Skip to main content

View Diary: WI-Gov: Scott Walker takes leads, but sits right at the 50% mark (275 comments)

Comment Preferences

  •  tommy you're not being rational here (0+ / 0-)

    There is nothing unusual about this special election that argues for some kind of "different" or "unique" likely voter screen, or no likely voter screen at all.  

    Your arguments are not remotely plausible.

    This poll is fine.  It might be wrong, the difference from last time might be statistical noise, and yes the screen could be off and skewed the turnout model.  But there's no reason to think there's some "better" way to do it that is clearly going to produce "more accurate" results.

    All the Wisconsin polls, not just PPP's, have been strangely volatile this year, I think reflecting an unsure electorate.  We've seen Walker on the ropes in a couple rounds of polls, and Walker modestly leading in others.

    But there's no reason to dismiss PPP's methodology here.

    44, male, Indian-American, married and proud father of a girl and 2 boys, Democrat, VA-10

    by DCCyclone on Mon Apr 16, 2012 at 07:18:13 PM PDT

    [ Parent ]

    •  So then make a case (0+ / 0-)

      Why is the turnout 31D/32R?

      No armwaving now, just state why they chose a likely voter model with a 31D/32R turnout model.

      Further state why likely voters in the June recall are lower in both Dems and Reps than either 2008 or 2010.

      Cut to the chase, you can't.  There is no logical argument that can be made that says both Dems and Reps are much less likely to vote in June, and indies are way, way more likely to vote.

      Screening for likely voters is good, but the only likely voter screen that makes sense for a race like this is self-selection.  Eliminate voters who say they won't vote.  Don't conjure up an absurd distribution that has no chance of occuring.

      Mr. Gorbachev, establish an Electoral College!

      by tommypaine on Mon Apr 16, 2012 at 07:29:08 PM PDT

      [ Parent ]

      •  It's just noise, and we have history here... (1+ / 0-)
        Recommended by:
        rosarugosa

        ...for just these types of "problems" in PPP polling, almost always proving irrelevant to the accuracy of the toplines.

        I played this game myself the last cycle, and got burned.  I kept questioning turnout models and such, based on party or race or ideology, and it turned out most of the time that PPP still came out consistent with the totality of polling far from an election, and proved accurate up close.  That doesn't mean there weren't any misses, NV-Sen was a big one, but they were actually a bit better than other polls even there, regardless of the turnout models.  I distinctly CO-Sen as one where the Hispanic vote share and party ID jumped around a lot across polls, but the topline always showed a dead heat, almost always 3 points or less either way.

        Another thing is that "adjusting" the turnout model based on party ID or other demographic categories doesn't change the topline much, often hardly at all.  That's something a lot of people don't realize, but it's always true when you actually do the math.

        Again, yes this poll could be wrong, as even the best public polls are wrong at least 5% of the time.  But trying to pick apart its internals to prove the point is folly, and I learned that myself through making the same mistakes I see you making here.

        44, male, Indian-American, married and proud father of a girl and 2 boys, Democrat, VA-10

        by DCCyclone on Mon Apr 16, 2012 at 08:19:17 PM PDT

        [ Parent ]

Subscribe or Donate to support Daily Kos.

  • Recommended (126)
  • Community (58)
  • 2016 (50)
  • Environment (38)
  • Elections (36)
  • Media (34)
  • Republicans (32)
  • Hillary Clinton (30)
  • Law (29)
  • Jeb Bush (28)
  • Culture (27)
  • Barack Obama (26)
  • Iraq (25)
  • Trans-Pacific Partnership (25)
  • Civil Rights (24)
  • Climate Change (23)
  • Labor (19)
  • Economy (19)
  • LGBT (16)
  • Health Care (15)
  • Click here for the mobile view of the site