Skip to main content

View Diary: Associated Press-GfK poll: Obama up by 15 among all adults, up by 10 among RVs (12 comments)

Comment Preferences

  •  I think the screen is too tight (2+ / 0-)
    Recommended by:
    Athenian, Stwriley

    It looks like, from my math, they cut the screen off at "always" or "almost always", and left out the part of the time folks.

    The thing is, Presidential races are actually the "part of the time" those folks show up and vote.

    That 9% of the sample is probably the difference between the one point lead here and something along the lines of the 4-5 that other polls are showing (yes, I think its conceivable that Obama is leading among this cohort by 30-40 points).

     

    •  It has evangelicals at 39 and the tea party at 32 (1+ / 0-)
      Recommended by:
      Stwriley

      Are those reasonable numbers? It sounds scary.

      Cheers

      I refuse to believe corporations are people until Texas executes one.

      by Athenian on Wed Sep 19, 2012 at 05:25:50 AM PDT

      [ Parent ]

    •  I'd say they cut deeper than almost always (1+ / 0-)
      Recommended by:
      Athenian

      The total pool of interviews is N-1512.
      The total pool of registered voters is N-1282
      The total pool of ABC/GfK likely voters is N=807.

      The first question (S4) asking for intention to vote on a 1-10 scale has "definitely vote (10) at 73%. Those voter intention numbers are in keeping with past behavior as seen in S3a (72% of all 1512 respondents).

      73% of 1582 would have left 1100+ interviews standing.
      73% of 1282 would have left 935 interviews standing.

      The pool of ABC/GfK likely respondents represents just over 53% of their total pool.

      The pool of such respondents out of 1282 registered voters is a hair under 63% of that sub-population.

      I'd conclude that another screen, beyond "are you certain to vote" has been applied. Quite what that is is difficult to say. Perhaps they only take the absolutely dead certain to vote pool which is shown in question S2 as 51% of 1512 respondents (though that would yield fewer than 800 in the likely voter headcount, so, maybe not)? Almost certain to vote would have boosted the pool to, guess what - 72%, of 1512, nearly the exact rate at which the respondents report voting in 2008.

      The resulting party ID in their constrained likely voter pool is a near even split D-R-I-(Other) 31-30-30-(8). That doesn't seem all that likely to me.

Subscribe or Donate to support Daily Kos.

  • Recommended (122)
  • Community (60)
  • Media (23)
  • Elections (23)
  • Civil Rights (22)
  • Culture (21)
  • Law (21)
  • Environment (21)
  • Trans-Pacific Partnership (21)
  • Josh Duggar (20)
  • Science (19)
  • Labor (18)
  • Economy (17)
  • Marriage Equality (16)
  • Ireland (16)
  • 2016 (15)
  • Bernie Sanders (15)
  • Hillary Clinton (15)
  • Climate Change (15)
  • Health Care (14)
  • Click here for the mobile view of the site