Skip to main content

View Diary: 'Likely voter' polling screens were skewed toward Romney (111 comments)

Comment Preferences

  •  Not after the fact (7+ / 0-)

    we knew unlikely voters were turning out in the early vote. At that point, I adjusted my predictions to the RV models.

    If our early vote looked poorly, I would've stuck with LV.

    So we had additional data that could help us determine which model was more likely -- actual votes being cast.

    •  I get the analysis of early voting (0+ / 0-)

      ...helped adjust models, but while you were doing that the GOTV campaigns were getting out more unlikely voters, and especially in same-day registration states.

      My argument is that polls do not obviate the need for boots on the ground actually getting out the vote.  And the topline poll numbers often mean less energy is devoted to downticket races that are not being polled.

      Likely voter screens were skewed towards Romney only because OFA was dooing something that most pollsters had not methodological way of handling.  If you found how to validly incorporate that methodological difference, more power to you.  I would be very careful about expecting similar results in an off-year election unless there is substantial intensive and coordinated GOTV to the standards OFA set this year.

      50 states, 210 media market, 435 Congressional Districts, 3080 counties, 192,480 precincts

      by TarheelDem on Fri Nov 23, 2012 at 12:25:17 PM PST

      [ Parent ]

      •  This analysis by Sinsinger undercuts your argument (3+ / 0-)
        Recommended by:
        IM, sethtriggs, PrahaPartizan

        as it also goes back to 2010 - when Democratic voter involvement decreased, and even considers 2004, when OFA was not even a glimmer in a progressive's eye.

        ASSUMPTION #1: Likely voter polls are more accurate than ones of registered voters

        ASSUMPTION: Incorrect

        Of the 50 state presidential polls conducting during the final month of the 2004 and 2008 presidential campaigns, the RV result was closer to the final outcome than the LV result in fully half of them. In just 38 percent of them was the LV screen closer to the final outcome than the RV screen. In six of the polls, incidentally, there was no difference between the RV/LV results in a poll.

        For what it is worth, those numbers track closely to a study I did of downballot polling in advance of the 2010 midterm elections.

        There is a real issue with likely voter screening - one which, if the pollsters wish to maintain some semblance of relevance, some degree of sterling reputation, they need to address.

        Instead of whining about how those assessing their predictive value are co-dependent on the polls themselves - as Newport does to try and defend Gallup's badly tarnished reputation.

        •  Missed Singiser article link (0+ / 0-)
        •  Didn't understand my argument, did you (0+ / 0-)

          My argument is that GOTV to turn out "unlikely" voters is more important than  proving the obvious point that likely voter screens can be way off because of faulty assumptions about that year's electorate.

          Emphasis on GOTV.

          But a likely voter screen that can incorporate early voting data as estimates of likely voters is going to be closer to the mark than a likely voter screen that ignores already voted unlikely voters and treats them as nonvoters.

          50 states, 210 media market, 435 Congressional Districts, 3080 counties, 192,480 precincts

          by TarheelDem on Fri Nov 23, 2012 at 02:40:18 PM PST

          [ Parent ]

          •  Oh, I got it, I just disagree with the conclusion (0+ / 0-)

            (implicit) that GOTV was the only thing explaining failure of LV screens to yield accuracy.

            •  By definition (0+ / 0-)

              ...the people who showed up in the polls being a different population than the those assumed in the likely voter screen is the total explanation for the the inaccuracy.  I agree that GOTV was was not the only factor.  There were errors built into the sampling process, those built into the wording of the questions used to identify who was likely, and twiddling of the demographics to conform with the pollsters' model of who likely voters ought to be.

              And then there is always a plain old random outlier that is taken as a trend in interpretation.

              50 states, 210 media market, 435 Congressional Districts, 3080 counties, 192,480 precincts

              by TarheelDem on Fri Nov 23, 2012 at 03:58:46 PM PST

              [ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site