Deming Questions # 17

gogue@chailly.ensmp.fr (GOGUE J.M. Societe MAST 39 50 99 67)

Tue, 18 Jun 96 10:48:44 +0200

--------------------------  DEMING QUESTIONS  -------------------------- 

A weekly paper including a set of questions copied in the Deming book 
OUT OF THE CRISIS   Chapter 5.  Questions to Help Managers.

By Jean-Marie Gogue

-----
# 17
-----

1.   (No 33 in the Book) - Knowledge about your customers' relations with 
your services

a)    Do your customers think that your product lives up to your expectations? 
What did your advertising and your salesmen lead your customers to expect? 
More than you can deliver? How do you know?

b)    (If applicable.) Are your customers satisfied with the service that you 
or your dealers provide? If yes, what is satisfactory about it? The quality of 
workmanship? The lag between your call and appearance of the serviceman? How 
do you know?

COMMENTS
Page 179 of Out of the Crisis we read : "In the olden days, before the 
industrial era, the tailor, the carpenter, the shoemaker, the milkman, the 
blacksmith knew his customes by name. He knew whether they were satisfied, 
and what he should do to improve appreciation for his product... With 
expansion of industry, this personal touch is easy to lose... But sampling, 
a new science, steps in and pierces that barrier."

Dr. Deming said that in front of an audience of 24 Japanese top managers in 
Tokyo, July 1950. We know that this historical conference (the text is 
available) was the spark which started the Japanese industrial revolution.

The Japanese chief executives understood the Dr. Deming's message. Listening 
to the Voice of the Customer is everybodie's job. Sampling techniques can 
help. The American chief executives did not understand the Dr. Deming's 
message. They believed they had to create new categories of specialists 
(Quality Control, Marketing etc.). Wrong.

2.   (No 34 in the Book) - Everybody sees the quality of your product with 
his own eyes

a)   How do you distinguish between your quality as your customer perceives 
it and quality as your plant manager and work force perceive it?

b)   How does the quality of your product, as your customer sees it, agree 
with the quality that you intended to give him?

COMMENT
Pretty good questions ! 

----------------------------------------------------------------------------

Jean-Marie Gogue
President
The French Deming Association
Versailles   France
gogue@ensmp.fr

===========================================================================

Web address - mea culpa

clauson@deming.ces.clemson.edu (List Moderator Account)

Wed, 19 Jun 1996 14:17:29 -0400 (EDT)

Mea Culpa to those that tried the web addresses I sent yesterday.

When  I wrote the URL down, I changed rampages to rampage.
                                             ^
Adding the 's' back in fixes it:

http://rampages.onramp.net/~dumont/perapp1.htm
              ^
http://rampages.onramp.net/~dumont/perapp2.htm
              ^

Sorry for the inconvenience.

Your red-faced moderator,

Jim Clauson

========================================================================
 

Aggressive response to new ideas.

sta010@abdn.ac.uk

Wed, 19 Jun 1996 18:35:01 +0100

I would like ideas about a very important practical problem. I am sure
we must all have met people who respond to a Deming idea, not with
surprise, or disbelief, but with aggression. Surprise and disbelief
are easy to understand, and probably hopeful. If you jump uncritically
at a new idea, you are hardly likely to understand it. But some people
seem to see the new ideas as a threat.

I taught Statistics for many years, where some of the concepts are
difficult, and there have been many bitter arguments, for example about
the nature of probability. So the problem is not confined to the Deming
Philosophy, but it raises more than usual. Typical examples are
performance appraisal, targets, competition, pay-for-performance.

So, following the SPK, I need a theory, or theories, about the reasons
for this agression, in the form of predictions about ways to avoid it:
if, that is, it is a good thing to avoid it.

The greatest heat often comes from consultants or academics: they may
feel that their livelihood is threatened if they have to admit to having
been wrong. I would feel threatened if I did not keep finding I had been
wrong, because it would mean I had stopped learning. But the "threat"
theory may be wrong.

Ideas please:

Our overiding aim, it goes without saying, is to spread understanding
and application of the Deming Philosophy. So:

1)  Is an aggreesive response a sign that we are making progress, or
    an obstacle to progress?

2)  If it is an obstacle, how can we present the new ideas effectively
    without causing an aggressive response?


David Kerridge

British Deming Association Scotland

dfk@abdn.ac.uk

=========================================================================

Re: Aggressive response to new ideas.

dts@sympatico.ca

Wed, 19 Jun 1996 18:26:59 -0400

At 06:35 PM 6/19/96 +0100, David Kerridge wrote:

>1)  Is an aggreesive response a sign that we are making progress, or
>    an obstacle to progress?

An aggressive response is most definitely a sign of progress - it means the
respondent has heard the message, and even if (s)he doesn't immediately
agree with it, the seed has been sown. Any emphatic response, positive or
negative, is preferable to a shrug and a change of subject.

>2)  If it is an obstacle, how can we present the new ideas effectively
>    without causing an aggressive response?

Even if aggression is not an obstacle, effective communication is vital. The
concepts of pay-for-performance, quotas, striving to be #1, etc., are
'motherhood' issues to many people, and we must be sensitive to that. One
way to attack them would be to engage the doubter in a discussion about a
common topic centered around one of the 'motherhood' issues. Take athletic
events, for example. If a particular runner wins a particular race, does
that mean that he is the fastest in the world? Can anyone guarantee that the
same runner would win on the next day, too, while competing against the same
group of runners? Why not? (Come to think of it, the upcoming Olympics are
very fertile ground for debating Deming's ideas).

Another way is to sow the seeds of doubt by attacking one of the issues
'head on' and putting up with an aggressive response. A few months ago, I
tested the receptiveness of Deming's views on commission sales and other
pay-for-performance incentives on a friend who is a commission salesman
(with an MBA). Needless to say, he was aghast. As an example of 'effective
motivation', he told me about an acquaintance of his who used to work at a
car dealership where every month, the owner would fire the salesperson with
the lowest sales - just to keep the sales force motivated. When I mentioned
the success of some recently opened 'no dicker' dealerships where the sales
staff are on salary, he dismissed them as a 'totally different class of
operations'. 

Just yesterday, however, he was talking about leaving his current employer
because even though he will have met his annual quota in about two weeks,
and he will end up getting a large bonus at the end of the year, he would
like to work for a company that pays a higher base and a lower commission,
'just to have more stability during the year'. I bit my tongue. Obviously,
he recognizes the benefit of Deming's ideas, although not yet on a conscious
level. But the seed has sprouted!

These are just a couple of suggestions. I'm looking forward to seeing more -
just as I'm trying to learn Deming's teachings and relate them to my
experiences, I'm often in a position of having to explain them to others, so
this debate will be very beneficial.

Alla Linetsky

===========================================================================

Re:Demo(crat)ing

"Anton O. Tolman, Ph.D."

Wed, 19 Jun 1996 16:48:02 -0700 (MST)

On the 17th, Patrick Owings wrote:

> The Deming philosophy represents a sensible alternative to the ignorant, 
> half-baked, hypocritical nonsense of political ideology - Democrat, 
> Republican, Independent, or anything else you care to name.   

Patrick:

I was interested in your conclusions and thoughts on the question of 
politics and Deming.  Should we start a Deming political party?  :)

Actually, the "ignorant, half-baked..." political ideologies also 
constitute a system, and are probably a system that is not very self-
aware.  This is why so many people get elected promising to "change 
the system" in Washington, but never really seem to achieve very 
much.  The main statistics of interest within that system would seem 
to be opinion polls rather than control charts.  

Robert Crow's recent post on the June Atlanta group's meeting was 
intriguing because it suggests that at least there is a semblance of 
change occuring at the state level with regard to the quality 
movement and transformation.  It will be interesting to see how far 
it goes.  I would be interested to know how many state legislators or 
federal legislators are even aware of Deming's name or contributions? 
  
I guess one of my conclusions on this subject would be that unless 
and until we are able to achieve wide-spread consideration and 
application of Deming's teachings among the general population, the 
political system is unlikely to change.  Even with that base, it 
would constitute a massive culture shift to see results.  That does 
not mean it is not worth doing......

________________________________
Anton O. Tolman, Ph.D.
Director, Psychological Services
Director, Program Services
Wyoming State Hospital
P.O. Box 177
Evanston, WY  82931-0177
Anton@wsh.state.wy.us
(307) 789-3464
---------------    

=======================================================================

Re: Aggressive response to new ideas.

"Andre' Everett"

Thu, 20 Jun 1996 08:30:27 -1200

In response to an interesting post by David Kerridge:
   David answered his own question with the following comment:
> I would feel threatened if I did not keep finding I had been
> wrong, because it would mean I had stopped learning. But the "threat"
> theory may be wrong.

   Were the Deming Philosophy to be blindly accepted by all, it would not be
challenged (aggressively or otherwise).  Without challenge, even continuous
improvement would slow (and radical or major, non-incremental change would be
lost).
   Thus:

> 1)  Is an aggreesive response a sign that we are making progress, or
>     an obstacle to progress?

   Yes, this is both:  Progress results from overcoming obstacles, which must
first be encountered and recognized.  Alcoholics Anonymous works on this
principle:  You're half way to solving your problem by admitting that you
have a problem.  In our case, identifying the challenges (aggressive
responses or respondents) allows us to address the issues and the issuer,
which will benefit not only our understanding of the Deming philosophy but
hopefully also the challenger's.

> 2)  If it is an obstacle, how can we present the new ideas effectively
>     without causing an aggressive response?

   Perhaps our goal should be to stretch, in some instances, to the point
where we do provoke aggressive responses.  My suggestion would be that we
choose when and where to fight, instead of allowing this to be at the
discretion of the sparring partner...

- Andre' M. Everett (PhD), Advanced Business Programme, University of Otago -
------  Box 56, Dunedin, New Zealand; tel 64 3 479 8047/8046; fax 8045 ------
---------------------- aeverett@commerce.otago.ac.nz ------------------------

==========================================================================

Four Days with Dr. Deming

Don Clarke

Wed, 19 Jun 1996 09:43:40 -0700

I just finished Four Days with Dr. Deming, by William J. Latzko and David M. 
Saunders.  A very nice read.  

I've read the two last books by Dr. Deming and several others about him, but 
Four Days is separated not only by its reader-friendly layout but also by the 
various points of view offered of the seminar: wonderings of the fictional 
executive attendant, interpretations and emphases by the authors, and quotes 
of Dr. Deming.  The epilogue offers a flowchart giving guidance on 
implementatng Deming's ideas, information on where to get help, and questions 
for discussion.

I believe Mr. Latzko is a DEN list member.  Congratulations on a wonderful 
job.  It was a great learning experience for me.  

I suppose this list has discussed his book before this, but if anyone has not 
read it, it is solid.   
-- 
Don Clarke
U.S. Geological Survey
clarke@usgs.gov

==========================================================================

Aggressive Response To New Ideas

David Young <100654.3270@CompuServe.COM>

20 Jun 96 17:19:18 EDT

David Kerridge asked:
>>1) Is an aggressive response a sign that we are making progress, or an
obstacle to progress?
    2) If it is an obstacle, how can we present the new ideas effectively
without causing an aggressive               response?<<

In my own experience I have found resistance to change to be useful. In a
'Darwinian' sense, the more aggressive and powerful the resistance, the more
elegant, persuasive and powerful I have been invited to make my response, if it
is to 'survive'.

The extremely strong challenges to the Deming philosophy that I have met at work
have prompted me to go away and really sort my thoughts/arguments out.

Alberto Knox in 'Sophies World' (page 283) says:
 Theres no need to have an opinion on something everyone agrees on.  
 The very best that can happen is to have energetic opponents. The more extreme
they become, the more powerful the reaction they will have to face.'' 
''The many men in Hegel's time who could reel off gross broadsides like the one
on the inferiority of women hastened the development of feminism. They produced
a thesis. Why? Because women had already begun to rebel. And the more grossly
they expressed themselves about women's inferiority, the stronger became the
negation.''

