Author: Bill Funk Date: Jul 6, 2007 18:42
On Thu, 05 Jul 2007 19:16:37 -0700, SMS geemail.com>
>What _are_ often useful from CR are their surveys, since they have large
>sample sizes and good statistical methodology. Their vehicle reliability
>surveys, and wireless carrier surveys have always been spot-on, jiving
>with the results of other surveys.
The sample that CU depends on is far from random (which would be a
good statistical methodology).
It consists only of those who get the survey (readers of CR), and
then, only those who choose, for whatever reason, to reply.
The idea that readers of CR represent all parts of the consuming
public is unrealistic, as is the idea that those who choose to return
the surveys also represent a cross-section of the consuming public.
THIS IS A SIG LINE; NOT TO BE TAKEN SERIOUSLY!