 4%.

    LOL

    “Thank you for your time, ma’am. Now, on that presidential library location, I have you down for Hell, Michigan. Is that right, or did you mean some other Hell?”

  24. There is a note on the polling site that they changed the labels to avoid confusion.

    Judge Mental: (which seems to perhaps imply that maybe the pollsters too have at some stage slightly updated or amended the way they presented the data, but I don’t know).

  25. nbc says:

    Because of some errors in the original reporting of that story, Birther Report may not have intentionally misrepresented the data, but it’s misrepresented all the same. Actually 71% of the 10% who said they wouldn’t vote for Obama if they had it to do over said they regretted their vote, and that looks like 7.1%, not 71%. Romney voters are more confident about their 2012 choice, but then Romney hasn’t been president for them to judge.

    Birthers properly doing math? Inconceivable… They are still learning their abc’s…

  26. The Magic M says:

    aarrgghh: as the “unskewed polls” guy and the romney campaign painfully found out

    I thought consensus was that Romney’s campaign knew the real numbers and just wouldn’t acknowledge them, either to avoid conceding defeat too early, or to not jinx it.
    You know, if their guy says he’s polling 45-55 with only a few days left, many people would stay at home and thus allowing the prophecy to self-fulfill.

    roxy7655: If the poll had compiled a total of how many approve of Obama’s performance PLUS how many disapprove because he “isn’t liberal enough,” I believe THAT figure would be well over 50%.

    I told WND this when they trumpeted their Wenzel polls that “X% believe Obama was not born in the US or ‘don’t know’ where he was born” (don’t remember the actual number) and led their readers to believe this somehow translates into “therefore at least those people will not vote for Obama”.

    The question was simply meaningless without a follow-up along the lines of “In the light of not knowing/believing US birth, will this influence your decision in the 2012 elections to switch from ‘for Obama’ to ‘for another candidate’?”.

    I said back then, birthers could show 100% believe in foreign birth, it still wouldn’t mean even 1% would vote for “the other guy”.

    (Of course, in the real world, this proved that the numbers were utterly wrong, because you wouldn’t have a significant part of the people believe the President was ineligible yet vote for him anyway. Or if it didn’t, it would mean “the other guy” was so bad that the majority would rather have a “usurper”. Go figure.)

  27. Dave says:

    There has been a lot of ridicule aimed at the pollsters who thought Romney would win, which is to some extent unfair (and yes, the unskewed polls guy is an idiot, I’m not talking about him).

    The argument against these pollsters goes, you do a random sampling, you account for sampling error, and you have the answer. It’s just science, and these pollsters were just in denial.

    But in predicting elections, there is a major wrinkle in this simple picture — you can randomly sample people, but there are always significant correlations between candidate preference and actually showing up to vote. And asking people “are you going to vote” is a poor predictor of who is going to vote. This is why each pollster has a “likely voter” model, and there is no clear answer to what is the right way to do it.

    So the pollsters who called the election wrong had a bad likely voter model. They were wrong, but they were not being ridiculous.

  28. CCB says:

    Dave,

    One of the ways pollsters choose likely voter models is to look at past elections. I cannot provide citations but I think I remember that the pollsters who predicted a Romney win had a likely voter model based on the recent 2010 congressional elections while those who predicted a loss had chosen a model based on the 2008 presidential elections.

    Another factor that influences all our decisions is confirmation bias. To quote Harry Nilsson in The Point, “You see what you want to see and you hear what you want to hear.” Whether we are conscious of it or not, we often make up our minds before hearing any or all of the facts and arguments and we then see and hear what supports our decision. This is easy to see in the persistence of the Birthers but it is less easy to see in ourselves.

    A really smart candidate would hire a polling firm that has worked for the other party.

  29. Keith says:

    aarrgghh: this might come as a shock but nothing in life is certain. some things however, like statistical analysis, are pretty reliable, even after 65 years, as the “unskewed polls” guy and the romney campaign painfully found out.

    Not shocking at all actually. I proposed “Dewey defeats Truman” in support of your point. The polls failed to randomize properly. If I recall correctly, they used a telephone call during the daytime – most Truman voters were not at home – they were at work.

  30. maybe the poll was twice as big as it needed to be in the first place.

    it may actually surprise some of you to learn that I actually admire some of his policies/initiatives.

    if we can just get through this birther thing.

