polls: Obama Loses 2/3rds Of His Lead In One Week

interesting article : Another Poll Shocker – Obama Loses 2/3rds Of His Lead In One Week

and very interesting reader’s comment in the discussion:

Here’s an anecdote. I received a robo call from Rasmussen earlier this year (during the primaries). The reason I picked up the phone was because the name “Rasmussen” came up on the caller ID!! Gee, I wonder who this is? OK I’ll play. The robo call then led you through a series of questions, to which you responded by typing in a number from your phone (e.g. “on a scale of 1 – 9 please describe yourself, very liberal = 1, very conservative = 9″). You could, of course, pass yourself off as anything from Bob the rightwinger to Bob the progressive! Since you don’t talk with a live person, it would be difficult to detect mischief with this approach.

So the question is – How accurate is a “random” poll when poll-ee can refuse to answer (i.e. not pick up the phone) or, knowing who the pollster is (as in my case) can deliberately give false answers (I didn’t BTW)? The traditional measures of “margin of error” go right out the window, and the accuracy of such polls should be questioned. And I’m sure the collection techniques of other pollsters are just as bad as Rasmussen’s…

comment by archtopon 30 Oct 2008 at 3:32 pmon