Skip to main content

View Diary: Exit polls (300 comments)

Comment Preferences

  •  Then the MOE is rather high, no? (none)
    If the MOE is 7%, sure.

    The Bush Budget: Borrowing a Million Dollars a Minute off your future...

    by Maxwell on Tue Feb 03, 2004 at 05:01:50 PM PST

    [ Parent ]

    •  More like 5% (none)
      Kerry's number was off by 4%, Dean's by 5% and Clark's by 3%.

      5% is a pretty standard moe.

      •  Margin of Error (none)
        Given there are multiple candidates (categories) in estimating vote percentage (rather than a simple bush-not bush or yes-no,where there is really only one variable X, since the other is simply  100 - x), shouldn't the accuracy (margin of error) be measured by a chi-square (goodness-of-fit) statistic rather than the normal distribution?

        I just come here for the fresh air like everybody else.

        by fredbellemore on Tue Feb 03, 2004 at 05:25:21 PM PST

        [ Parent ]

        •  Depends on the pollster (none)
          but, to the best of my knowledge, most pollsters use the convention that if x is the moe then any of the figures could be off by x.  One piece of evidence for this interpretation is the small sample size (usually on the order of 500-600 people) which would peg MOE somewhere in 3-5%.

          I would agree that a "goodness of fit" measure would be better, but then everyone would realize that close polls don't tell you much...

Subscribe or Donate to support Daily Kos.

  • Recommended (136)
  • Community (67)
  • Elections (25)
  • Environment (24)
  • Media (23)
  • Culture (23)
  • Civil Rights (22)
  • Law (22)
  • Trans-Pacific Partnership (21)
  • Science (21)
  • Josh Duggar (20)
  • Labor (19)
  • Economy (18)
  • Marriage Equality (17)
  • Ireland (17)
  • Rescued (16)
  • Bernie Sanders (16)
  • Hillary Clinton (15)
  • Memorial Day (15)
  • Climate Change (15)
  • Click here for the mobile view of the site