🚀 go-pugleaf

L. H. Garlinghouse posted (in part):
>...Wheeler's empircal rule
>goes into great detail to explain that there is NO assumption about
>the normality of the distribution when you use the empircal rule....
I am not familiar with Wheeler's empirical rule and I would not be
surprised to find that it works in many cases. My question is does he
tell you when it won't?

My argument is why use such a rule where a simple rigorous approach will
work as in the case of the data presented by Gunter in Quality Progress
in March of 1989, "The Use and Abuse of Cpk, Part 2". He presents 80
measurements of flatness from a machining operation. The data are clearly
non-normal and yield a Cp of 2.4. However, a simple transform, the
reciprocal, provides an excellent fit and shows a true Cp of 0.97.
Correspondingly, the estimated ppm defective rises to 2,000 from 0.1.
This recalls Twain's comment on the difference between a lightning bolt
and a lightning bug.

>...there is a strong liklihood (at
>least in my mind) that the original poster, or certainly many list
>members, do not know, care nor have the time to care about the REAL
>way to answer the question.  They have about 15 minutes (the
>attention span of the Top Dog when it comes to questions of Quality
>and statistics) to give an answer.
Apparently my experience has included a more incisive collection of
hounds with longer attention spans. It surprises me that the Quality
management profession still has a large population of individuals who
have not learned the lessons of presentation from their occasionally
dreary confreres in the accounting world. None of the collies, shepherds,
spaniels and Chihuahuas I've encountered over the years would settle for
one number to characterize their P & L statements much less their balance
sheets. So they get an assortment of numerical descriptions. Each of
these numbers is more obscure than its predecessor--if you don't know or
care what they mean. Quality managers who cannot present their
no-more-obscure numbers during their 15 minutes of fame (or infamy) at
the monthly or quarterly review meetings need to recall an old rhetorical
dictum. "If you have nothing interesting to say, find an interesting way
to say it."

>In my 35+ years in manufacturing I have NEVER run accross a Plant
>Manager or even a VP of Mfg. (other than at a Quality Seminar) that
>had any understanding of even the most basic concepts of SPC much
>less capability indices.  I have run accross maybe three (3) Quality
>Managers that did.  My point is that as simplistic as it is,
>Wheeler's empirical rule will be more than accurate enough for many
>industrial purposes....
Again, my experience with manufacturing management has been far more
salutary. I repeat my point: never use a shortcut when a rigorous
approach exists that takes no more time to execute. This is especially
true if the risk of making a wrong decision is significant as in Gunter's
example.

I recommend setting up a simple spreadsheet that will analyze data for
non-normality, apply suitable transforms, test for goodness of fit and
then give a suitable Cp. It will be a few hours well spent--once--to set
it up. Many quality managers have powerful statistical analysis packages,
e.g. Statview, JMP, SAS, SPSS, etc. which can do far more and such folks
need not have to resort to IBM's Lotus 123 or Microsoft's Excel.

>In some cases the world is probably better served by the Quality
>professional using his/her disposable time to work on the
>psycho/political dimension of SPC and Quality rather than
>spinning their wheels with arcane statistical methods, that while
>valid, etc. are beyond the needs of the person asking the questions.
>In so much of industry (at least the parts of it I am familiar with)
>a "good enough" answer is a significant breakthrough.
I suspect I'm not preaching to the choir but to the dogged. But to those
who find themselves between the terror of the "arcane" and the fear that
"good enough" might not be I offer the following advice from Henry V:

In peace nothing so becomes a man
As modest stillness and humility.
But when the blast of war blows in our ears,
Then imitate the action of the tiger.
Stiffen the sinews, summon up the blood,
Disguise fair nature with hard-favour'd rage;
Then lend the eye a terrible aspect.

To all I repeat my Swahili benediction (which roughly translated means: I
wish that you will go forth and do well.)

Kwenda mzuri,

Frank Isackson
re:SOLUTIONS
205 West Walnut Avenue
Unit A
Monrovia, CA 91016
626 358 1340 Voice
626 358 1322 Fax
FIsackson@aol.com