Friday, September 28, 2012

Poll denialism


People are starting to make some connections:
One of the odder little subplots of the 2012 election has been the growth of poll denialism among Republicans. As Mitt Romney's chances have grown ever dimmer, a cottage industry has sprung up on the right claiming that presidential polls suffer from liberal bias and Romney is really doing better than they say. "When the published poll shows Obama ahead by, say, 48-45," explains conservative pundit Dick Morris, "he's really probably losing by 52-48!"

Now, this is hardly in the same league as climate denialism or evolution denialism.
Or vaccine denialism or rape denialism or . . . well, you get the idea. Here in America, one of our major parties has responded to the revolution in personal communication that began with 24 hour cable news and progressed to smartphones and the blogosphere by developing an independent, self-reinforcing, weather-dominator-scale doomsday device of cognitive dissonance.

There is no unpleasant reality, the right has discovered, that cannot be shouted down by 128-bit quad-core sophistry. The latest fact to be dragged, gagged and bound, towards the conservative memory hole is Romney's dismal prospects in the upcoming elections:
congressworksforus posted a comment in On Polls and Polling · 2 days ago
I can guarantee you that the Romney campaign has access to far better polling than any polling organization that's posting public polls.
The fact they are NOT changing course simply tells me that Erick is wrong, and that the public polls are horrible this time around.
The fact that Obama has (if you pay attention) essentially given up on being re-elected (could he possibly make any more mistakes than he has the past couple of weeks) tells me that his campaign knows it is toast as well.
Turnout will favor Republicans. The GOP built a ground game after 2008 that was effective (but not overwhelmingly) in 2010, but was *incredibly* effective a few months ago in Scott Walker's recall election (he got more votes than when we was first elected!)
Believe me, Obama is not winning Ohio, nor is he winning Florida, and without either, he ain't winning. 
Yep, we'll just take your word for it, anonymous Restate commenter. Poll denialism is rich with the sort of useful idiots who feel that good old horse sense and five minutes on WUWT qualifies them to dismiss radiative physics. Chait's take here
This was the week that the political world discovered the burgeoning world of conservative polling denial. Just like other, better-established fields of conservative reality denial, the polling denial movement has its own levels of insanity. At the core sit the most fanatical of the denialists, like, a popular site that offers its own twist on public opinion data, which currently has Mitt Romney leading Barack Obama by 7.8 percent nationally.
The poll denialists’ argument holds that the polls — all of them, except Rasmussen, conducted by a right-wing pundit with a terrible record of accuracy — are over-sampling Democrats, finding nearly as many of them as showed up at the polls in 2008, which they consider a high-water mark for Democrats unlikely to be repeated. Pundits have patiently explained that polls do not make assumptions about the party identification of voters but merely report what voters tell them. And the most plausible explanation for the higher number of Democrats in polls is that increasing numbers of conservatives who reliably vote Republican are identifying themselves as independents to pollsters.
So poll denialism is silly, and the conspiratorial explanation undergirding it is deranged.
Steven Taylor is on the same page:
I am astonished at the degree to which many who are rooting for Romney seem to be in total denial about the polling.  For example, the following from Katrina Trinko at NRO:
But regardless of partisan breakdown, Republicans should be wary of taking any polls as completely accurate.
“Part of the reason the Democrats won in 2008 was that when it looked as if McCain was going to lose, some Republicans stayed home,” argues McLaughlin. “So if President Obama is in a dead-even race with Mitt Romney in so many swing states, if the Democrats can convince enough Republicans they’re going to lose, it could take a one-point loss for the Democrats to a one-point win.”
Emphasis mine.
The thing that is remarkable about the above is that it not only in based in an approach that privileges preference over reality, it comes with a built-in fairy tale to explain any non-preferred results!  Using the logic above the polling can be wrong whilst predicting the actual outcome and, better yet, the wrong polling (that was actually right) wasn’t just wrong, but it caused the wrong outcome to occur!
Life would be better for all of us if we were all, regardless of partisan preferences, a tad more grounded in empirics.
Can I just say I wish the pundits and national media types would be as quick and forthright in confronting climate science denial? But of course, the difference is that most reporters don't understand climate science, while the concept of calling people up and asking them who they are going to vote for is a little easier to wrap your mind around. And again, as we saw with Todd Akin's comments, the immediate and obvious connection between the ideological worldview to the facts being denied helps people see deniers for what they are. The polls are brutally discouraging to my candidate = I deny the polls; the most scrupulous practitioner of journalistic false balance has got to be able to connect the dots on that one.


