September 15, 2004

Please. No Wagering

As Fraters pointed out earlier today, the Star Tribune's latest Minnesota Poll is showing John Kerry with a nine point lead over the President.

The left is tittering with glee; they think they'll hold Minnesota this year.

And they may just.

But Powerline was pointing out the flaws in the Minnesota Poll two years ago.

And so was I. This bit was from five days before the 2002 election:

In the meantime, the poll shows Mondale leading Norm Coleman by 8 points, 47-39.
Eight point in five days?

The Minnesota Poll should come with a disclaimer: "For Entertainment Purposes Only".

Posted by Mitch at September 15, 2004 03:28 PM | TrackBack

GOP chairman Ron Eibensteiner pointed out the Minnesota Poll's historial bias four days ago. He called for the resignation of the poll director Rob Daves.

Star Tribune Editor Anders Gyllenhaal defended the poll in a prepared statement. "We have been operating one of the most respected and accurate polls in the country for more than half a century and always have been completely open and forthcoming about how we conduct our polls," Gyllenhaal said. He called Eibensteiner's personal attack on poll director Rob Daves "shameful and misdirected."

Gyllenhaal could get a job as Dan Rather's spokesman. No other poll, not Zogby, not AP, has anywhere near the MN Poll's 9 point spread.

Posted by: Gideon at September 15, 2004 07:12 PM

Actually, all polls should come with that disclaimer--polling is still more art than science. And while I think Kerry is winning Minnesota, I agree that nine points just seems like way too much (just as I think an 8-point lead for Bush nationally seems like way too much.)

So while polls are fund to dissect, and give us some idea of kind of where we sort of stand, we all know they don't necessarily reflect reality.

Oh, and the Mondale poll? Lest we forget, it was taken in the immediate aftermath of the Rallemorial, before it really began to crystalize just how bad a mismanaged fiasco that was. I will maintain to my dying day that if I could hop in a time machine, hogtie Rick Kahn, and hide him in a basement, that Mondale wins that election and the Democrats hold the Senate in 2002. (This is not to say that the Minnesota Poll isn't off from time to time--I think the Pioneer Press/MPR poll has a better track record, and more of a conservative tilt. Go figure.)

Posted by: Jeff Fecke at September 16, 2004 12:43 AM

"Actually, all polls should come with that disclaimer--polling is still more art than science."

Tell Kos.

Posted by: mitch at September 16, 2004 06:56 AM

Check out the PP poll today. It's got Bush ahead at 46-44. Can you say that the Strib is a bit of an outlier? Nobody's as bad as they are at predicting reality. Prediction: at the last minute they'll get closer to everybody.

Posted by: nerdbert at September 16, 2004 09:16 AM

Nothing should be done about the Star Tribune poll.
DFLers think it is accurate, giving them false confidence. Get out the vote efforts are less effective if you think that your side is already ahead.
Republicans know the Strib polls are biased already.
Sounds like a win-win to me!

Posted by: Steve Meyer at September 16, 2004 09:48 AM

"Can you say that the Strib is a bit of an outlier?"

No, nor can you say the Pi Press/MPR is an outlier. Without further data, we can only conclude that one or both polls is wrong. The most likely answer is that Kerry leads by 2-3%--but I wouldn't wager money on that until another poll shows up.

Posted by: Jeff Fecke at September 16, 2004 10:17 AM

"No, nor can you say the Pi Press/MPR is an outlier."

You're right! I should have said out-liar! Mae culpa!

On an only slightly more serious note, the CNN/USA Today/Gallop poll showed the same thing as the PP/MPR poll -- a statistical dead heat. Two polls with a history of being at least somewhat accurate agree, while the one with a long history of inaccuracy disagrees. How seriously should we then take the methodology of the Strib poll?

Posted by: nerdbert at September 17, 2004 12:36 AM