I see an aggressive response as a sign of progress, as an invitation to present
Deming's philosophy more elegantly, patiently, professionally and convincingly.
This benefits everybody.

I do not see it as an obstacle. Rather than looking for ways of presenting the
new ideas effectively without causing an aggressive response, I would welcome an
aggressive response as a way of discovering how to present the ideas more
effectively.

David Young
100654.3270@compuserve.com
=================================================================================

Re: Aggressive response to new ideas.

tpowers@uscsumter.uscsu.sc.edu (Tom Powers)

Thu, 20 Jun 96 13:56:58 EST

"""""""""""""""""""""""David Kerridge wrote"""""""""""""""""""""""""""
I would like ideas about a very important practical problem. I am sure
we must all have met people who respond to a Deming idea, not with
surprise, or disbelief, but with aggression... But some people
seem to see the new ideas as a threat.
""""""""""""""""""""""""""""end quote"""""""""""""""""""""""""""""""""


     I hope no one thinks such a reaction is unique to Deming ideas. Many 
     people react to ANY new idea as a threat. In part, this is because 
     they are accurately perceiving the situation. And it should surprise 
     no one that many people react to threats by demonstrating "aggressive" 
     behavior which is not really aggressive at all, but profoundly 
     defensive.
     

""""""""""""""""""""""""another Kerridge comment""""""""""""""""""""""""
The greatest heat often comes from consultants or academics: they may
feel that their livelihood is threatened if they have to admit to having
been wrong. I would feel threatened if I did not keep finding I had been
wrong, because it would mean I had stopped learning. But the "threat"
theory may be wrong.
"""""""""""""""""""""""""""""""end quote""""""""""""""""""""""""""""""""

     First, please entertain the possibility that there may be other 
     paradigms than your own: that is, that people may react strongly, not 
     simply because they perceive their livelihoods threatened, but because 
     a cherished value -- not necessarily an economic one, but perhaps one 
     in which their personal identity is wrapped up -- is threatened. 
     "Threatened," that is; not "challenged" or "examined" or "engaged." 
     Academics, in particular, often see the world in wildly diverse ways. 
     A behaviorist or quantification paradigm may itself be inconsistent 
     with broader reality as a humanist philosopher sees it, and any 
     attempt to impose it upon a community of free thinkers may be seen -- 
     not in terms of the idea and its merits but in terms of the threat of 
     its being imposed -- as destructive to the very concept of academic 
     integrity. This should not be surprising to Deming followers, who more 
     than most others, should understand the importance of paradigmatic 
     thinking, for better or for worse. 
     
     Second, please recognize that ideas do not get communicated in a 
     vaccuum. They come wrapped in context, and presented in an 
     environment. For example, an idea I'm encountering a lot now is that 
     anything which smacks of TQM is simply a tool of administrators to 
     amass all power in their own hands while giving the appearance of 
     participatory management; thus, TQM does nothing but legitimze an 
     illegitimate thing. It's a trick to make it look like "we" are 
     cooperating in our own destruction, while shielding those who REALLY 
     make the decisions from the consequences of their decisions, and 
     directing those consequences toward those whose participation was 
     visible but minimally significant. Absent a context of trust, almost 
     ANY new idea will probably be taken to be "just another trick," or, to 
     quote a common phrase "BOHICA: Bend Over, Here It Comes Again!" This 
     is, of course, mostly a problem of a pre-existing system, but when are 
     ideas NOT presented in context of an existing system? Seldom can we 
     really build on truly virgin ground.
     
     As for the environment, that's a related problem. An idea which might 
     be easily accepted as at least a subject for debate in one time might 
     be summarily rejected in another. Some issues are too hot to handle in 
     some environments, and not a problem at all in others. The present 
     environment in which academics work is one of increasing political 
     intrusion into academics. The justification for these intrusions is 
     usually phrased with terms like "accountability," "performance 
     standards," "objective criteria," and even "TQM." To debate seriously, 
     and thus to lend scholarly credence to, such issues simply plays into 
     the hands of those who are attempting to effect the intrusion. The 
     mere fact that the issue is considered open for discussion at such a 
     level is taken as evidence that it's valid, and therefore that the 
     intrusions are themselves justified. Even where there is relative 
     trust within an organization, and where what I described above as 
     "context" does not enter in, the question of "whether this is the 
     thing to do in this environment" must enter.
     
     A couple of examples: a few years ago, some of the faculty at my 
     institution objected to the fact that prayers -- always Christian 
     prayers, though probably by default rather than by intent -- were part 
     of university ceremonies which faculty were required to attend. Our 
     faculty contains Jews, Muslims, a Hindu, a Bahai, and a large number 
     of people without any particular religious orientation, plus a large 
     number of devout Christians who object on religious and philosophical 
     grounds to breaching the separation of church and state. Was it not a 
     violation of individuals' religious freedom to compel attendance at 
     functions of which religious services not consistent with their own 
     values were a part, particularly when the ceremony was not that of a 
     church, but of the university which supposedly contains us all?
     
     Only a few people -- all anti-separationist Christians -- objected to 
     the idea of doing away with prayers in and of itself (though there 
     were many who wanted to keep the prayers, but chose to vote for doing 
     away with them out of respect for their fellows who believed 
     otherwise.) This IS, after all, a public institution. But this is also 
     South Carolina, and religion is a hot political issue just now. The 
     Christian Coalition, as a matter of fact, has won control of the 
     state's Republican Party, and controls many, if not most, of the 
     county organizations too. What would be the effect of public support 
     for higher education in the state, and of legislative support for 
     appropriations for higher education, if we were to take such a step, 
     and let ourselves be branded "anti-Christian" in the media and in the 
     public mind? Indeed, what would happen if the media in this region 
     discovered that we were even discussing the issue? Further, the state 
     government was conducting some highly-charged investigations into 
     higher education and the means of funding it. Was this the time even 
     to breathe such an idea? Opposition was strong and yes, even 
     "aggressive." But the opposition was based, not on resistance to the 
     idea itself, but to fear of the consequences of consideration of the 
     idea in this particular environment.
     
     Another example: perhaps it would be a good idea to keep track, even 
     statistically, of what people actually do with their days. But right 
     now, our legislature has mandated that institutional funding no longer 
     be based upon enrollments, but upon meeting certain "performance 
     standards." One of these is "faculty workload." No one knows just how 
     that will be defined, but the strong fear is that those who made the 
     law, and have no idea what goes into being a faculty member, will be 
     expecting some time-clock-like system which would probably be 
     destructive of the entire way higher education is done. While 
     reviewing the way higher education is done certainly might be in 
     order, it should come from serious study on its own, and not be 
     imposed by default and through the back door. Thus, why now make 
     available information which will probably be fed into what is already 
     a threatening development? Why not wait until the cloud passes, or 
     until it's already upon us anyway, and do it when we can benefit from 
     it more than we'll be threatened by it? The environment makes what 
     could be a neutral or even a welcome idea into a threat.
     
     I think all these things need to be considered.
     
     More basically, people can respond vigorously to new ideas because 
     they really ARE a threat. No, David, it's not necessarily that people 
     are afraid to find out that they've been wrong. It's that they're 
     afraid to permit any method which would enable to boss to prove that 
     they've been wrong, and thus to justify their firing. Where the system 
     provides security, not for excelling, but for not screwing up, the 
     most important imperative is to avoid the impression that you've 
     screwed up. Here, you can't learn from your mistakes, because if you 
     acknowledge the mistake, you're fired! Further, even though the boss 
     may have the impression that you've screwed up, many present systems 
     still prohibit his doing anything about it unless he can PROVE it. A 
     new idea which offers means of proving it simply puts a gun into the 
     hands of the man who's already trying to kill you. "Doing it better" 
     is a luxury for the secure. "Doing it differently" destroys security. 
     What do you expect? "Doing it better" may be essential for 
     institutional survival, but destructive of individual security. The 
     way the system is set up will have a lot to do with whether that 
     contradiction exists, and with which side of it will prevail.
     
     Recent postings comparing Deming to Maslow have a point, I think. 
     Unless fear is first cast out, new ideas will be received in an 
     atmosphere of fear and distrust. So, David, perhaps your problem is 
     not that you're presenting a Deming idea, but that you're presenting a 
     Deming idea out of sequence. Perhaps another Deming idea needs to come 
     first. 
     
     I know it's not for a tyro like me to say something like this, 
     particularly to someone like you, but I'm going to say it anyway: it 
     IS important to consider the system, and not simply the reactions of 
     individuals within the system.
     

""""""""""""""""""""""""""""David then asked"""""""""""""""""""""""""""""
1)  Is an aggreesive response a sign that we are making progress, or
    an obstacle to progress?
"""""""""""""""""""""""""""""""""""end quote""""""""""""""""""""""""""""


     Could be either, neither, or both. See above.
     

""""""""""""""""""""""""""""""and finally"""""""""""""""""""""""""""""""""
2)  If it is an obstacle, how can we present the new ideas effectively
    without causing an aggressive response?
"""""""""""""""""""""""""""""end quote""""""""""""""""""""""""""""""""""""

     Consider presenting them in sequence, with an eye toward context, 
     environment, and SYSTEM. Unless fear first be case out, and trust 
     instilled, all new ideas will appear at least mildly unsettling, and 
     probably severely threatening. And most people don't respond well to 
     threats.


Tom Powers
Professor of History
The University of South Carolina at Sumter
TPOWERS@USCSUMTER.USCSU.SC.EDU

========================================================================

Agressive Behavior

jr1crow@mindspring.com (James Robert Crow)

Thu, 20 Jun 1996 21:03:36 -0500

        I would like to thank Tom Powers for an excellent discussion of a
problem which we all face from time to time.  An excellent book on this
subject is "Excape From Freedom," by Erich Fromm.

In this book Fromm examines Germany's move from democracy to a totalitarian
state in the twenties and thirties.  The German's in power at the time
thought they could control Hitler.  They were wrong.  Another excellent book
by Fromm is "Man For Himself."  An inquiry into the psychology of ethics.
Here is a short discussion from that book on the use of power.

"Productiveness is man's realization of the potentialities characteristic of
him, the use of his powers.  But what is "power"?  It is rather ironical
that this word denotes two contradictory concepts: Power of = capacity and
power over = domination.  This contradiction, however, is of a particular
kind.  Power = domination results from the paralysis of power = capacity.
"Power over" is the perversion of "Power to."  The ability of man to make
productive use of his powers is his potency; The inability is his impotence.
With his power of reason he can penetrate the surface of phenomena and
understand their essence.  With his power of love he can break through the
wall which separates one person from another.  With his power of imagination
he can visualize things not yet existing; he can plan and thus begin to
create.  Where potency is lacking, man's relatedness to the world is
perverted into a desire to dominate, to exert power over others as though
they were things.  Domination is coupled with death, potency with life.
Domination Springs from impotence and inturn reinforces it, for if an
individual can force somebody else to serve him, his own need to be
productive is increasingly paralyzed."

I have been using this for about 25 years as a management development tool.
It ties in well with McGreggor's theory X, theory Y.  How do you look at
your position in management, and the power associated with that position?
How would you concept of you position and of people affect your behavion and
the leadership you provide?  This also ties in well with Deming's point
number 8.  "Drive out Fear."

Ideology is the enemy of mankind.  I would include in that all the "ism's",
religion, political parties etc.  Once we become attached to a particular
thought pattern it clouds our vision.  The world becomes black and white
instead of shades of gray.  It affects our ability to see other people as
human beings, and thus robs us of our humanity.  We lose our ability to
think objectively, and it becomes a struggle of us vs. them.

