Saturday, August 7, 2010

David Warren: More errors and retractions

David Warren’s correction at the end of this article does not address the other erroneous information in his piece about the BP spill. And unfortunately, what precedes his most recent retraction includes yet another error.

Warren writes: “Rasmussen's trick consists of pegging people by the nature of their work or station in life, into these two broad classes”.

Rasmussen’s “trick” (perhaps the only accurate characterization) does not define the “Political Class” or “Mainstream” by “the nature of their work, or station in life”, as Warren claims. It collects no such information to reach its “conclusions”. Rasmussen’s “trick” is as follows, from its website:

"The Political Class Index is based on three questions…

The questions used to calculate the Index are:

-- Generally speaking, when it comes to important national issues, whose judgment do you trust more - the American people or America’s political leaders?

-- Some people believe that the federal government has become a special interest group that looks out primarily for its own interests. Has the federal government become a special interest group?

-- Do government and big business often work together in ways that hurt consumers and investors?

To create a scale, each response earns a plus 1 for the populist answer, a minus 1 for the political class answer, and a 0 for not sure.

Those who score 2 or higher are considered part of the Mainstream. Those who score -2 or lower are considered to be aligned with the Political Class. …In practical terms, if someone is classified with the Mainstream, they agree with the Mainstream view on at least two of the three questions and don’t agree with the Political Class on any."

This is the kind of data Mr. Warren would like to see replace Statistics Canada. The problems are obvious enough, as discussed here:

“The Political Class (Rasmussen) defines occupies a mere 7% of respondents. With this gimmick he can report nonsense like the Tea Party numbers above. After all, who really cares if 13% of the 7% of Elitists don’t like tea? If he surveyed 1,000 people, then we’re talking about 9 who fall into that category. That’s 9 out of 1,000, or less than 1%.

...This has absolutely no statistical value whatsoever. The only purpose it can possibly serve is to create an imaginary group against which to compare other results whose significance you wish to artificially enhance. This permits Rasmussen to imply that an invented class of people have decidedly different values than the rest of us. And since his test for what constitutes a Mainstream American is so broad, the whole process is worthless...”

Correction, please.

No comments:

Post a Comment