RE:Developer of XmR chart

[Kromkowski@aol.com]

>>The curious thing is that neither Dr. Deming nor Shewhart ever used and X and MR chart - at least not that I have found. However, when Dr. Deming did apply a control chart to individual values, he always grouped them into a subgroup and then constructed the average and range chart. Dr. Shewhart did the same.<< 

Is this correct? What about the fires example Ch. 11 OOTC, p 325.

I have information regarding a December 1949 Fortune 
Magazine, which noted:  "Statistical Quality Control .. is
among the sharpest management tools developed in half a
century."  I don't have article from which this quotation 
comes but I wonder if it sheds any light on the orgin of 
chart. 

In addition what about the citations by Deming in OOTC Ch. 
11 notes 12, 13.  This seems to point in part to Tippett in 
Bometrika 17 (1925), as a very early source.


I have long been interesting in the derivation of the 
coefficients used.  I once remember an explanation in a 
book by Logothetis, I think was spelling, but I don't have 
book, nor can a remember the name.  But I have a 
recollection that the book was published by the British 
Deming Association.  David K. once told me that he 
remembered the author but that he went back to Greece.

I would find an explanation on the DEN of the derivation of 
the d2 or E2 (which is obvious just 3/d2) very useful and 
perhaps enlightening.

Before, I knew about Control Charts or Deming, in a very 
brief middle school teaching career while I was in law 
school, I got in the habit of converting raw test scores 
and the sums of multiple test scores into z-scores (I was 
then a total autodidact, "uncorrupted" by formal statistics 
courses) and I then was using below minus 2.5 
(which "seemed" about right to me) as a means of confirming 
my suspicions about students in need a special help.  Of 
course, this was in some ways akin to what he says is 
totally wrong at p. 340, OOTC,  "One sees much wrong 
practice ....  It is totally wrong to take any number of 
pieces such as 8, 20, 50, or 100, measure them with 
calipers or other instruments, and take 6 standard 
deviations of these measurements as the capabiltiy of the 
process.  The first step is to examine the [available]data, 
as by a run chart(CH. 1), or by x-bar and R-charts, to 
decide whether the process of manufacture and the system of 
measurement show statistical control.  If the do, then the 
capability of the process will be obvious from the x-bar 
and R-charts.  If the do not, then there is no capability."

JDK


Message posting through the Clemson CQI Web Server.