On the subject of liberal vs. conservative.  I have been keeping up with
that discussion, and I do not believe I have seen an operational defination
of either term.  Since we have not arrived at any kind of consensus on what
these two words mean can we perceed with that discussion?  Do we want to do
that anyway because we will be getting into the area of personal values, and
Harry Truman said, "Two intelligent people never agreed about anything."  
Robert Crow
President, The Crow Group
jr1crow@mindspring.com

====================================================================

Aggresive Response to New Ideas

jr1crow@mindspring.com (James Robert Crow)

Thu, 20 Jun 1996 21:03:33 -0500

>To: den.list@deming.ces.clemson.com
>From: jr1crow@mindspring.com (James Robert Crow)
>Subject: Aggresive Response to New Ideas
>Cc: 
>Bcc: 
>X-Attachments: 
>
>        Generall when we hear something for the first time, or we hear
something that is totally different from what our understanding of a subject
has been there is a tendancy to reject it out of hand.  The first time I
heard that Deming was opposed to performance appraisals and pay for
performance I wondered just how old this guy was and where he had been.
Anybody knows that these two practices are the cornerstone of any modern
human resource program.
>
>     However having implemented two of these programs, one for a Fortune
200 company and another for a high tech Japanese company I was aware of the
problems we experiences during the implementation and afterwards, so
possibly I was a little more receptive than the average bear.
>
>     Quite frankly I am taking an agressive posture on these subjects.  I
am looking for people who can give me good arguments for the continuation of
appraisals, pay for performance, and quotas.  The reason is that I am
concerned that I don't know as much as I think I know.  I am concerned that
I will get blind sided, by something I did not think about.  By continually
testing my knowledge base I can learn, and just possibly make a few converts
in the process.
>
>     In March the local chapter of AQP sponsored a debate between myself
and another person on the subject of performance appraisals and pay for
performance.  This was a fairly well attended meeting and the group got to
hear both sides of the issue.  I think for most of them it was the first
time they had heard anyone attack these two sacred cows.
>
>     As a result of this debate the editor of HR Atlanta, circulation 7000,
has asked both of us to write a position paper, which will be published in
an upcoming issue.  I don't back down from discussing this subject with
anyone.  I am looking for people to shoot holes in my positions.  This gives
me a way to continually test what I think I know.
>
>     I recently made a presentation to a group in Atlanta on the subject of
why TQM fails and what to do about it.  I asked the group to give me their
feelings on the subject and then we worked with what they came up with.  I
also cited a study I did several years ago of companies in the Atlanta area
and their implementation efforts.  One of the conclusions was that "Not
taking a system approach was a major cause of failure.
>
>     As you know one element of a system is interdependence.  During this
discussion we talked about performance appraisals, pay for performance, and
reward systems, and how they ignored the existance of a system and in fact
resulted in sub-optimization.  One of the participants, a consultant,
recently retired from IBM, wasn't buying any of it, even though the group as
a whole could see the point and agree with it.
>
>     We then did the Green M&M exercise which I have developed as a
substitute for the Red Beads exercise.  A part of the exercise is that I
reward, and punish performance.  His group performed outstanding the first
round and each person earned a production bonus. (a penny).
>The second round they did not do well and I questioned their dedication,
ability to follow procedures, etc.  The third round they did well again and
when I approached them with their production bonus, this person said, "Get
out of here with your bonus."  Even he could see that the bonus was
meaningless.  The system was controlling perfromance.  This made a much
stronger point than I could have ever made in a verbal argument.
>
>     Adults learn by doing.  If you can incorporate some exercises in you
presentation which allows the group to expericenc the power of the system on
performance, the difference between competition and cooperation, and the
difference between motivating forces, and dissatisfiers in a work
environment the group will begin to pick up on the idea and will work to
convert doubting Thomases.  It is more powerful if you can discipline
yourself to keep you mouth shut and allow the group to do the work.  This is
hard to do.  I think I was much better at this before I thought I knew so much.
>
> 
>
Robert Crow
President, The Crow Group
jr1crow@mindspring.com

=======================================================================

Renewing American Civilization Review Project

"Kerr, Donald"

Fri, 21 Jun 1996 07:06:00 -0700

The following is a list of Team members for the Renewing American
Civilization Review (RACR) Project.  The DEN RACR Team will review
Speaker Newt Gingrich's lectures on his foundational principles
necessary to the renewal of American Civilization.  The transcripts from
a series of 10 lectures can be downloaded from
http://www.pff.org/pff/i-renew.html.  The team will then develop a
DEN-based feedback response.  Our purpose is to open up a dialogue with
Speaker Gingrich with the hope of improving his and our understanding of
the Deming Philosophy and The System of Profound Knowledge.  We believe
our collective contributions will make a difference in truly Renewing
American Civilization.

The purpose of this DEN post is to complete the Team Formation phase by
June 28. which will be followed by a brief off-line team planning and
scheduling phase.  The first team lecture review will begin the first
week of July.

If you would like to be a vital part of this team (no citizenship or
credentials required) please let me know directly at the following
address: donald.kerr@alliedsignal.com

The DEN RACR Team

John Hunter -- wdsg@aol.com
Wayne Levin -- levin@astral.magic.ca
Anton Tolman -- anton@wsh.state.wy.us
Ian Bradbury -- lnuspth1.rzyj8c@gmeds.com
Jonathan Siegel -- jmseigel.info.research@worldnet.att.net
Matt Barkley -- mbbjr@vnet.ibm.com
Patrick Owings -- owingsp@msn.com
Rip Stauffer -- ripstaur@aol.com
John Seddon -- 101454.573@compuserve.com
Stan Adams -- sadamstimd@aol.com
Myron Tribus - 104055.2663@compuserve.com
John Paul Fullerton - jpf@myriad.net
Bill Cooper - BCoop1116@aol.com
Don Kerr - donald.kerr@alliedsignal.com

I would like to thank each of you for volunteering.  This is going to be
a Great Adventure!

Don Kerr

P.S.  I guess a "Speed Racer" metaphor would be silly...so what! Go DEN
RACR!...Go DEN RACR!...Go DEN RACR...Gooooo!

P.S.S.  To the team...go ahead and start reviewing and making notes on
the "Quality as defined by Deming" lecture.  Matt suggested we start
there and work out to the others...sounds good to me.  Also, double
check you addresses. thanks.

====================================================================

Re: Agressive Response To New Ideas...

"Patrick Owings"

Fri, 21 Jun 96 13:02:59 UT

Tom Powers' recent comments are very enlightening.  As I reflect on my own 
situation, I can see (I believe) what, in part, may be necessary for the 
Deming Philosophy to become accepted in an organization.

I met Dr. Deming in 1991, in Greenville, South Carolina, at one of his Four 
Day Seminars.  At the time, I really knew very little about him, but as the 
seminar progressed and I had the opportunity to meet and talk with Gipsie 
Ranney, and others, I became awe-struck.  I remember remarking to my 
operations manager who was attending with me that being in the same room with 
Dr. Deming was akin to being in the same room with God, i.e., an overwhelming, 
humiliating experience.  I remember a chance meeting with Dr. Deming in the 
corridor one evening.  I struggled to say something intelligent to the man, 
but all I could do was stand there with my mouth open and no words coming out. 
 I'm sure he passed by without even knowing I was there.  When I returned home 
Dr. Deming's books collected dust while I struggled to run my company.  It 
wasn't until I returned to school in an Executive MBA program that I began 
pursuing the Deming Philosophy with a purpose.  On more than one occasion, I 
became very angry at some of my professors because they in essence dismissed 
Dr. Deming's theories in favor of your traditional MBA approach to things, 
i.e., highly analytical (think about as opposed to synthesis and Russell 
Ackoff's teaching).  Fortunately, one of my professors had received his Ph.D. 
under Ackoff at Wharton and he encouraged me to continue my pursuit.  I also 
have an engineering degree, so I have nothing against being analytical, nor do 
I have anything against MBAs.  What I do have a problem with is the way we are 
all taught to think in most business schools.  After finishing my MBA program, 
I had the time to pursue major change within my company.

Since I am at the top of my own company, I did not have to be as concerned 
about strong opposition to my desire to operate the organization using the 
Deming Philosophy.  I simply made it very clear that we were going to do it 
together.  The first year I spent 250 hours with top and middle management and 
first line supervision, together in a classroom setting, teaching them the 
Deming Philosophy.  I made it clear that no one would be fired or chastised 
for questioning or criticizing anything we were learning.  I made certain that 
I personally protected first line supervision from middle management, and 
middle management from top management.  All were encouraged to come directly 
to me with their concerns.  We have developed an open, unafraid dialog where 
ideas and concerns are addressed honestly and without hidden agendas (they 
soon surface).  The results are astounding and have even been observed by our 
customers as representing dramatic improvement in our company's service and 
performance.

There is much more to this story and I will have to continue at a later date.  
The point is (yes, there is a point) that the leadership at the very top must 
be converted because THAT is where the problem is.

Patrick Owings     

==========================================================================

VARIATION: p- np- u- and c- charts vs. x-Bar and Ind. x/moving R charts

Kromkowski@aol.com

Fri, 21 Jun 1996 12:00:28 -0400

I was wondering if any one could comment on the inability of the p-, np-, u-,
and c- charts to provided information on the "flattness" of a system's
variation.  The spread of variation in these charts is a always a function of
sample size, if I understand correctly. It seems to me that there may be
quite a bit of difference between a system that produces defects in a narrow
band and a system that produces defects in a larger (although with the limits
predicted by p-, et al charts) band.

Unlike x-Bar and X/moving R, these charts (p-, et al) do not seem to imbed
within the calculation of limits information about how one data point varies
from the next.  I'm not sure about this, but it seems that the underlying
assumption of  the p-, et al charts is that individual data points are
_independent_ of the prior and the next (like the hypothetical fair coin
toss).  How does this rub against the idea of SYSTEM.

In the context of the red beads and using the data from TNE example,  won't
an Ind. X/ moving R chart provide essentially the same results?

Comments?

John D. Kromkowski
Attorney at Law
The Jefferson Bldg. -- Suite 103
105 W. Chesapeake Ave.
Towson, MD  21204

Tel. 410 821-6116
Fax. 410 828-0705

also Kromkowski@aol.com

====================================================================

OTHER: List - Subject Lines

Kromkowski@aol.com

Fri, 21 Jun 1996 12:00:29 -0400

Notwithstanding some of the fuzziness about the aim of DEN, and fully
cognizant of the issue of destruction of a system, I was wondering if a more
systematic approach to "Subject Lines" of posts might be helpful.
Specifically, I would suggest that Subject Lines follow the following format:

SYSTEM - unique desc.
VARIATION -unique desc.
PSYCHOLOGY - unique desc.
THEORY OF KNOWLEDGE - unique desc.
ANNOUNCEMENTS -unique desc.
OTHER - unique desc.

Obviously, OTHER in time and after consideration my Moderator might be
expanded.
Having these categories might help us identify what gets discussed and what
doesn't.

