Tuesday, December 13, 2005

Maybe "Zogbyfied" will someday be an adjective

Zogbyfied: used to denote a poll as meaningless because of surveyor bias, especially if the surveyor accepted money to support a conclusion. Usage example: "That latest CNN poll is so Zogbyfied that you can't trust it."

John Zogby recently released a poll supposedly showing most Americans have an unfavorable view of Wal-Mart. It naturally got a lot of attention from the Wal-Mart-bashing mainstream media. Reader Marshall Manson pointed me toward Joel Mowbray's column "Polling a Pollster" in the Washington Times, which exposed Zogby's anti-Wal-Mart bias. Even Dan Rather would probably agree that it's hardly an "unimpeachable source" that accepts money to testify against Wal-Mart, then conducts polls. In politics, a judge would have to recuse himself. In academia, one preferably wouldn't even touch the topic, lest the methodology used be questioned.

I don't know how Zogby's reputation ever survived his idiotic prediction of a Kerry landslide. With a hat tip to Marshall, here are some other pieces critical of Zogby:

John Zogby's Creative Polls (details Zogby's history and biases)

Voters vs. Zogby: a once-golden pollster is tarnished (about Zogby's recent polling fiascos)

A Snowy Graveyard For Pols And Polls (explains problems with polling practices)

Remember what I've said before: polls are meaningless compared to actions, because talk is cheap; people can give any answer they'd like without having to weigh the consequences. Moreover, polls can be conducted poorly. The example I've used before is higher gasoline prices' effect on SUV sales. Survey after survey can say that gas prices won't affect sales, but what counts in the end are the actual sales numbers. Similarly, the surveys about the 2004 presidential campaign were meaningless compared to the actual casting of ballots.

2 Comments:

Blogger Mike said...

Three kinds of lies: lies, damn lies, and statistics.

Wednesday, December 14, 2005 4:57:00 AM  
Blogger Perry Eidelbus said...

Yes indeed. It's the first quotation in the great book "How to Lie with Statistics," which my father read back in the 1950s. I have a modern reprint, and it is as valid as ever.

Wednesday, December 14, 2005 2:24:00 PM  

Post a Comment

Subscribe to Post Comments [Atom]

Links to this post:

Create a Link

<< Home