  1. This all from a guy who believes it doesn't matter who you sample to get your survey results? No wonder your "ideological worldview" allows you to accept the status quo of main stream media polls.

    Go take a look at the details of the analysis of the 2008 election polls. A few pollsters got it right. The liberal MSM was way off--too optimistic. (And don't bother quoting what others said about the analysis. Go look up the paper for yourself. There's a nice table showing the details.)

    We're in echo-chamber land these days. I've given up on finding any significant group that cares about integrity. This nation is toast, as predicted long ago. It just isn't obvious enough quite yet.

    (Sixty years ago, one person summarized 18th century Scotsman Tytler's observations this way: "A democracy cannot exist as a permanent form of government. It can only exist until the majority discovers it can vote itself largess out of the public treasury. After that, the majority always votes for the candidate promising the most benefits with the result the democracy collapses because of the loose fiscal policy ensuing, always to be followed by a dictatorship, then a monarchy." -- sure looks true to me.)

  2. Ehhh, I won't have time later so here's an immediate hint. It will be interesting to see what you do with these facts...

    The article you quote claiming Rasmussen's results are awful is based on the idea that the spread between candidates is more important than anything else.

    Here's a snippet of how that worked out in reality for 2008 (the most-cited study about accuracy is by Panagopoulos. It's easy to find.) Note that all polls are rounded to whole percentages, while results typically are to a tenth:

    1) Real result: 45.7 vs 52.9
    2) Two polls nailed it (error less than 0.5 and whole-number spread was correct.
    3) Two polls had the R value correct and D rounded down instead of up (52 instead of 53, spread=6.)
    4) The other polls were worse. For example, the best of the rest had the spread correct at 7 but the absolute values at 44 and 51 -- off by two for both candidates.

    Believe it or not, this last poll was rated as "most accurate"... even above two that nailed both candidates... while Rasmussen was downgraded severely for having a spread of 6. His poll was in category #3 above.

    When pundits snipe about denialism based on silliness like this, it seems to me it isn't Rasmussen who is ignoring reality. I could dream that there would be science-minded folks who let the numbers speak for themselves without torture... but that seems beyond hope.

    1. I suggest you look at Nate Silver's post, "Poll Averages Have No History of Consistent Partisan Bias":

      Unfortunately it's not enough to see yourself as science-minded. You actually need some understanding of what you are talking about. Without that, it is sheer folly to declare the great bulk of the professionals to be obviously wrong.

      Rasmussen is awful. They have a terrible reputation, and a strong Republican house effect. But even if they were an average pollster or even a good one, it would still be an idiot move to exalt them and vilify a dozen different pollsters because at this point you happen to like their numbers better.

  3. In a similar vein from CNN - vaudeville journalism from Fox.

    Your take is probably even better :D

    1. Like many Republicans, Romney has struggled to find popular musicians who will let him use their work. Given Fox's theme of "There's plenty of time; he'll catch fire any day now" I might humbly suggest Chairmen Of The Board:

      Give me just a little more time
      And our love will surely grow
      Give me just a little more time
      And our love will surely grow

      Life's too short to make a mistake
      Let's think of each other and hesitate
      Young and impatient we may be
      There's no need to act foolishly
      If we part our hearts won't forget it
      Years from now we'll surely regret it


      You're young and you're in a hurry
      You're eager for love but don't you worry
      We both want the sweetness in life
      But these things don't come overnight
      Don't give up cos love's been slow
      Boy, we're gonna succeed with another blow

      Give me just a little more time
      And our love will surely grow
      Baby please baby
      Baby please baby

      Love is that mountain we must climb
      Let's climb it together your hand in mine
      We haven't known each other too long
      But the feeling I have is oh so strong
      I know we can make it there's no doubt
      We owe it to ourselves to find it out


      Give me just a little more time
      And our love will surely grow
      Baby, please baby
      Baby, please baby

      [Repeat And Fade]