John D. Kromkowski
Attorney at Law
The Jefferson Bldg. -- Suite 103
105 W. Chesapeake Ave.
Towson, MD 21204

Tel.  410 821-6116
Fax. 410 828-0705

also Kromkowski@aol.com

=====================================================================

Re: Psychology -vs- Physiology

jkaliste@skipjack.bluecrab.org

Sat, 22 Jun 1996 12:29:18 -0400 (EDT)

> From den.list-request@deming.ces.clemson.edu Wed Jun 19 18:54:56 1996
> Return-Path: 
> Received: from eng.clemson.edu by deming.eng.clemson.edu.eng (4.1/SMI-4.1)
> 	id AA12723; Wed, 19 Jun 96 18:54:55 EDT
> Received: from deming.eng.clemson.edu.eng by eng.clemson.edu
(SMI-8.6/SMI-4.1)
> 	id SAA23123; Wed, 19 Jun 1996 18:54:24 -0400
> Received: by deming.eng.clemson.edu.eng (4.1/SMI-4.1)
> 	id AA12718; Wed, 19 Jun 96 18:54:54 EDT
> X-Envelope-From: jkaliste@skipjack.bluecrab.org  Wed Jun 19 18:54:51 1996
> Received: from eng.clemson.edu by deming.eng.clemson.edu.eng (4.1/SMI-4.1)
> 	id AA12712; Wed, 19 Jun 96 18:54:51 EDT
> From: jkaliste@skipjack.bluecrab.org
> Received: from deming.eng.clemson.edu.eng by eng.clemson.edu
SMI-8.6/SMI-4.1)
> 	id SAA23120; Wed, 19 Jun 1996 18:54:21 -0400
> Received: from skipjack.bluecrab.org by deming.eng.clemson.edu.eng
4.1/SMI-4.1)
> 	id AA12709; Wed, 19 Jun 96 18:54:49 EDT
> Received: from cfn_client33.bluecrab.org (cfn_client33.bluecrab.org
[206.205.140.83]) by skipjack.bluecrab.org (8.7.4/8.7.3) with SMTP id
AA10918 for ; Wed, 19 Jun 1996 18:54:14
-0400 (EDT)
> Date: Wed, 19 Jun 1996 18:54:14 -0400 (EDT)
> Message-Id: <199606192254.SAA10918@skipjack.bluecrab.org>
> X-Sender: jkaliste@skipjack.bluecrab.org
> X-Mailer: Windows Eudora Version 1.4.4
> Mime-Version: 1.0
> Content-Type: text/plain; charset="us-ascii"
> To: den.list@deming.ces.clemson.edu
> Subject: Re: Psychology -vs- Physiology 
> 
>On Saturday, June 15, John Kaliste (?) wrote:
>
>         It is great to have Dr. Tolman aboard.  I am no phd, but I feel
> that the psychological does not exist, only the psychophysiological.
People
> observe what they are trained to understand, and focus on how they
have been
> trained to help.  
>
>John, I would like to ask for more definition of what you mean by the 
>"psychological does not exist, only the psychophysiological"?  For 
>the hard-core psychoanalysts, psychology, meaning the internal 
>perception of events and internally generated reaction to events, is 
>all that matters.  

Nothing exists outside of the human mind.  My sensibilities do not 
convey externally. My eyes see only a fraction of what my brain translates 
into my perception of what I think I see.  Pyschophsyiology. I am no analyst.
I believe it was Gardner Murhpy who described how the mind throws + noise +
into the + channel+ , tightening of muscles, drumming fingers on the table. 

Watch Bill Clinton on t.v.,it is easy to see when he is telling a 
An analyst uses a couch to get people to relax, so they can attempt to get
past this defense.  
 
>Our ability to understand new concepts and new ideas is constrained 
>by the language we use to describe those concepts and the rigidity of 
>our own cognitive system.  Deming's point that the system cannot know 
>itself reminds me of this point.  My belief is that there is no 
>essential distinction between "psychology" and "physiology". 
 Thank you 
>The current dominance of biochemical explanations for behavior is
nothing more 
>than a reductionistic explanation ....
         
Some one I love very deeply is bi polar.  This is caused by a
chemical inbalance in the brain.  I have seen this person have all of the
properties of being asleep.  They will physically spasm, and become awake.
After many endless days days, they will become manic, attempt geographical
cures, and become suicidal, from lack of sleep. Medication relieves this.
The physical lack of sleep causes a mental>> stress. 

>Would you say anabolic steriod intake, combined with intense
>exercise,
> would not affect behavior? 
>
>I will probably add more on this later.  These were just some 
>prattlings at the end of the day.  I hope they are helpful.

You are not prattling.  We need you, your knowledge,and skills.

Thanks,
John
jkaliste@skipjack.bluecrab.org

==============================================================

Re: Aggressive response to new ideas.

gathomas@mail2.quiknet.com (George A. Thomas)

Sat, 22 Jun 1996 12:05:12 -0700

David Kerridge asked:

>1)  Is an aggreesive response a sign that we are making progress, or
>    an obstacle to progress?

An "aggressive" response from someone who could influence progress suggests:
the individual is not open to alternative thinking at the moment, will
resist collaborative efforts, and would probably be an obstacle to progress. 
>
>2)  If it is an obstacle, how can we present the new ideas effectively
>    without causing an aggressive response?

There are so many different reasons and combinations that underpin an
individual's  resistance to change that one would have to know much more
about the history and context of the individual to suggest a presentation
strategy that might not cause an aggressive response.  Unless the person is
a "show-stopper" and warrants the effort, I will generally focus my time and
energy on those who can, at the very least, enter into dialogue.       


*****************************************************************************
"I never said it would be easy, I just said it would work."  W. Edwards Deming
*****************************************************************************
George A. Thomas
Mananagement Consulting & Training
(916) 684-4381
gathomas@mail2.quiknet.com
*****************************************************************************

===========================================================================

SPC: p-charts etc

sta010@abdn.ac.uk

Sat, 22 Jun 96 21:55:35 BST

John D Kromkowski asks an interesting question:
>I was wondering if any one could comment on the inability of the p-, np-, u-,
>and c- charts to provide information on the "flatness" of a system's
>variation.  The spread of variation in these charts is always a function of
>sample size, if I understand correctly......


Yes, that's right. The control limits in these charts are calculated
from the theoretical model, which does sound risky, since no one should
ever believe that a theoretical model is "true". But we don't need to.
The control limits are meant to show the best that the system could do,
in an ideal state. We know we are never in the ideal state, but we want
to know if we are getting close to it. So the theoretical model which
describes that unattainable ideal is a good standard to check the real
variation against.

>................................It seems to me that there may be
>quite a bit of difference between a system that produces defects in a narrow
>band and a system that produces defects in a larger (although within the limits
>predicted by p-, et al charts) band.

Since the theoretical model shows the best you can possibly get, if we
get narrower limits for variation than that, something is very fishy.
Usually it means that someone is cheating: for example not recording
defects if there are too many of them. As Dr Deming comments in one of
the MIT Deming Library tapes (which everyone should see), this happens
much more often than people think.

Equally we find, quite often that the spread is much wider than the
theoretical model tells us to expect. This occurs is defects are not
independent, and occur in clusters..
>
>Unlike x-Bar and X/moving R, these charts (p-, et al) do not seem to imbed
>within the calculation of limits information about how one data point varies
>from the next.  I'm not sure about this, but it seems that the underlying
>assumption of  the p-, et al charts is that individual data points are
>_independent_ of the prior and the next (like the hypothetical fair coin
>toss).  How does this rub against the idea of SYSTEM.

As I said above, we don't really believe this: but as we improve the system
we get nearer and nearer to it.

I usually draw both charts for investigation purposes, that is the chart
which treats the points as individual values, and the appropriate other
chart. Usually they tell the same story. If they don't, you may find
extra information about the process. For routine purposes, as opposed
to initial investigation, one is enough.
>
>In the context of the red beads and using the data from TNE example,  won't
>an Ind. X/ moving R chart provide essentially the same results?
Yes, because in those cases the models fit well. But drawing both
helps you to find if this is true.


David Kerridge

dfk@abdn.ac.uk

======================================================================

Aggressive response to new ideas

sta010@abdn.ac.uk

Sun, 23 Jun 96 18:50:57 BST

Thanks to all who have contributed, both on the net and privately.

As several have said, there are several different situations, which
should be regarded separately.

1   We may be giving a presentation to a group who are completely new
    to the ideas. Then the order in which we present them is important.

2   If we are trying to introduce the ideas within a company, the
    important of "Drive out fear" is very great. No one can be expected
    to react well to ideas that create fear if there is fear already
    in the organisational structure.

3   We may be explaining Deming ideas to those who already have some
    knowledge, either of the Deming Philosophy or of some other system.
    This is the situation which gives me most difficulty, because it
    involves unlearning - ten times more difficult than learning from
    a "clean slate".

Dr Deming seemed to imply that you only have one chance with education.
I hope he did not mean that those who have been wrongly taught should be
written off: it does not sound like him. This was the main reason I raised
the topic.

Clearly it is far better to avoid the problem, by making sure that all
introductory teaching is of the highest quality. But we face the situation
in which harm has been done.

Does anyone have advice on helping people to unlearn? I do not know of
any research on this, apart from that mentioned by Howard Gardner in
"The Unschooled Mind". But there must be a good deal of research and
experience out there somewhere.

I also have some comments and questions on fear and reactions to fear,
but will save these for another posting.

David Kerridge

dfk@abdn.ac.uk

======================================================================

Re: VARIATION: p- np- u- and c- charts vs. x-Bar and Ind. x/moving R charts

jkaliste@skipjack.bluecrab.org

Sun, 23 Jun 1996 10:46:03 -0400 (EDT)

>I was wondering if any one could comment on the inability of the p-, np-, u-,
>and c- charts to provided information on the "flattness" of a system's
>variation.  The spread of variation in these charts is a always a function of
>sample size, if I understand correctly. It seems to me that there may be
>quite a bit of difference between a system that produces defects in a narrow
>band and a system that produces defects in a larger (although with the limits
>predicted by p-, et al charts) band.
>
>Unlike x-Bar and X/moving R, these charts (p-, et al) do not seem to imbed
>within the calculation of limits information about how one data point varies
>from the next.  I'm not sure about this, but it seems that the underlying
>assumption of  the p-, et al charts is that individual data points are
>_independent_ of the prior and the next (like the hypothetical fair coin
>toss).  How does this rub against the idea of SYSTEM.

        The pattern displayed by the relationship of the individual points
is what it is all about.  If you have an upper and lower limit, you will have
a center.  The distance from center is what we are measuring.  If you see
a saw tooth pattern, slowly rising or falling and then abruptly and perfectly
corrected many times, you may have tampering. If you have that much control,
you should be able to have a centered line, or almost a lack of pattern. If 
you are lacking variation, it is time to tighten your limits.
        Look for a pattern, five points in a row on one side of center, or
too little variation, to see if something is wrong, or right.
        This is an indication only.  System knowledge is understanding the
cause of the effect.

                                                      Thanks,
                                                       John 
        
jkaliste@skipjack.bluecrab.org

======================================================================

Re: OTHER: List - Subject Lines

Julie Beedon

Sun, 23 Jun 1996 09:52:23

 
, I would suggest that Subject Lines follow the following format:
>
>SYSTEM - unique desc.
>VARIATION -unique desc.
>PSYCHOLOGY - unique desc.
>THEORY OF KNOWLEDGE - unique desc.
>ANNOUNCEMENTS -unique desc.
>OTHER - unique desc.
>
Main problems then is that we would be breaking into parts that
which we should be learning to synthesise into a whole...  

Julie Beedon
VISTA Consulting - for a better future
julie@vistabee.win-uk.net

======================================================================

Re: Aggressive response to new ideas.

Ian Griffiths <100326.1264@CompuServe.COM>

23 Jun 96 17:45:10 EDT

In response to David Kerridge's question:

I doubt whether we are going to find a useful theory. We probably have to
CONSTRUCT it.

Maybe it's helpful to view the problem from a human motivation viewpoint. The
victim (person with alcohol problem, drug-dependency, or commitment to
management style that 'frustrates' basic human needs) is motivated to change. He
sees the benefits of change. But he sees them unclearly. The benefits are
distant and there is a risk they will prove illusory (at its simplest, the
'patient' may die before the cure is complete!). The costs are more obvious and
painful -- short-term disruption to the existing steady-state pattern of
behaviour. The status quo may not be 'good' but it has become 'comfortable'.

