The problem with polling is because of the response rate. A generation or two ago people were honored that someone would solicit their opinion. No so today, for whatever reason. Two days ago the Pew Foundation revealed the percentage of those persons contacted who are willing to give their opinions. Take a guess as to what that percentage is: According to Pew, their "success rate" is nine percent, or one out of every eleven persons they speak with. Thus they cannot get a reasonable random sample and the Central Limit Theorem does not apply. Therefore most polls are GIGO.





Speak your mind

3 Comments so far

  1. Ed on November 6, 2012 4:14 pm

    The question then is, what type of bias does this introduce into the sample? People with time on their hands? Lonely people? I personally never even pick up the phone anymore unless I know the number. Just not worth the hassle.

  2. Steve on November 8, 2012 11:30 am

    Cui bono? Those conducting the survey and selling the results or those participating in the survey?

  3. Ardent1 on November 8, 2012 4:35 pm

    I disagree with the author’s views. I am a FIRM believer in the Law of Large Numbers and two simple facts support this: (a) Nate Silver’s recent performance and (b) the WSJ’s ongoing quarterly survey of Economists — how an individual Economist is off but the average prediction of all the Economists surveyed is better, respectively.

    If it truly was GIGO, Nate Silver’s performance would have been no different than a tossing a coin. N. Taleb devoted his first book to this subject by calling it “Fooled by Randomness” because people don’t fully understand randomness.

    Finally, the Obama camp got it right so please don’t use the “sour grapes” excuse when someone did their homework.



Resources & Links