UnSkewed Polls Keeps on Sucking

UnSkewed Polls, the blog that thought it was correcting the Democratic bias inherent in most polling data, and that so many conservatives faithfully believed would rub Nate Silver’s nose in the mud, just keeps on acting as though they truly believe that the correct methodology is whichever one most successfully confirms…er, “confirms” their hypothesis.

This time it’s a post-election survey, asking whether Obama and the Democrats fraudulently stole the election.  Here is the the “first question of the survey. (click on image to see it in full screen).

UnSkewedPoll

A rather leading structures and the available answers are skewed to produce an outcome unsupported by any actual evidence.

Another question asks you to list your income range, with their categories being:

  • Under $30,000
  • $30,000 – $99,999
  • $100,000 – $199,999
  • $200,000 – $249,999
  • $250,000 or more

They really need to unskew their categories.

They also ask if you think the mainstream media was actively advocating the re-election of Barack Obama, as a simple yes/no question, without asking the same question for Romney.

And let’s talk about their sampling method… Not only is UnSkewed Polls visited primarily by conservatives and the methodologically illiterate,* but the invitation on the front page reads;

AAA_Unskewed

Biasing the respondents before they enter the poll? In the immortal words of Walter Sobchak, “They’re a bunch of fucking amateurs!” Imagine how the right wing would respond if Public Policy Polling invited people to take their poll by opening with, “Do you think Republicans tried to illegally suppress the minority vote? Take our survey!”

Dean Chambers the wizz behind UnSkewed Polls poll unskewing method has at least admitted that Nate Silver was right and that his own adjustments to the polls’ numbers of Democrats and Republicans sampled was wrong, and has even lost confidence in the conservative pollsters who got it wrong.

“Nate Silver was right, and I was wrong,” Chambers said in a phone interview… “…those assumptions — my assumptions — were wrong.” …

But he said he probably won’t go back to “unskewing” polls next time. He actually thinks conservative-leaning pollsters like Scott Rasmussen have a lot more explaining to do.

“He has lost a lot of credibility, as far as I’m concerned,” Chambers said. “He did a lot of surveys. A lot of those surveys were wrong.”

And yet he hasn’t really learned the lesson that good methodology matters. Fuckin’ amateurs.

______________________________________________________________
*Or so I presume; how could anyone who’s either a non-conservative or who understands methodology stomach frequenting the place? I only did so to build up a set of readings on the Nate Silver flap for my Methods students, so I can show them real-world effect of knowing one’s ass from one’s elbow when it comes to polling and statistical analyses.

Advertisements

About James Hanley

James Hanley is Associate Professor of Political Science at Adrian College and a Fellow of the Institute for Social Policy and Understanding. The views expressed here do not reflect the views of either organization.
This entry was posted in Politics in General. Bookmark the permalink.

5 Responses to UnSkewed Polls Keeps on Sucking

  1. Dr X says:

    It would appear that Dean Chambers knows absolutely nothing about surveys and polls.

  2. Troublesome Frog says:

    I’m optimistic about polling. Polling accuracy is getting more and more attention these days, and your call is either right or wrong. Sooner or later, reputations will suffer and bad actors will be squeezed out.

    At least, I hope so. It’s clear that political commentators can make whatever predictions they want and never suffer blows to their reputations. But polling companies don’t really have personalities or followings. They just have numbers and predictions. If those are gone, there’s not much left.

    I love watching the numbers people win, especially when people notice. It gives me hope for humanity.

  3. Scott Hanley says:

    I was intrigued by the whole polling debate. I was confident that Silver, et al’s methodologies would be good, if the raw polling data was relatively unbiased. And I knew that the side that seemed to be losing would be full of accusations and excuses. But if it turned out that the polls were biased, then it would be a fascinating sociological game to figure out how that happened.

    One reason I saw offered for Romney’s confidence was that they saw they were polling well among self-described independents, but the the polls didn’t reflect as many Republican voters as they expected. So it was easy to conclude “We’re doing well in the middle and the polls aren’t counting all of our people.” Post-election, a plausible explanation might be that many people who still vote Republican aren’t eager to wear the brand name any more.

  4. James Hanley says:

    Post-election, a plausible explanation might be that many people who still vote Republican aren’t eager to wear the brand name any more.

    I hadn’t thought of that. It seems very plausible.

  5. Scott Hanley says:

    I played around a few minutes with the polling function at HuffPo, looking at party identification over the last few months of the election. In the last two weeks there’s a strong break for Independents to declare for one party or the other, and it does look like they were more R than D. Lends a little support for the thesis, anyway. Also, looking over a two-year period, Republican identification has been noticeably more voluble than Democratic identification.

    http://elections.huffingtonpost.com/pollster/party-identification-adults#!mindate=2012-01-01&maxdate=2012-11-06&smoothing=less&estimate=custom

Comments are closed.