Human motivation is describable by a number of theories (Herzberg, Laborit,
Lawler, Maslow, Girard, etc) each of which deals quite well with human
motivation in some circumstances. But none of these theories is complete. Each
accounts for only part of the picture. In principle these partial theories can
be integrated into a more complete theory of human motivation -- we would need
to take account of interactions in the larger 'system'. 

The system here does not really have to encompass the complete range of human
behaviour. Managers and consultants (who are the real problem for the Deming
community) are DIFFERENT from the average man-in-the-street. It's really just
this subgroup of humanity whose behaviour and motivation we would like to
understand better -- so we can find how to influence it for the better. Viewed
this way, the problem would become simpler. That's important because the problem
is, I think, tough enough that we should take every opportunity to abstract and
simplify.

In fact, of course, there is diversity among managers. They do not all behave
the same. But I think that MOST managers do fall into a narrow enough range of
behaviour and motivation patterns that a useful theory could be developed by
focusing attention on some sort of an 'average manager' and ignoring 'outliers'.
Our theory would not attempt to describe perfectly the motivation of EVERY
manager or consultant. We do not need to succeed with every one. It would be
nice just to succeed a lot more than we do now!

I have already seen a 'theory' of the type I am describing. This is in the form
of a model of human motivation, and has been developed to help managers in large
organisations deal more effectively with motivational issues. It is being used
by the French Army to help junior officers cope better with the reactions of the
soldiers under their command (who are often more mature and far more experienced
than the officers). Modern armies are struggling to adapt to the need to allow
soldiers to think and not just obey orders -- often a big break from traditional
military management. It is also being used by industrial organisations to help
managers to understand the motivational factors affecting real sales people.

The point about this approach is that it is not anecdotal or based on memories
of a few 'case studies'. It attempt to be rigorous and to build on the work of
serious psychologists. It makes predictions and is testable. And it can be used
to influence the behaviour of people.

Ian Griffiths
email: 100326.1264@compuserve.com
tel: (44) 01383 872937

==========================================================================

Re: Agressive Response To New Ideas...

"Myron. Tribus" <104055.2663@CompuServe.COM>

23 Jun 96 17:07:05 EDT

Patrick Owings post was encouraging.  It reminded me that many things need to be
taken into account,  beyond leadership. For a successful project,  certain other
things need to be in place.

If you do not have this                 Then you will get this

Leadership                              No action
A clear aim                             No sense of direction
A compelling philosophy                 No followers
A useful vision                         Confusion.  Where are we headed
Strategy for action                     False starts
Skills                                  Anxiety
Resources                               Frustration
Rewards                                 Bitterness
Organization and
   Communication                        No coordination

Myron Tribus

===========================================================================

Re: VARIATION: p- np- u- and c- charts vs. x-Bar and Ind.

"William J. Latzko"

Sun, 23 Jun 1996 18:22:20 GMT

At 16:00 6/21/96 +0000, you wrote:
>I was wondering if any one could comment on the inability of the p-, np-, u-,
>and c- charts to provided information on the "flattness" of a system's
>variation.  
>
>In the context of the red beads and using the data from TNE example,  won't
>an Ind. X/ moving R chart provide essentially the same results?
>
John:

The purpose of a control chart is to determine if special causes exist or
not. A control chart is an operational definition of a special cause.
Remember that Shewhart in designing control charts balanced two mistakes. He
developed the control limits empirically to balance the two mistakes.

Shewhart used a model of the normal distribution. Contrary to some texts, he
did not assume an underlying normal distribution. Nor did he rely on the
Central Limit Theorem. With samples of four that he used, a condition of the
Central Limit Theorem is not fulfilled. Further, the Central Limit Theorem
does not hold for all underlying distributions. Shewhart simply said that an
assignable cause is one where points fall on or outside the three sigma
control limits. Today we use the term special cause rather than assignable
cause.

When we deal with attributes, we use approximations to the normal
distribution in constructing p-, np-, c-, and u-charts. Shewhart's
operational definition of the special cause still holds for these charts. 

 However, we must be very, very careful that we do not mix apples, oranges,
bananas, and other fruit when we get our data to estimate the
non-conformance. If we have mixtures, we introduce the variation between
items. The total variation is the variation within samples plus the
variation between samples if the correlation coefficient is zero. If it is
not, we have to adjust the total variation by the covariance.

Each point on the control chart has an within sample variation. Using these
points in a moving range chart we obtain an estimate of the variation
between points. Since Shewhart's analysis is based on variation within
observations, the use of a moving range chart gives us a different
variation, which by coincidence may be the same. However, it no longer
matches Shewhart's original operational definition of an assignable cause.

You suggested to use the example in The New Economics. I did this for both
the Nashville and the Newport Beach data. The control limits were as follows:

               X, MR chart           np-chart
City           LCL     UCL          LCL    UCL

Nashville        3.30   15.79        0.73   18.36
Newport Beach   0.69   19.44        0.63   18.12

In one case the chart for individuals has a narrower range than the np-chart
while in the other case the opposite is true.

If one adds one more reading of 18 for Nashville and 19 for Newport Beach we
get an interesting result. This change causes Nashville data to indicate a
special cause for the chart for individuals, but not for the np-chart. On
the other hand, the Newport Beach data shows the chart for individuals to
consist only of common causes while the np-chart shows a special cause to be
present.

Using the inappropriate chart can give misleading results. In some cases one
fails to act while in others, tampering can occur.

My suggestion is that rather than try modifying the methodology of analysis,
we look at the data to be sure that it is homogeneous. The very interesting
situation described in your message is common and most often is due to a
problem in data definition.

Bill
---------------------------------------------------------------------------
William J. Latzko
215 - 79th Street
N. Bergen, NJ 07047
Voice:     201-868-5338
Facsimile: 201-868-5338
E-mail:    latzko@worldnet.att.net
------------------------------------------------------------------------------

=========================================================================

Enthusiasm and Pride in Work

Diane Korzeniewski

Sun, 23 Jun 1996 16:48:09 -0400

I attempted to send a message a few weeks back and let's just say that it
was a "lesson
learned".  I have been reading randomly for the past six weeks, or so.  I'll
give a little
background on myself before I get into my topic.  

I am a product designer (structures)  with General Motors Midsize Car
Division.  I am
currently assigned to an assembly plant in the Atlanta area (6-8 months).  I
attended 
my first 4 day Deming seminar sponsored by GM Powertrain in March of this year
and it had some rather strong effects on my thought process.  I hope to
continue 
expanding my understanding of Dr. Deming's work.  With the help of a network
such as this, along with friends I have made thru the seminar, I believe it
will be
fulfilling.  I also plan on attending some future seminars, though I don't
think I'll make
the one in October.   Because I am a fairly new student of Deming, many of
my posts 
are likely to be questions (as I sift thru all that there is). 

I have been pondering a question for some time and would be interested in
hearing
the thoughts of others.   The subject is pride of workmanship.   Joy in
work, enthusiasm,
pride, etc. etc.  Is there much of a difference between these or do they
parallel?  If 
my work environment enables me to have pride in my work, does it also lead to
enthusiasm?  

Do we have an idea (based on any known studies), what percentage of productivity
can be unleashed if management is successful in creating an environment
conducive
to pride and enthusiasm?  How can such a thing be measured......or is this
not possible?

Diane M. Korzeniewski
GM MCD Engineering Design
collabr8@mindspring.com
ToCOLLABR8@aol.com

===========================================================================

Re: VARIATION: p- np- u- and c- charts vs. x-Bar and Ind. x/moving R charts

Andrew Worsley

Mon, 24 Jun 1996 10:33:44 +1000

  I found that Dr. Wheeler's recent book "Advanced Topics in Statistical
  Process Control" from SPC Press makes several interesting points in
  discussing attribute charts.

  In particular he concludes that as the attribute charts are based theorectical
  probability model which makes extra assumptions (beyond Statistical
  Control) it is better to use Individual X and moving R charts if you have
  any doubts about these assumptions.

  As he puts it, the X & R charts limits are based on the distribution
  of the data (empirical limits) and hence much more robust than the
  attribute charts which assume indepedance and constant probability
  per unit area et cetera. With attribute chart's he says you must check to
  see if the theory is appropriate.

  Dr. Wheeler also compares limits obtained with the two different methods
  and even suggests that the X & R chart may pick up special causes
  missed because of it's robustness in his example. From memory one
  point he concides is that attribute charts are considerably more
  efficient in the use of data than X & R. That means to get the same
  degree of certainty about the value of the control limits takes
  significantly more data with X & R charts. But this becomes
  irrelevant if collect more data.

  He does mention though that the X & R charts doesn't show up as
  clearly the kind of tampering with data Deming mentions in "Out of
  the Crisis" in which the peaks are smoothed out. Which might be
  important if it is common as David Kerridge suggests.

    I can post more details from the book such as the exact assumptions which
  the different charts assume of the data and perhaps some examples if
  people are interested. Otherwise I suggest reading the book.

  Frankly I find the whole subject of SPC fascinating. I would be interested
  to here of any examples of whole one can apply it to everyday life.
  e.g. I have tried to plot petrol consumption as kilometer / litre but found
  that longish trips happen sufficiently frequently to disturb almost every
  2nd or 3rd point. If I used more petrol this might become less of a problem
  as one point corresponds to two weeks!

	Andrew Worsley

Australian Artificial Intelligence Institute		 Email: amw@aaii.oz.au
6th floor, 171 Latrobe St.				 Phone: +61 3 9663 7922
Melbourne, Victoria, 3000, Australia			 Fax:   +61 3 9663 7937

==========================================================================

David Kerrige speaks

BCoop1116@aol.com

Sun, 23 Jun 1996 20:41:49 -0400

David, thanks for the help and work you did with the information on the
various charts.  To those of us who are not true math majors these
discussions really help thanks again

Bill Cooper
bcoop1116@aol.com

============================================================================

Latzko speaks on shewhart.

BCoop1116@aol.com

Sun, 23 Jun 1996 20:42:07 -0400

Bill, 
thanks for the very coherent explanation that you gave us on Dr. Shewhart and
the control chart.  Dr. Deming would be proud of these types of discussions
on this net which indeed are helping all to continue to learn and grow.  I
intend to extract some of this material Bill and use it on a client who keeps
wanting to argue the Shewart works.  (hope you don't mind)  I remember some
one asking Dr. Deming why he believed and used Dr. Shewhart's  work so much
and he said "because it works".  Maybe that might be good enough for some of
us as a reason.

Thanks again for the work.  Look forward to seeing you in October.

Bill Cooper
bcoop1116@aol.com

===========================================================================

Myron wrote about leadership.

BCoop1116@aol.com

Sun, 23 Jun 1996 20:42:14 -0400

Myron, talks about the "inaction of leadership".  It is the biggest problem
that we all face in this issue of change strategies or Deming theories.  Why
is it that the very leadership that we must have, and all seek, has become
almost a  non existent capability of our current leaders.  I would like to
see us enter into a discussion of how to get the leadership of and
organization to move forward on any issue that involves change and risk.  

I am sorry that we all as a group did not get Dr. Deming to spend more time
in his writings on this subject of leadership.  He addresses it briefly, in
various writings, but never in his four day seminars did he really focus on
how we might help these leaders move forward on implementing a customer
focused quality program.  Maybe we in this net can use our collective wisdom
and provide to our members what Dr. Deming said about leadership in various
forums and see if we can help some of the people who are dealing with the
various frustrations.

