Skip to main content

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

Comment Preferences

  •  Polimetrix (1+ / 0-)
    Recommended by:
    Larry McAwful

    Good discussion here on the methodology

    •  Thanks. (1+ / 0-)
      Recommended by:
      George

      I'll check 'em out.

      A conservative is just a liberal who hasn't needed a second chance yet.

      by Larry McAwful on Wed Nov 28, 2007 at 08:22:48 AM PST

      [ Parent ]

    •  Read it and it still fails (1+ / 0-)
      Recommended by:
      Geotpf

      Here is what is problematic:

      The Polimetrix approach to political polling is to draw a true random sample from the list of registered voters provided by election officials then go into their panel of volunteers and select the closest possible match for each sampled voter based on a set of demographic variables. They then interview each panel member that best "matches" the randomly selected voter.

      Basically, what they are saying is that everything is random UP UNTIL they select the people in their "registered group". The argument is that it a random sample up until they inject bias (because that is the nature of a registered group).

      They go on to say:

      The matching uses a complex statistical algorithm (something so complex that MP finds it difficult to decipher much less evaluate).

      Niiiice. A proprietary algorithm that no one can evaluate because it is difficult to understand.

      But then they go on to state what I got from my initial reading of the above paragraph:

      The end result of this process may approximate a random sample, but does not produce a "true" random sample.

      Approximation meaning it is similar to but not totally unlike a "true" (love those quotes guys!) random sample.

      Boil this down to: "I can't believe it's not random sampling!"

      Internet polling: It's bad enough to build a business model! (http://www.polimetrix.com/)

      Please note:

      Readers should know that my arrangement with Polimetrix provides complete editorial freedom. I can write whatever I want, and no one from the company has ever tried to influence, direct or even review anything I have written (including this post).

      Pollster.com is inherently biased towards Polimetrix.

      When war is considered to be more noble than peace - we have lost everything.

      by feloneouscat on Wed Nov 28, 2007 at 08:46:10 AM PST

      [ Parent ]

      •  Whatever (0+ / 0-)

        Do you think that everyone on a telephone pollster's random sample answers the phone and takes the poll?

        •  Uh, noooo, but that make it MORE not less random (1+ / 0-)
          Recommended by:
          Geotpf

          What Polimetrix are trying to do is create a pseudo-random sample. They even admit it is not a true random sample.

          So here are the facts:

          1. Pollster.com is not capable of answering questions about the validity of internet polling because there is an arrangement with Polimetrix, which does internet polling
          1. Pollster.com ADMITS that Polimetrix methodology does not produce a true random sample. This violates the whole point of polling and generates a truly biased result.
          1. Were all polling to adopt Polimetrix methodology, god help us all - polling is used extensively in Public Health. Were we to limit it to those with computers only, the bias would be heinous.

          Then there is there comparison:

          Internet numbers:
          White: 1067
          Black: 145
          Hispanic: 80

          Telephone numbers:
          White: 563      
          Non-White: 88      
          Hispanic: 50      
          Black: 21

          WTF?!? Notice there is an extra category in the telephone? Yet they have the effing GALL to say you can do comparison?

          This just doesn't fly. Period.

          You may say "whatever", I say it is a joke.

          When war is considered to be more noble than peace - we have lost everything.

          by feloneouscat on Wed Nov 28, 2007 at 10:35:49 AM PST

          [ Parent ]

          •  OK (0+ / 0-)

            You think that a self-selected subsample of a random sample is more random than random.

            •  Never mind... (0+ / 0-)

              I've been corrected by my wife who knows more about sampling procedures than I do. I was looking at it from the wrong end. Being brain dead on THAT one...

              My apologies for looking like a dumb ass...

              When war is considered to be more noble than peace - we have lost everything.

              by feloneouscat on Sat Dec 01, 2007 at 11:39:37 AM PST

              [ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site