Skip to main content

View Diary: Zogby "interactive" polls are junk (322 comments)

Comment Preferences

  •  What is your point? Zogby is as good as others on (0+ / 0-)

    your list because none of them was accurate. Zogby poll is as good as any other polls you mentioned whether it be by 3.3 to 8.3%.

    You failed to show which poll had it right or within 0.5% as the polls in Europe do.

    I don't buy the argument that Zogby is junk because it showed that Hillary is tanking.

    •  Very well. (0+ / 0-)

      I'm in my fifth statistics course now, have an MS in applied mathematics and writing an article which faces peer review that is a statistical analysis.  

      Zogby is junk because an online survey is easy to freep.

      A margin of error simply denotes the width of the range that a given percentage (usually 95%) of survey results taken from the same population at the same time with the same methodology are expected to fall within.  

      The margin of error pronounces no judgment upon the quality of the methodology or how representative of the population the survey sample is.  

      A 1% margin of error at 95% confidence is reasonable, but by lowering the confidence requirement, the margin of error can be narrowed.  For example, if I don't like the 6% MOE that 95% percent confidence offers, I can settle for 90% confidence and get an MOE closer to 4%.  

      Example:  I go to Netroots Nation this summer and sample hundreds of attendees on their choice for President.  Then I survey 300 Austinites on their choice [1].  The results come back with 74% favoring the Democrat and 22% favoring the Republican.  My margin of error is 8% at 95% confidence, so I settle for 90% confidence and get a 5% margin.  The most rigorous and up-to-date mathematics support this;  I even use a Rao score interval (2000) instead of the Wald interval.

      Yamaneko

      [1]  Methods of determining margins of error tend to be written assuming that considerable proportions of respondents choose each of the two possible responses, e.g. 54% vs. 46% or 31% vs. 69%.  The math breaks down at Cheney levels of approval and I'm a bit less comfortable than my professor at Bushian levels.  Thus the random Austinites.

      Dems in 2008: An embarassment of riches. Repubs in 2008: Embarassments.

      by Yamaneko2 on Wed Nov 28, 2007 at 11:35:28 AM PST

      [ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site