Bill Booper
bcoop1116@aol.com

=========================================================================

Re: VARIATION: p- np- u- and c- charts vs. x-Bar and Ind.

dts@sympatico.ca

Mon, 24 Jun 1996 10:47:17 -0400

At 10:33 AM 6/24/96 +1000, Andrew Worsley wrote:

>  Frankly I find the whole subject of SPC fascinating. I would be interested
>  to here of any examples of whole one can apply it to everyday life.
>  e.g. I have tried to plot petrol consumption as kilometer / litre but found
>  that longish trips happen sufficiently frequently to disturb almost every
>  2nd or 3rd point. If I used more petrol this might become less of a problem
>  as one point corresponds to two weeks!
>

Perhaps the difficulty you're encountering has to do with how you're
measuring fuel consumption - every two weeks, whether you have driven the
car or not. I've been tracking the fuel consumption of my cars for years,
but with respect to distance driven, rather than time. Hence, the abscissa
is the distance, and the ordinate is the amount of fuel. I find it easiest
to reset the trip meter every time I fill up the car. The slight variation
in the amount of fuel per fill-up is too slight to make a significant
difference.

I have, however, noticed a very pronounced non-random seasonal variation in
my cars' fuel consumption (my current car is much more affected by it than
my previous cars). In the winter, the consumption goes up by almost 20%, and
the change is quite rapid. It's almost as if there is a step around 0 deg.
C. This may not be an issue in Melbourne, but it raises an interesting problem.

All current literature on SPC (that I've seen) suggests that one's goal is a
stable process where the data are randomly scattered about some average,
which is a horizontal straight line. What do we do when we know that the
line is neither straight, nor horizontal? If we lump the non-random
variation together with the random variation, we'll obviously have
meaningless results that will tell us that the process is not in control.
But I haven't seen any SPC models that take non-random variation into account.

Another example of non-random variation is the gradual degradation of the
process - I suppose that's where the 'run up' and 'run down' trends come in
- but must one sit and wait until the degradation shows up on the chart when
one knows that the output begins to deteriorate exactly N hours after the
last clean-up?

I'd be interested in learning if anyone has been using SPC for processes
with non-random variation, and how they've gone about it.

Alla Linetsky

========================================================================

Re: VARIATION: p- np- u- and c- charts vs. x-Bar and Ind.

Kromkowski@aol.com

Mon, 24 Jun 1996 15:02:19 -0400

Bill

Your points are well taken and I hope already generally intergrated into my
thoughts.  

What I had hoped to stimulate by the request for comment was not necessarily
a change in method (from p- et al to X, et al), but a deeper understanding of
the tools for understanding variation.  Sometimes the use of tools can be
mindless and not lead to understanding but understanding balance and riding a
bike are also two different things.  Put another way: show me goods works and
I will show you the faith that undergirds them. There are a lot of hacks
talking about variation without using the tools.

You point that >>Shewhart used a model of the normal distribution. Contrary
to some texts, he did not assume an underlying normal distribution<<  is
important.  However, isn't this is only for the the X-bar and Ind. X charts?
 

You suggest that >>When we deal with attributes, we use approximations to the
normal distribution in constructing p-, np-, c-, and u-charts<<  But this is
really not the case, is it? Unlike the  X- charts,  the p- et al charts are
not _approximating_ anything.  The p- et al charts, on the other hand
_require_ and underlying assumption of the poisson distibution. (may need
some help here, I usually have to look this stuff up).  The poission
distribution is the heads/tails (or yes/no) of the mythical fair coin  (a
special type of normal distribution, right?).  And as part of this assumption
isn't there the further assumption of the independence of each event?

The X-bar and Ind. X charts on the other hand recognize a kind of "system
memory"  where outcomes are connected to one another. I want to draw an
analogy to brownian motion but I fear I am in over my head here for a lunch
time post.   These charts are, as I have argued before, making an
"appoximation" of a normal distribution.

You continue:

>>If one adds one more reading of 18 for Nashville and 19 for Newport Beach
we
get an interesting result. This change causes Nashville data to indicate a
special cause for the chart for individuals, but not for the np-chart. On
the other hand, the Newport Beach data shows the chart for individuals to
consist only of common causes while the np-chart shows a special cause to be
present.<<

Think about this for a moment:  "if one adds one more reading".  You mean,
with in the context of the experiment, "if the next willing worker happens to
draw on the next turn . . .."   The Red Bead experiment is not hypothetical.
 

I put the data together when I calculated an Ind. X chart (I like to have at
least 28 points, why? who knows?)  and I don't have it right with me here,
but how does "adding" the data proint you propose (cheating) change the
limits?

How, if at all, would doing the Red Bead experiment with a paddle with only
10 depressions or a million marbles, change what chart one uses?  What if we
don't put the balls back into 
the bin and mix?

John D. Kromkowski

also Kromkowski@aol.com

==========================================================================

Pride in work; joy in work

jcastellano@desire.wright.edu

Mon, 24 Jun 1996 18:53:40 -0500 (EST)

At the time I received Diane Korzeniewski's posting about the difference 
between pride in work and joy in work, I had just finished reading my 
graduate MBA class reflection papers.  I gave them the following 
statement which Tom Johnson develops so brilliantly in his book 
"Relevance Regained"-------Accounting based targets should not be used to 
control people or processes-----and asked them to take a position and 
defend it.

Twenty out of twenty-one students agreed with Johnson's premise.  The 
class is also reading the Deming Dimension by Neave so they are also 
familiar with Dr. Demings views on the dangers of arbitrary numerical goals.

What was most interesting were the personal stories and examples from 
their work experiences about the dangers and dysfunctional consequences 
from using accounting based targets to control people and processes.  The 
examples included distortion of figures, distortion of the system, 
internal competition which resulted in sub-optimization of the overall 
system, fear, and barriers to improvement.

I am convinced that pride and joy in work are not possible in a culture 
that manages by accounting targets and holds employees accountable for 
these numerical goals.  My graduate students during the past four years 
have convinced me that most organizations are clueless when it comes to 
recognizing the dangers from such a management philosophy.

Such a culture fails to nourish the "soul" of the organization and their 
employees.  How can anyone have pride in work and joy in work in such an 
organization?  Every quarter my students tell stories of how numbers and 
processes are distorted in order to "make the numbers". Putting your 
employees in such positions not only does tremendous damage to their 
souls, but also destroys any chance of pride and joy in work.

Deming once said there maybe and excuse for ignorance but there is also a 
penalty for it.  The penalty for using accounting based targets to 
control people and processes is the destruction of any pride and joy in work.


Joe Castellano
JCastellano@desire.wright.edu

======================================================================

Re: Aggresive Response to New Ideas

"Anton O. Tolman, Ph.D."

Mon, 24 Jun 1996 13:05:59 -0700 (MST)

I have been enjoying the discussion on how to deal with aggressive 
responses to teaching Deming principles, especially the wonderful 
posts by Robert Crow.  I would suggest two ways of understanding how 
an aggressive response might occur.  This might help us understand 
how to approach persons with such a response.  

First, along the lines that I have posted before, people tend to 
think irrationally.  This means that they develop their own theories 
of what is "supposed" to be, what "should" be, and how life "ought" 
to be run or how a company ought to be run.  These theories or 
explanations or models for themselves tend to be irrational because 
they are not based so much on "reality" or data, but are based 
instead upon training, history, learning, and models by powerful 
others.  For example, if I learned from a powerful administrator that 
things "should" be such and such, then as I develop in my own career, 
I tend to believe that that is the way things ought to be.  Why?  
Because I want to ultimately become powerful like he/she was and that 
is what they believed.  I don't want this to seem like a logical 
conclusion reached through a careful thought process, because most of 
the time it is not.  Anyway, then you come along and tell me, "No, 
performance appraisals belong on the dungheap."  Well, that flies in 
the face of what I thought was self-evident, what I thought was the 
way things should be -- and so I react emotionally.  

Clinically, I borrow from the Bible and say that we often "kick 
against the pricks" meaning that we beat ourselves against the way 
the world really is.  I think most people, after being presented with 
a forceful, logical, Deming-based argument or demonstration such as 
the funnel or red-bead experiment, begin to see the light -- begin to 
see how this different way of viewing reality fits better than the 
old one.  However, we have to first overcome the irrational shoulds 
in order to make our point.

Second, cognitive dissonance theory fits well here and is really just 
a spin on what I said above.  Many experiments have indicated that 
when people can find an external reason for their behavior, they are 
much more willing to discount its value.  For example, if I pinched 
my sister because a friend told me they'd give me a dollar, then that 
is the reason I did it.  I don't particular like pinching my sister, 
I am not a mean person, I did it for the money.  If I pinched my 
sister with no external reason, I experience an unpleasant state 
called in psychology, dissonance.  This means that I cannot reconcile 
my behavior with the type of person I think I am.  I can resolve this 
by decided that I _am_ mean, and acting that way, or by apologizing 
and becoming friends with my sister, but I am motivated to eliminate 
the dissonance.  For people who have adopted or "bought" the old way 
of thinking that Deming rails against, to consider these new thoughts 
may create dissonance.  Think of the supervisor who for years has 
graded people and watched how it affected them, but who thought this 
was the way it "should" be or who was trained to believe this.  One 
way to resolve the dissonance is to more firmly entrench myself in my 
position, to prove to you that you are full of it, and failing that, 
to become aggressive enough to get you to back off and take your 
ideas away.  The other way, of course, is to embrace it, examine it, 
and use the new ideas.  

These understandings help us look at different ways to approach 
people who are resistant.  For example, can we offer them an external 
reason for the existence of the previous system?  Sure, it made sense 
then, we all believed it, it was the zeitgeist of the time, etc., but 
there is a new opportunity now to see it in a new way.  We can also 
attack irrational thinking by the use of experiments such as the red-
bead (or Green M&M) ones or by helping them explore ways not to 
"should" on themselves or others.  One way to do this is to point out 
the famous bumper sticker -- "Sh*t Happens" and explore why do we all 
react to this.  If should were true, would this sticker be true?

There are lots of other ways to perhaps impact people's reactions to 
our message.  Let's continue to explore them.

Anton O. Tolman, Ph.D.
Director, Psychological Services
Director, Program Services
Wyoming State Hospital
P.O. Box 177
Evanston, WY  82931-0177
Anton@wsh.state.wy.us
(307) 789-3464
====================================================================

Re: VARIATION: p- np- u- and c- charts vs. x-Bar and Ind.

Andrew Worsley

Tue, 25 Jun 1996 10:34:52 +1000

> At 10:33 AM 6/24/96 +1000, Andrew Worsley wrote:
> 
> >  Frankly I find the whole subject of SPC fascinating. I would be interested
> >  to here of any examples of whole one can apply it to everyday life.
> >  e.g. I have tried to plot petrol consumption as kilometer / litre but found
> >  that longish trips happen sufficiently frequently to disturb almost every
> >  2nd or 3rd point. If I used more petrol this might become less of a problem
> >  as one point corresponds to two weeks!
> >
> 
> Perhaps the difficulty you're encountering has to do with how you're
> measuring fuel consumption - every two weeks, whether you have driven the
> car or not. I've been tracking the fuel consumption of my cars for years,
> but with respect to distance driven, rather than time. Hence, the abscissa
> is the distance, and the ordinate is the amount of fuel. I find it easiest
> to reset the trip meter every time I fill up the car. The slight variation
> in the amount of fuel per fill-up is too slight to make a significant
> difference.

  Actually I am measuring consumption per distance travelled, obviously
  I didn't explain that clearly. It's just that we make mostly small
  trips so that we only fill up once every two weeks. If we drove more
  we fill up more often and get more data points. Once in a while we go
  on a long trip and the "mileage" improves up to around 10k/l (from
  5-6) which is an clearly an out of control point. But this happens so
  often that I am excluding over half my data points!

  Another source of data is measuring your weight. But you need scales
  that are particularly accurate, at least .2kg (.5lb?) to detect the
  minor fluctuations and even then you need to subgroup them in (say
  4-7 pts) to get enough variation to avoid insufficient sensitivity
  problems that Dr.  Wheeler talks about.

