Skip to main content

View Diary: The 2012 polling hall of shame (269 comments)

Comment Preferences

  •  Yes (1+ / 0-)
    Recommended by:

    the question will be was it "luck" or good methodology. At the very least it seems it can't be dismissed out of hand.

    •  The answer is pretty clearly luck (0+ / 0-)

      Which is really lucky because it should be much better than that.

      The weighting procedure that Rand used was screwy. Not biased, but it generated a huge margin of error because it assigned too high weights to a few individuals. That in turn made the reported results really noisy. Unless they cleaned up the final poll, they got really lucky.

      Had they followed proper weighting methodology (in particular trimming weights from their raking procedure) they would have had a much more reliable measure. They even have a paper that suggests using panel estimation techniques to measure vote intent, which would be even better. They will release all the microdata after the election to academic researchers. We will then get a really good read on Rand.

      •  What? (0+ / 0-)

        I thought that Rand had Obama by about 3.3% and he won by more than 2.4%. That's better than a LOT of the pollsters listed here. So why so dismissive?

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site