Skip to main content

View Diary: Back in the polling game! (144 comments)

Comment Preferences

  •  What, this data? (0+ / 0-)

    From the 2009 exit polls for the NJ and VA governor races:

    57% of NJ-Gov voters approved of President Obama's job approval.

    48% of VA-Gov voters approved of President Obama's job approval.

    And what results did PPP get in their last survey for each race (released the day before the election)? Try 45% in NJ and 41% in NJ. They only missed the mark by 12 and 7 points, respectively. No biggie right? Right???

    Then of course you have NY-23 which they said Hoffman was pulling away with after releasing a poll the day before the election showing him up by 17 points. I don't think I need to remind everyone that Owens, the Democrat, ended up winning that race by 2 points.

    •  Off-cycle and special election? (0+ / 0-)

      While I certainly appreciate you bringing data to the mix, those are pretty edge cases.  If you'll recall, NY-23 was the one where the Republican dropped out like 3 days before the election and then endorsed the Democrat, and the 2009 elections were in an off-year where the turnout percentages are usually really messed up, and you're likely to have issues with a good likely voter screen (that election day wound up being less Democrat-unfriendly than most anticipated).

      Further, the NY-23 case is already in Mr. Silver's scoring system, and is a non-trivial factor in PPP's above-average score already.

      I'm not saying we should excuse those errors, but I don't see why they should outweigh Nate's analysis of their whole recent history.

      •  I'm focused on internals, not the topline... (0+ / 0-)

        ...numbers for these examples.

        The internals of their polls are screwy. Their weird sampling results in inaccurate secondary questions and approval/favorable numbers.

        Since the 2008 election, we have only been presented with 3 cases to judge the accuracy of pollsters: NJ-Gov, NJ-Gov and the special election in NY-23.

        In all three cases PPP failed miserably when looking at the internal numbers--especially with regards to approval/favorable numbers.

        I don't trust PPP because they haven't proven themselves to be a reliable pollster. A lot of that probably has to do with them being an automated pollster, but that's still no excuse for being so wrong.

        •  I don't know about that. (0+ / 0-)

          First of all, I don't know that there's any reason to believe that the exit poll favorability numbers are more or less accurate than the pre-election polling favorability numbers.

          You seem to be hyper-focused on personally disagreeing with unverifiable internals, and ignoring the verifiable final horse race numbers.

          •  You seem to forget all the fuss that was made... (0+ / 0-)

            ...leading up to those 2009 races about how it was a "referendum on President Obama", "a check on his presidency", etc., etc.

            The actual polling showed he had little or no influence on the election and was still rather liked by the voting public. In fact, the VA-Gov electorate was made up of McCain +8 voters and yet the President still achieved a 48% approval rating among them.

            The media rarely focuses on the topline number because the internals are more fun for them to play around with. That leads to BS CW thinking which turns out to be very wrong come election time.

            PPP has been doing it a lot recently. In fact, they have a recent post on their site claiming the reason Democrats are down in swing states is because of the President.

            How do they claim that? Well, their approval numbers for the President are low in those states so Dem deficits must be all his fault, no?

            They claimed the same thing with the NY-23 race and were proven wrong. They had the President's approval at just 39% in the CD (he won it by 8 points in E2008--51/43).

            •  fact check (0+ / 0-)

              Here's what Jensen actually wrote:

              Obama won all of these states in 2008 but his unpopularity in them now is a major contributor to the problems Democratic candidates are having in them this year.

              I'm inclined to disagree with that for a couple of reasons (the most important may be that Jensen doesn't even mention economic conditions). But as unsupported as "a major contributor" is, it doesn't equal "the reason" or "all his fault."

    •  The facts (4+ / 0-)
      Recommended by:
      Terri, boadicea, BruinKid, xgz

      Last PPP poll of VA-GOV:

      McDonnell 58, Deeds 40

      Actual results:

      McDonnell 58.62, Deeds 41.26

      Last PPP poll of NJ-GOV:

      Christie 47, Corzine 41, Daggett 11

      Actual results:

      Christie 48.5, Corzine 44.9, Daggett 5.8

      So I don't know what numbers ("12 and 7") you THINK you're citing, but they're clearly not based in any reality.

      •  As I said, I'm not talking election results... (0+ / 0-)

        They've been fairly accurate with regards to final election results (but so were others). I'm talking internals: Electorate make-up, approval and favorable numbers, etc. The "12" and "7" numbers are very much based in reality (as I noted earlier):

        Their final NJ poll pegged the President's approval at 45%.
        Actual result: 57% (57-45 = 12)

        Their final VA poll pegged the President's approval at 41%.
        Actual result: 48% (48-41 = 7)

        •  those numbers can't be directly compared (0+ / 0-)

          PPP uses a different approval question than the exit poll does. For one thing, it offers a "not sure" option -- which 10% of New Jersey respondents chose. It's actually rather remarkable that you didn't mention that.

          Also, the four-point scale of approval used in the exit poll (including "strongly" and "somewhat") tends to elicit higher levels of approval than the binary choice of approve/disapprove. This has been commented on for years.

          You may have some good examples of sketchy internals, but this certainly isn't one of them.

    •  Anecdotes (0+ / 0-)

      Nate looks at every released poll.

      You cite some polls.

      I trust every more than some.

      In theory, there is no difference between theory and practice; but in practice, there always is a difference. - Yogi Berra

      by blue aardvark on Mon Aug 09, 2010 at 05:41:32 AM PDT

      [ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site