> 
> I have, however, noticed a very pronounced non-random seasonal variation in
> my cars' fuel consumption (my current car is much more affected by it than
> my previous cars). In the winter, the consumption goes up by almost 20%, and
> the change is quite rapid. It's almost as if there is a step around 0 deg.
>C. This may not be an issue in Melbourne, but it raises an interesting problem.

   I believe the efficiency of a car engine does vary a lot with tempreture
   and hence it is standard practice to adjust the air intake for summer
   or winter, even in Melbourne.
> 
> All current literature on SPC (that I've seen) suggests that one's goal is a
> stable process where the data are randomly scattered about some average,
> which is a horizontal straight line. What do we do when we know that the
> line is neither straight, nor horizontal? If we lump the non-random
> variation together with the random variation, we'll obviously have
> meaningless results that will tell us that the process is not in control.
> But I haven't seen any SPC models that take non-random variation into account.

  I have seen this sort of situation in timing data for running or
  swimming laps. It can also vary over time. I don't know the cause as
  the data was collected well before being plotted. Dr. Wheeler
  suggests that you need to plot the data as close as possible to where
  it's collected - so as soon as you see a OOC (Out Of Control) point
  you can look around for special causes, perhaps note the conditions
  on the graph.

  If the cause is understood perhaps you can control it - or (such as
  for the weather) alter the process to suit the change? If not perhaps
  you can subtract it out.

> 
> Another example of non-random variation is the gradual degradation of the
> process - I suppose that's where the 'run up' and 'run down' trends comein
> - but must one sit and wait until the degradation shows up on the chart when
> one knows that the output begins to deteriorate exactly N hours after the
> last clean-up?

   According to the example of the cigar lighter (from one of Dr.
   Wheeler's books) degradation from wear does show up rather
   suddenly.  I presume that it's affect is minor until a certain level
   is reached and then suddenly it dominates the other (common) causes
   of variation. i.e. until generates OOC points you don't need to
   worry about it, as it's affect is below that of other common causes.

> 
> I'd be interested in learning if anyone has been using SPC for processes
> with non-random variation, and how they've gone about it.

  Non-random variation is probably a good definition of OOC variation.
  I think that with such variation, if you can't control it or subtract
  it out you are limited in what you can do.

  Another interesting situation is when the data is highly
  auto-correlated.  (i.e. successive points are correlated) This often
  occurs when measuring some changing attribute (tempreture/pressure).
  It's value is not completely random as it's next value depends on
  it's current value, although after enough time, it's actual value is
  random.  The weather seems to be a good example [I was told 75% of
  the time the weather tomorrow will be like today].  In this case Dr.
  Wheeler suggests if it's correlated above .8, you can expand the
  control limits by a factor which he gives to allow for this. But I guess
  you lose sensitivity.

  I see similar auto-correlation cycles in the lap time running data. I am
  not sure if this is fitness comming and going or a physiological cycle of
  the body of the order of a month. Any doctors of medicine out there?

  I too would like to know how other people handle these situations.
  Does it represent the limits of what we can do? e.g. If you can't
  eliminate these sorts of variations or calculate them out, does that
  mean you can't reduce variation much more?

	Thanks
		Andrew Worsley

==============================================================================

Re: VARIATION: p- np- u- and c- charts vs. x-Bar and Ind.

guym@statware.com (Guy W. March)

Tue, 25 Jun 1996 8:43:19 -0700 (PDT)

:> 
:> All current literature on SPC (that I've seen) suggests that one's goal is a
:> stable process where the data are randomly scattered about some average,
:> which is a horizontal straight line. What do we do when we know that the
:> line is neither straight, nor horizontal? If we lump the non-random
:> variation together with the random variation, we'll obviously have
:> meaningless results that will tell us that the process is not in control.
:> But I haven't seen any SPC models that take non-random variation into account.
:> 
[snip]
:> 
:> I'd be interested in learning if anyone has been using SPC for processes
:> with non-random variation, and how they've gone about it.
:> 
:> Alla Linetsky
:> 
:> ========================================================================
:> 
:> 

I agree with Andrew that the subject of SPC is fascinating. 

Alla, we have addressed the non-random variation for some our clients by using
some statistical methods to take the non-randomness out of the data. 

On the "run down"/"run up" data, we run a regression analysis on the data, then
plot the residuals to the Control Chart. 

The gasoline consumption problem might be addressed by running a time series
analysis on the data, and plotting the residuals on a Control Chart.  The
trick is, of course, finding the best model for each of these.  

Both these methods create a horizontal plot of the data.  But we have also done some work with using control limits from these calculations to calculate the 
control limits on the original data.  Then we run a Control Chart on the
original data with the control limits set as calculated.

This can lead to possibly having 4 charts instead of the traditional 2:
A de-trended chart and recalculated limits chart for both a location chart and 
a dispersion chart.

Another chart we have been working with is suggested by your comment about:
"output begins to deteriorate exactly N hours after last cleanup".  This is 
what we call a Step Chart or Phase Chart.  The process goes thru a number of 
phases that would have separate control limits.  For example: History, Trial, 
Implementation.  This chart calculates the control limits for each phase and 
plots the data on the same chart.

Some real interesting stuff.

Oops, I don't think I've introduced myself to this list.  
I'm Guy March, Training Coordinator for Statware.  We produce Statit, a Data
Analysis tool that, among other things, does SPC charts.  Runs on Unix, Win95
and NT.

I am involved in Customer Training and Support and have been in that field for
many years. I have a technical educational background, Nuclear Engineering and 
Numerical Analysis. 

With my schedule, I have mostly lurked on this list, but as you say, Andrew,
SPC is fascinating and I just had to comment on this idea.

Guy W. March
guym@statware.com

===========================================================================

Masters Degree...Help!

"Kerr, Donald"

Tue, 25 Jun 1996 08:54:00 -0700

I'm looking for DEN advice and direction on what Masters Programs are
available related to The System of Profound Knowledge.

Here's my situation.  I have a B.S. in Mechanical Engineering 12/90.  5+
years experience in engineering, project management and leading Total
Quality (Training & Facilitation) in the Aerospace industry (NASA &
AlliedSignal).  I also have been leading the Greater Houston Deming
Alliance for about four years.  All my education has come from 5 years
of self-study and practice on the front lines.  In a world of
credentials, this has been a hard sell in landing a job managing
organization-wide Continuous Improvement.  Internal opportunities are
non-existent because we are downsizing.  I'm frustrated and have decided
I need more "official" education or "Pigskin" to accomplish my aim. 
Communicating Deming's Ideas in an 20 minute interview with a
traditional HR type just doesn't work.  ISO9000 is their bottom line.

My questions are many and open to any new ideas you may have.  Are there
any good distance learning programs?  Are there any MBA or other
interdisciplinary programs that I can take courses in
philosophy/epistemology, psychology, variation, and systems theory? What
alternatives to an traditional MBA do I have?  How do I find gold in
such a wide array of programs?  What is good way to do research?  I want
to explore and create new ideas with the System of Profound Knowledge
and at the same time get a marketable degree.  (I'm even considering
Religion studies.)  Am I asking the right questions?

As you can see, I'm struggling.

ANY suggestions you might have would be greatly appreciated.

Thanks!

Have a Great Adventure!
Don Kerr

===========================================================================
    

Re: Pride in work; joy in work

John Woods

Wed, 26 Jun 1996 10:52:29 -0500

Joe Castellano talks about a culture that is based on numerical accounting
goals and holds individuals responsible and says:

>Such a culture fails to nourish the "soul" of the organization and their 
>employees.  How can anyone have pride in work and joy in work in such an 
>organization?  Every quarter my students tell stories of how numbers and 
>processes are distorted in order to "make the numbers". Putting your 
>employees in such positions not only does tremendous damage to their 
>souls, but also destroys any chance of pride and joy in work.

I'd like to suggest that the organization and the people who make up the
organization are the same thing.  By dehumanizing the organization, you
ignore this fact of life, and there is indeed a penalty.  This penalty came
be summed up in one word: mediocrity.  On the other hand, we should
appreciate that no one would knowingly take actions that result in
mediocrity.  So the problem is one of lack of knowledge and understanding of
the principles on which quality management practices are based and how to
effectively implement those principles and practices.  We all have a
responsibility to keep trying to help people appreciate these ideas,
especially an understanding of the systems view of organizations.  For from
that idea, we can derive everything else.

John Woods
jwoods@execpc.com

====================================================================

Agressive Response to new ideas

jr1crow@mindspring.com (James Robert Crow)

Wed, 26 Jun 1996 13:41:07 -0500

        I would like to pursue this a little further if we can.  Yesterday I
was talking with a Vice President of Customer Service, and Stragetic
Planning for a rather large temporary service company in Atlanta.  He is
interested in optimization, but doesn't really understand it.  I had send
him the two articles I have written on the subject of competition and
measuring perfromance.

     One of the articles has an example regarding ranking plants.  He opened
the conversation by stating that he didn't agree with my point on ranking.
In the course of our conversation I determined that he has recently
instituted just such a policy at his company where all the branch offices
are now ranked.  He also made the statement that figures don't lie, and that
with these figures he will be able to tell which are the best run operations
and which are the worst.  Since he had the article on performance appraisals
in which I show that while there is variation in performance, all of the
variation is within the control limits and is the random variation being
caused by the system, I suggested that we might do a control chart using his
figures for the branches to see if there were any outstanding branches, or
would the chart  only show the common cause variation being caused by the
system.  He resisted this, saying that he only had two quarters of data and
doubted that it would be sufficient.

     My problem is how to break through the wall and get him to look at
things from a different perspective.  The comments by Anton Tolman, and Joe
Castellano are very appropriate here.  He has learned the way it should be,
and is not very willing to listen to an outsider who is questioning some
recent changes he has made.  The objective here is to turn him on without
turning him off.  Any ideas?

Robert Crow
President, The Crow Group
jr1crow@mindspring.com

========================================================================

Re: Masters Degree...Help!

ICMUSA@aol.com

Wed, 26 Jun 1996 13:58:19 -0400

In a message dated 96-06-26 11:40:06 EDT, you write:

>To:	den.list@deming.ces.clemson.edu (Deming Network)
>
>I'm looking for DEN advice and direction on what Masters Programs are
>available related to The System of Profound Knowledge.
>

>My questions are many and open to any new ideas you may have.  Are there
>any good distance learning programs?  Are there any MBA or other
>interdisciplinary programs that I can take courses in
>philosophy/epistemology, psychology, variation, and systems theory? What
>alternatives to an traditional MBA do I have?  How do I find gold in
>such a wide array of programs?  What is good way to do research?  I want
>to explore and create new ideas with the System of Profound Knowledge
>and at the same time get a marketable degree.   Am I asking the right
questions?

ANY suggestions you might have would be greatly appreciated.
>
>Don Kerr
*******************************
Don,
You bet you are asking the right questions. I've just begun a Ph.D. at Walden
University, There are a number of areas of focus including social change,
psychology, theories of learning and motivation.  One of the components is in
general systems theory.  I will be examining breath, depth, and application.
 I planning on examing the TQM movement and possible application to
non-traditional schools.  One of my mentors gave a presentation on TQM last
night, including a video by Deming.  I have been very satisfied by the
program thus far.  It's non-tradition education at its best.  I believe they
have an M.A. program.  You would not want to make a final decision without
looking at what Walden may have.  For information, you can contact
DKarnes@waldenu.edu.  I'm sure he can answer your questions.

Kind regards
Myron Goodwin     
mgoodwin@waldenu.edu

====================================================================

Re Masters Degree Help!

jmartin@mail2.lmi.org

Wed, 26 Jun 96 14:15:35 EST

     This is my first post to the den.  I have been reading the den for six 
     months and really enjoy the lively discussions.  This is the second 
     request I have read for info on an MBA program on Dr. Deming's 
     teachings.  I received my MBA in 1984 and I don't remember Dr. 
     Deming's teachings ever being mentioned.    
     
