🚀 go-pugleaf

L. H. Garlinghouse wrote of his concern that Juran's Quality Handbook now
pays obeisance to Shainin's PRE Control Charts and inquired as to my
reasons for seeking their interment. They are several but as one of
mine--the first--is the lack of peer review, I shall avoid most charges
of hypocrisy by citing the article by Ledolter and Swerson, "An
Evaluation Pre-Control" in the Journal of Quality Technology 29, 2, 1997.

Legend
# confirmed by this writer
* Ledolter and Swerson

1) #* Shainin, et. al. consider the technique as proprietary or
intellectual property and thus purportedly free of the obligation to
undergo peer review
2) * A series of simulations on normal, normal contaminated and
exponential distributed variates showed the error rate for false signals,
missed signals and ARL to be substantially higher than competing
methodologies, Shewhart and CUSUM to be precise.
3) # Calculation of the ARLs for various processes shows the procedure to
be inferior to Shewhart Charts (sample size = 2)
4) # * The presentation of the method (in print) assumes the
implementation of the method operates in an intellectual vacuum, i.e no
other methods of analysis or training are at hand.
4) * The technique is particularly feeble when dealing with low
capability processes.
5) #* Owing to it's lack of a statistical representation, one cannot
distinguish common from special causes.
6) * it provides no useful pedagogical path for using ever more powerful
techniques of process control.
7) * It is a univariate technique only.

(This is a superficial restatement of Ledolter and Swerson's article
which is written at a level suitable for almost anyone with any training
in the mathematical foundations of SPC.)

Garlinghouse has interpreted my remarks as suggesting the technique is
used frequently. I have no independent knowledge of this (Ledolter and
Swerson apparently do.) I've been asked about the technique on numerous
occasions but by would-be clients who were well-enough informed to both
ask about it and be concerned about the educational level of their
employees.

Garlinghouse also refers to these charts as "rainbow charts." I freely
recommend to all who are engaged in the establishment of SPC (and who use
computers) to use the colored bar technique in the training and early
implementation stages. This does work well.

Finally--and probably unnecessary--I must doff my figurative hat to
Dorian Shainin and his progeny whose contributions to the practice of
Quality Control are significant. Their legacy over the last 50 years is
substantially greater than that of but a handful of others. Their jewels
of achievements just do not need this bit of paste.

Frank Isackson
re:SOLUTIONS
FIsackson@aol.com