  31. aarrgghh says:

    Keith: Not shocking at all actually. I proposed “Dewey defeats Truman” in support of your point. The polls failed to randomize properly. If I recall correctly, they used a telephone call during the daytime – most Truman voters were not at home – they were at work.

    … well cheeze, louise, y’coulda just said so inna first place!

  32. John Reilly says:

    It is possible to look at data and reach correct conclusions. Look at Nate Silver, a gay liberal Democrat working for the New York Times. He managed to get his predictions to be 100% accurate without leaving midtown Manhattan. Apparently he found a way to avoid confirmation bias.

    In 1936, Literary Digest, which had pioneered election polling, used its telephone polling technique to predict that Gov. Landon would beat Pres. Roosevelt. That was a daunting claim, seeing as the defeat of a sitting president is a rare occurrence. What Literary Digest missed was that people who did not have telephones (often because they were poor) broke very heavily in favor of Pres. Roosevelt.

    In 1948 the problem was that the polling companies, not having seen much movement during the campaign, simply stopped polling. The undecided voters then broke heavily for Pres. Truman. And they missed that. Truman supporters would have you believe that the break occurred because Gov. Dewey had cursed at a train engineer.

  33. Rickey says:

    CCB:

    One of the ways pollsters choose likely voter models is to look at past elections.I cannot provide citations but I think I remember that the pollsters who predicted a Romney win had a likely voter model based on the recent 2010 congressional elections while those who predicted a loss had chosen a model based on the 2008 presidential elections.

    Actually, there were no public pollsters who predicted a Romney victory. Only two polls taken in November had Romney ahead (Rasmussen & Gallup) and they had him up by 1%, with a margin of error of +/- 2-3 points. You can’t call those predictions when the difference is that narrow. Pew and ABC/Washington Post had the most accurate popular vote spread, each showing Obama ahead by 3 points (he won the popular vote by 3.9%).

    Romney apparently had internal polling which showed him winning, but we don’t know anything about the numbers or the methodology.

    Nate Silver, or course, didn’t focus on the popular vote numbers. He focused on the state-by-state numbers, which consistently showed Obama with a substantial electoral vote lead during the last six weeks or so of the campaign.

  34. scott erlandson:
    maybe the poll was twice as big as it needed to be in the first place.

    it may actually surprise some of you to learn that I actually admire some of his policies/initiatives.

    if we can just get through this birther thing.

    Troll in the dungeon — there’s a troll in the dungeon.

  35. interestedbystander says:

    Rickey: Actually, there were no public pollsters who predicted a Romney victory. Only two polls taken in November had Romney ahead (Rasmussen & Gallup) and they had him up by 1%, with a margin of error of +/- 2-3 points. You can’t call those predictions when the difference is that narrow. Pew and ABC/Washington Post had the most accurate popular vote spread, each showing Obama ahead by 3 points (he won the popular vote by 3.9%).

    Romney apparently had internal polling which showed him winning, but we don’t know anything about the numbers or the methodology.

    Nate Silver, or course, didn’t focus on the popular vote numbers. He focused on the state-by-state numbers, which consistently showed Obama with a substantial electoral vote lead during the last six weeks or so of the campaign.

    And thank God for Nate or my nerves would have been shattered. Didn’t even bother staying up for the results this time (I’m in UK time zone). Got up early to find it was all over. I LOVE Nate Silver.

  36. Dr. Kenneth Noisewater says:

    scott erlandson: maybe the poll was twice as big as it needed to be in the first place.it may actually surprise some of you to learn that I actually admire some of his policies/initiatives.if we can just get through this birther thing.

    There is no “birther thing” to get through.

  37. The Magic M says:

    interestedbystander: Didn’t even bother staying up for the results this time (I’m in UK time zone).

    The first US election I ever watched live (from Germany) was Bush vs. Gore, and at 5 a.m. there still was no offical result, so I felt cheated.

    The second US election I ever watched live was Obama vs. Romney, and it entertained me more than the last Superbowl (where I rooted for the Broncos).

  38. The Magic M says:

    scott erlandson:
    it may actually surprise some of you to learn that I actually admire some of his policies/initiatives. if we can just get through this birther thing.

    Just when I thought the concern troll had become extinct around these pastures…

  39. Dave B. says:

    And oddly enough, in the comments on this yougov page
    https://today.yougov.com/news/2014/02/12/poll-results-barack-obama-birth/
    you can’t say “Wong Kim Ark.”

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.