Those Unskewed Republican Polls


Alex Pareene of Salon.com links to this website, which changes the demographic weighting of public opinion polls to conclude that Mitt Romney is actually a huge favorite to win the election in November. Unskewedpolls.com is all the rage on the right these days. They take a whole bunch of polls, change the makeup of the polling samples, and declare that Mitt Romney is actually leading by 8 points nationally.

Here’s how this works. The polling companies have to decide how many self-declared Democrats, Republicans and independents to include in their sampling pool, with the goal being to have a pool that is as representative of the American voters as they can make it. They do this by using other polls that ask people which party they identify with and, like it or not, more Americans identify as Democrats than Republicans. But Dean Chambers, the creator of that site, doesn’t believe that.

So he takes those polls and reweights them based on the polling sample used by Rasmussen, a polling company that always skews Republican. But even Scott Rasmussen says that what Chambers is doing is invalid:

Even the founder of Rasmussen Reports, whose surveys often show higher Republican numbers, cast doubt on Chambers’ methods: Scott Rasmussen told BuzzFeed in an e-mail that “you cannot compare partisan weighting from one polling firm to another.”

“Different firms ask about partisan affiliation in different ways,” explained Rasmussen. “Some ask how you are registered. Some ask what you consider yourselves. Some push for leaners, others do not. Some ask it at the beginning of a survey which provides a more stable response while others ask it at the end.”

That’s why it’s important to do what Nate Silver does so well, really dig into the assumptions of these polls to weight them accurately. So I propose that Dean Chambers should make a wager with Nate Silver on which one’s predictions comes closest to the actual results of the election. And make the wager big enough to actually matter. I’m guessing that Silver would be happy to do it. Would Chambers? It depends on whether he actually believes this stuff or if he’s just found a new con.

Comments

  1. trucreep says

    Every statistic and poll you see should be taken with a grain of salt. Republicans and Democrats, right left center – doesn’t matter.

  2. says

    This sort of Woo just works by making people feel like they are at least doing something even if totally irrati…. sorry think I just confused this post with the last one.

    But I suppose that doesn’t really matter since all Woo is pretty-well interchangeable.

  3. Abby Normal says

    Every statistic and poll you see should be taken with a grain of salt.

    I don’t care for added salt. I’ll subsitute Mrs. Dash instead.

  4. Gvlgeologist, FCD says

    I’m really looking forward to seeing what he says… after the election. I’d love (if I could stomach it) to see what they say on Cons….pedia (can’t even bear to spell it).

  5. Ben P says

    I’m really looking forward to seeing what he says… after the election. I’d love (if I could stomach it) to see what they say on Cons….pedia (can’t even bear to spell it).

    I think that’s exactly it.

    I’d guess that the motivation behind this is about 30% trying to help Mitt Romney win the election by encouraging conservatives, and about 70% laying the groundwork for after the election to say that “Well, Obama only won because the liberal media kept sticking it to Romney and showing that Obama was ahead. Look, these properly re-weighted polls show that Romney was ahead.”

  6. Die Anyway says

    I get two or three of those calls per week. They always ask “Do you consider yourself Republican, Democrat or Independent?” Since I am registered as a 3rd party affiliate, none of those answers is correct. Then they ask if I’m going to vote for Romney or Obama. I’m most likely going to vote for the candidate from my party and not for either of the two choices they give me. It ends up being a totally bogus poll.

  7. ajb47 says

    “Proofiness: The Dark Arts of Mathematical Deception” by Charles Seife contains a whole chapter on how polls get manipulated or are just plain wrong. And it all gets waved away as “+/- 3%”. It takes a whole lot of work to get polls correct, starting with the wording of the questions and the allowed answers.

  8. vmanis1 says

    Perhaps after the election, they will claim that the Obama campaign bussed in people from Indonesia and Kenya, and had them impersonate U.S. citizens in a grand orgy of voter fraud. I can’t wait to hear Michelle Malkin’s explanation.

    (If you think bussing people in from these countries is impossible because of the oceans in between, please wait for the new site, Republican Geography, that shows the real geography of our planet, shorn of all the liberal `oceans’ that supposedly separate continents.)

