"Why
Not Use Experiments that Change
One Factor at a Time"
November
19, 2002
A
Presentation by Gary McDonald
Dr. McDonald is the former Head of the Operations Research Dept. at General
Motors Corp., and the former Director of GM’s Enterprise Systems Laboratory.
He first pointed out the experimenters’ preference for rapid feedback over
statistical considerations as motivation for their use of such designs.
Dr. McDonald then described the engineer’s interest in One-Factor-At-a-Time
designs, and the practical difficulties and modeling inadequacies of such
designs, e.g., their geometric imbalance, and gave examples.
Dave Doane
giving Gary McDonald a Certificate of Appreciation from the Detroit Chapter