Tuesday, June 13, 2006

A Poll Autopsy – How CBS Manufactures Its Own Evidence

CBS News released a poll this morning, claiming a 2-point drop for President Bush’s Approval numbers even after the death of the vicious terrorist Musab Zarqawi. It seems the "Fake But Accurate" meme is still hard at work at the Tiffany Network. These numbers are so out of line when compared to the other available polls, that I was immediately suspicious. Also, CBS only polled “adults”, not even asking for registered voters. Further, the respondent pool only numbers 659 adults, significantly below the threshold for satisfactory credibility in a national poll. The poll was taken June 10th and 11th, not even the full weekend. CBS does not verify that its “nationwide” poll made any effort to reach a representative cross-section by geography or age or race; no reference is made to the demographic characteristics of the respondent pool., which casts strong doubt on the methodology used in this poll. The collective impact is the creation of a hit piece, especially when the internal data from this poll, or rather its lack, is compared to previous polls in which CBS took part. The one demographic cited by CBS is that they counted 196 self-identified Republicans in the poll, against 244 self-identified Democrats and 219 self-identified Independents. CBS re-weighted the poll to count only 192 Republicans, against 236 Democrats and 231 Independents, or 29.1% Republican, 35.8% Democrat, and 35.1% Independent. This is a deliberate skewing of party representation, invalid in the face of the 2000 Census, exit polls from either the 2002 or 2004 Federal elections, and common sense in the fact that Republicans are the majority party in both houses of Congress.


Kathy said...

DJ - Thanks for analyzing this poll. The internals for it are awful. It was as someone once said, Statistics are like a light pole to a drunk, more for support than enlightenment.

Jimbo said...

What else would you expect from the network of Dan Rather