  9. Phillip IV says

    Ben P @ #5:

    I’d guess that the motivation behind this is about 30% trying to help Mitt Romney win the election by encouraging conservatives, and about 70% laying the groundwork for after the election to say that “Well, Obama only won because the liberal media kept sticking it to Romney and showing that Obama was ahead. Look, these properly re-weighted polls show that Romney was ahead.”

    I think it’s really mostly about not being able to face up to reality. The GOP is beginning to fall into the demographic hole they’ve been warned about for years, but the consequences of facing that reality (move back to the middle or permanent regional party status) are simply and truly inconceivable for the teabaggers – the idea of deviating from orthodoxy is worse than then end of the world for them. Literally for those of them who actually look forward to the end of the world.

  10. raven says

    Is there such a thing as too much lying?

    There might be.

    When you start lying to yourself and believing it, things can go wrong.

    That little kid who said the priest touched her, is just making it up.

    It’s just a cough, it doesn’t bleed that much.

    George Bush was a great president. Obama crashed the economy and bailed out Wall Street.

    Romney/Ryan don’t really mean it when they say they will cut your social security and Medicare, and cut taxes while the deficits climb ever higher. And start a huge war with Iran.

    Romney/Ryan will win because god is on our side and we lie a lot.

  11. raven says

    It looks like lying has become so important, so sacred a ritual, and so ubiquitious to the christofascists/Tea Partiers, that they can’t stop no matter what.

    It’s hard to say where this ends up.

    1. As a long term strategy, it doesn’t look that effective.

    2. To normal people, it doesn’t look all that good. The truth still has some value to some people.

  12. baal says

    thanks #9, vmanis1, I laughed in my cube :).

    I’ve been hearing this story as one of the talking points of this week. Yep, if you only ask card carrying republicans, Romney is ahead. So very astonishing!

  13. says

    The polling companies have to decide how many self-declared Democrats, Republicans and independents to include in their sampling pool, with the goal being to have a pool that is as representative of the American voters as they can make it. They do this by using other polls that ask people which party they identify with and, like it or not, more Americans identify as Democrats than Republicans.

    Actually, most don’t do this. They do often weigh for other demographic factors, such as age, gender, and race, but usually not party ID. That’s because party ID is fluid, and if you try to weigh for it based on what you think it should be, you could end up missing the fact that it’s changed and pushing your numbers in the wrong direction. Instead, they use demographic data from the census, which is much more reliable and unlikely to whipsaw in the span of 2 years.

    That said, you’d have to think that every single pollster — not a few, not most, but all of them — have completely flubbed their weighting procedures by a minimum of 5 points and more like 10-15 points in order for this joker to be correct. Not only that, but all of these elementary mistakes go in the same direction. To say that this is implausible is an understatement.

  14. The Lorax says

    Remember people, all of their statements, they’re not technically lies. Like every claim, there’s always a grain of truth…iness.

  15. DaveL says

    I think it’s really mostly about not being able to face up to reality. The GOP is beginning to fall into the demographic hole they’ve been warned about for years, but the consequences of facing that reality (move back to the middle or permanent regional party status) are simply and truly inconceivable for the teabaggers.

    I can’t remember if it was about the popular vote in 2008 or an approval rating afterwards, but I remember them complaining that including the black vote made Obama’s share “artificially” larger than his “real” share. Because blacks’ votes didn’t really count.

  16. matthewfedder says

    That’s why it’s important to do what Nate Silver does so well, really dig into the assumptions of these polls to weight them accurately.

    I would say that he does NOT dig into the assumptions of those polls. His goal is to create a consensus of the polls, and assume there is “wisdom of the crowds” in the combination of all polls.

    Polls are weighted by historical precision, and adjusted based on whether historically they have been biased. While corrections are factored in for likely-voter vs. registered voter and (I believe this year for the first time) all phones vs. landline-only, he does not look at the polls’ assumptions to see if there are any methodological problems to correct for.

    For example, Ras polls get bumped down 1-1.5% (IIRC) because of their historical bias towards Republicans, but get weighted pretty highly, because once you account for that bias they have in the past performed well. They’re NOT bumped down due to partisan weight directly – only indirectly, because it pushes them below the consensus of pollers. .

