Proprietary Statistical Methods

Tue, 19 May 1998 12:47:56 -0400
"Emil M Friedman"

In response to your request in your Amstat News column, I have two
comments.

I think that the item II.B.6, "Keep all statistical methods in the
public domain...." is a little too general. The phrase, "...specific
implementations...may be proprietary" is too narrow to cover a case
where a data analysis or experimental design method would give one's
employer a competitive advantage.

I don't understand the phrase (II.A.1) "transparent treatment".

Dr Emil M Friedman emfriedman@goodyear.com
Goodyear Technical Center - Dept 410D

Opinions are my own, are subject to change, and do not represent
Goodyear's.

Preliminary response from Dr. Gardenier. That item derives from W.
Edwards Deming's code of conduct for his consulting business. Dr.
Friedman is right to raise this issue. If an employee devises a new
method on the job, either general employment law or the specific
employment contract may dictate that the method is the intellectual
property of the employer. Any suggestions on how the Committee could
reword this item appropriately?

"Transparent" treatment of data handling means that the reader should be
able to understand clearly and accurately how the data were selected and
processed from reading the report. The opposite is to have the reader
misled by the report into thinking that the data selection and treatment
was honest and appropriate to the study when, in fact, it may have been
biased by selection or treatment or both. Would another phrasing express
this better?