  17. says

    I’d guess that the motivation behind this is about 30% trying to help Mitt Romney win the election by encouraging conservatives, and about 70% laying the groundwork for after the election to say that “Well, Obama only won because the liberal media kept sticking it to Romney and showing that Obama was ahead. Look, these properly re-weighted polls show that Romney was ahead.”

    It’s also laying the foundation for conspiracy theories about voter fraud. If Obama wins by roughly what the polls say — and whoever wins, the outcome will be very close to the aggregate of the polls — then either the poll denialists were wrong, or someone skewed the actual vote itself. These are people who have a very hard time admitting being wrong.

    The douchebag at unskewedpolls has put forth such an outlandish conspiracy theory claiming that ACORN (yes, the nonexistent ACORN) will commit fraud to a degree precise enough to match what the polls are saying. He says this after swearing that it’s not a conspiracy theory.

  18. Chiroptera says

    trucreep, #1: Every statistic and poll you see should be taken with a grain of salt.

    If by “taken with a grain of salt” you mean “should not be accepted without doubt until verified by other, independent researchers,” then this is true. In fact, it is true for all results from all experiments and observations in all fields, including the natural sciences.

  19. anteprepro says

    I get the feeling that ACORN will become the new Illuminati in another decade or so.

    Remember people, all of their statements, they’re not technically lies. Like every claim, there’s always a grain of truth…iness.

    Perhaps they should stamp the entire Unskewed Polls site with “Not Meant to be a Factual Statement”.

  20. oranje says

    Rather obvious name, isn’t it? Like Fox News using “Fair and Balanced” as a tagline, which by default implies the problems with other organizations. I give them credit for continuing to try to invent a new reality. Sisyphean task, that.

  21. jba55 says

    Every time I see a poll all I can think of is Homer Simpson saying “Oh, people can come up with statistics to prove anything. 14% of all people know that.”

    oranje: “I give them credit for continuing to try to invent a new reality. Sisyphean task, that.”

    I don’t know if it’s that hard, I have an ex girlfriend who lives in an alternate reality. It looks exactly like this one except she is never wrong, regardless of the evidence.

  22. dogmeat says

    That said, you’d have to think that every single pollster — not a few, not most, but all of them — have completely flubbed their weighting procedures by a minimum of 5 points and more like 10-15 points in order for this joker to be correct. Not only that, but all of these elementary mistakes go in the same direction. To say that this is implausible is an understatement.

    On the plus side, he’s stumbled upon the ultimate interstellar drive system, the “Infinite-Improbability Drive!!!”

  23. Randomfactor says

    Even Rasmussen will stop weighting their polls so blatantly soon, so as not to be TOO far off from the actual results.

    If they don’t pretend to be non-partisan some of the time, their value to the Republican Party goes down.

  24. fastlane says

    Is there a real world point to these? Does exaggerating the points motivate the (R)’s to go to the polls because they think it wouldn’t be a wasted effort or something?

    Or is it really all just denial?

    I suppose there might be a bit of both in this, but I’ll be damned if I know. Any other guesses?

  25. caseloweraz says

    “Unskewedpolls.com is all the rage on the right these days. They take a whole bunch of polls, change the makeup of the polling samples, and declare that Mitt Romney is actually leading by 8 points nationally.”

    And now we know what rationalization hamsters is/are really all about.

    Too bad those wheels are only metaphorically spinning. If they were real, we wouldn’t have to worry about alternative energy.

    BTW: The term “rationalization hamsters” dates from 2010, according to Wiktionary.

    http://en.wiktionary.org/wiki/Citations:rationalization_hamster

    Third item: “The rationalization hamster is an analogy for the thought processes used by women to turn bad behavior and bad decisions into acceptable ones to herself and her friends.”

  26. says

    DaveL:

    I can’t remember if it was about the popular vote in 2008 or an approval rating afterwards, but I remember them complaining that including the black vote made Obama’s share “artificially” larger than his “real” share. Because blacks’ votes didn’t really count.

    That’s absurd. They clearly do count, but only three-fifths as much.

  27. says

    Interesting take… The polls are using pretty close. I think GOP is just desperate. They had to have known when they put forth a moderate that Ron Paul people were bolting… The fact that Romney is losing should be no surprise at all.

Leave a Reply