     Fordham University in New York has a Deming program that Joyce Orsini 
     can provide the info on.  I do not have an e-mail or phone for Joyce, 
     but I am sure someone else on the den would have that info.
     
     Marymount University in Arlington, VA has a TQM certificate program.  
     I don't know if it stresses Dr. Deming's teachings.  Contact point is 
     Dr. Fry at (703) 284-5963.
     
     
     Jim Martin
     Board Member, Washington Deming Study Group
     jmartin@lmi.org
     
=========================================================================

Masters Degree...Help!

Expira AB

Wed, 26 Jun 1996 21:56:31 +0000

Comments on Masters Degree on Profound Knowledge,

Of course the answer to this has to come from those people being able t=
o give
authorized Masters Degree in Deming's Knowledge. Anyway, I want to give=
 a
piece of hope to those who want to learn about Deming but can not liste=
n to
him in person any more. I have myself been waiting for distance educati=
on in
Deming's philosophy and in the meantime I keep on learning and applying=
Deming, maybe more than I ever would do by carrying out a Master's Degr=
ee.

I attended Deming's four day seminars twice, -89 and -90. I shared some=
 letters
with him, with more spiritual content than scientific facts. The Deming=
 advice that
has meant most to me is: Show a positive, kind attitude towards anyone =
wanting
to learn. Keep learning all the time. Keep mind working. Read the book =
and
subject you feel important for the moment, even Spanish language of tha=
t gets up
to the "Pareto surface".

If your logic is stuck and emotions are down: go to church, concert or =
some other
personal inspiring event. Often the body is lacking oxygen when we sit =
here
writing emails to the Clemson-host. Physical training: body building an=
d jogging
might be much more important than Out of The Crisis certain hours.

Sometimes we must learn from the Japanese samurais to keep our brain in=
 a No
Mind state, meditating, praying, filling ourselves with holy spirit etc=

Marie-Louise
expira@postbox.postnet.se

========================================================================

Aggressive Response to New Ideas

Walker Associates

Thu, 27 Jun 1996 15:19:20 +0000

Robert Crow has a client who has just gone for ranking branch 
offices and asks for ideas on how to address it.   Here is one 
approach.

As the client has just made a decision to go for ranking, it will be 
counter-productive to pursue this aspect directly, so soon after the 
decision.   This is why salesmen are taught not to criticise a 
product the client has just bought.   The question is, Why does the
client need to know who is best or worst?  i.e. What is the
purpose of the ranking.   My guess is that he will explain that it is
in order to make improvements, and this is where you can offer to
help him.

Unfortunately, of course, you may find that the competition that 
ranking introduces makes it more difficult to get branch offices to 
share their good ideas with each other (Why prejudice your chance of 
coming top?)   You may also find that the competition causes the 
branch office to distort the data they collect in their own favour.

Thus you can praise your client for wishing to manage by fact and 
manage on the basis of data.   But then you need to help him collect that
data and use it more effectively in the PDSA improvement cycle.   
Hopefully this will lead away from the use of data for ranking and a 
focus on process measures rather than results measures.

Regards, Peter Walker.
-------------------------------
Walker Associates
Tel:+44 1908 510593 Fax:511772

===============================================================================

Fordham address

Bill Towns

Thu, 27 Jun 1996 07:05:47 -0600

To the Den List:  Joyce Orsini's Address is:

                            Fordham University 
                            113th 60th Street 
                            New York, New York  10023

I also have her phone number but do not feel comfortable with giving it out
on such a public venue.   My name is Bill Towns, I am at the University of
Kansas and my e-mail is wct@union.wpo.ukans.edu.   I wrote my thesis on
Deming as an educator and am interested in any application of Deming's
principles to education.   Thanks

======================================================================

"joy in work"

"Walter G. Prevalnig" <104337.2101@CompuServe.COM>

27 Jun 96 13:09:51 EDT

re: Diane Korzeniewski's question regarding "joy in work".

Dr. Deming spoke about intrinsic motivation. I believe that this is a
prerequisite to "joy in work". No matter what the task is, be it grand or
mundane, if we are intrinsically motivated to do it we will experience "joy in
work". In the absence of intrinsic motivation that same task becomes drudgery.

Walter G. Prevalnig . Building Learning Organizations
104337.2101@compuserve.com

===================================================================

Re: Enthusiasm and Pride in Work

sta010@abdn.ac.uk

Thu, 27 Jun 1996 21:44:52 +0100

Diane Korzeniewski asks if we can estimate the percentage of productivity
that could be unleashed if management is successful in creating an 
environment conducive to pride and enthusiasm.

I don't think we can, because the long term effects are much greater
than the short-term effects: and above all, it depends what we include
in "Productivity"

In a sense, management has gone through three stages:

1   Productivity in the sense of total volume of production is everything

2   Quality matters more than crude numbers: it affects profits more

3   Creativity, including innovation in product and process, and the
    ability to respond to a rapidly changing world, are most important
    of all. Who needs the world's best stage-coach?


People can, and sometimes do, work under miserable conditions and still
produce quantity. They are less likely to produce quality: that requires
intelligent application. And they cannot produce innovation: they will
fight every change, knowing they will never gain from it.

So if you measure on the ultimate scale, long run survival, the benefits
of joy in work are too large to measure. On the crude numbers scale, they
may be less impressive in the short run.

David Kerridge


BDA Scotland

dfk@abdn.ac.uk

========================================================================

Re: Myron wrote about leadership.

sta010@abdn.ac.uk

Thu, 27 Jun 1996 22:01:08 +0100

Bill Cooper suggests that Dr Deming did not say enough on leadership.
I think he did, but we may not recognise it.

In many things, Dr Deming did not tell us how to push the system towards
the desired result, whether quality, or profitability, or in this case
leadership. What he showed us was how to remove the obstacles, so that
the things we desire may flourish.

This is an important change of view. Most people, leaders as well as
workers, would rather do a good job than a bad job. Under the present
system, managers are overworked, overstressed, and naturally cannot
do the good things that would help everyone else - like learning new
ideas.

Just take a few "Deming" ideas. Stop reacting to common cause variation,
and you have more time, and less stress. Have fewer suppliers to deal
with, and the manager has fewer disputes to resolve, and muddles to
sort out.

These are not the only reasons for changing to the new ways, of course.
We usually stress the benefits to quality or to the workers. But the
whole basis of the Deming Philosophy is "Everybody wins".

I believe the answer is that the 14 points are (among other things)
a way to produce leadership. The old systems destroy leadership as
surely as they destroy motivation, quality, and profits.

David Kerridge


dfk@abdn.ac.uk

========================================================================

New ideas

jkaliste@skipjack.bluecrab.org

Mon, 1 Jul 1996 19:50:45 -0400 (EDT)

        The key word is new.  I often try to impose my will on others.  It 
may be beneficial to establish a good working relationship before exhibiting 
an aggressive new idea.  The person on the receiving end of this new idea must
have status of some type.  People in authority fall victim to status 
equalibration, Dr. M. Sherif's term for people who have status in one aspect,
who desire to have it proportionally in all.
        A conflict develops when a side seeks to impose it's will, or gain
advantage over others. You will build temporary coalitions of strange 
bedfellows who will unite in opposition to you. This situation is at best win
lose.  It could have many losing sides, with none winning.
        Keep in mind that the key is allowing all parties to save face, win, 
if you will.  This will stabilize the relationships. As Dr. Deming says,
it is important to have long term relationships.
        Listen a lot.  It is unwise to pour anything in to a full container.
Be patient, let people talk themselves out. Let them empty themselves of all
of their arguments, then let them have time to think of more.  This is not
High Noon.  It is difficult to listen.  Let people have time to think, 
or they will think about what they will say next when you are talking,
and ignore all your hard earned knowledge.
        You can lose a few battles and still win the war.
        Is this leadership ?

                                            Thanks,
                                                 John
jkaliste@skipjack.bluecrab.org

=========================================================================

Profound Knowledge

Steven_S_Prevette@RL.gov

Tue, 2 Jul 1996 16:20:06 -0700

--IMA.Boundary.827943638
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Description: cc:Mail note part

     I have been following the Profound Knowledge and Qualification vs 
     Understanding discussions with interest, and let me offer two models:
     
        Star Trek analogy
        Robert McNamara's "In Retrospect"
     
     The Start Trek analogy was taught in a management course at the Navy 
     Officer Candidate School when I attended in 1981.  Briefly, the idea 
     is that "Dr. McCoy" represents the emotional/experiential point of 
     view - all of his solutions are based on feelings and gut reactions.  
     On the other hand, "Mr. Spock" represents pure logic - or the use of 
     data (and the theory of variation in our context).  Neither approach 
     works well stand-alone, but our hero, Captain Kirk, is able to bring 
     the two differing approaches together for a well-rounded answer to the 
     crisis at hand.  The Navy did not refer to this as "profound 
     knowledge", but I think this model may be useful.
     
     I believe that Robert McNamara's "In Retrospect" contains important 
     information on over-reliance on data in decision making.  I know that 
     in my current job I am attempting to swing managers away from strictly 
     "gut reaction" decision making, to including data in their decision 
     making process (and the theory of variation).  However, I am mindful 
     not to overswing the pendulum as Mr. McNamara did.
     
     [Another Navy story]  While I was at Naval Postgraduate School for an 
     MS in Operations Research, there were frequent references to Robert 
     McNamara and how he gave operations research a bad name by 
     over-reliance on data, without taking into account the experiences of 
     senior military leaders.  I have found reading "In Retrospect" a 
     worthwhile experience, and I think by negative example, he points out 
     the need for "profound knowledge".  
     
     I do think it would be very worthwhile to put the question of profound 
     knowledge to Mr. McNamara, if possible.  His book appears to catalog 
     the shortcomings of the decision making process during the Vietnam 
     war, but doesn't appear to offer an answer.  But I still recommend the 
     book, if only as a unique example of going too far with data, without 
     considering the human element.  
     
     - Steve Prevette
       Performance Measures Consultant
       Westinghouse-Hanford Corp.
       509-373-9371
       steven_s_prevette@rl.gov
--IMA.Boundary.827943638--

=========================================================================

"joy in Work"

"Walter G. Prevalnig" <104337.2101@CompuServe.COM>

02 Jul 96 11:38:39 EDT

Ned Namson wrote: < In the absence of intrinsic motivation.....which was made
concerning Dr.Deming and joy /pride in work statements. This confused me a bit.
Intrinsic motivation is internal motivation,no?.......>

Ned, I concur with your comment and apologize. In the interest of brevity I
erred in not providing an operational definition for my  use of the term "
intrinsic motivation".

For this posting I used the term "intrinsic motivation" to represent   that part
of the internally generated frustration signal that starts us behaving to get
what we want.	

Recognizing that all behaviour is internally motivated , the purpose of my
posting was to make the point that if I choose to perform a task because it gets
me what I want and is therefor needs fulfilling, I experience "joy in work". If
I choose to do the same task due to an external condition such as fear ( if I
don"t do it I loose my job,  etc.) then that same task becomes drudgery.

Walter G. Prevalnig,  Building Learning Organizations
104337.2101@compuserv.com

==========================================================================

Intrinsic motivation?

RonDavison@aol.com

Tue, 2 Jul 1996 13:42:21 -0400

 Someone asked a few days ago, "Isn't intrinsic motivation the same as
internal motivation?"  Therefore, isn't it always present?

No, intrinsic motivation does not refer to motivation that lies within you,
but motivation that lies within a particular task.  For instance, if you love
to compose music because you love to compose music, you find motivation to
compose intrinsic to the task.  On the other hand, if your motivation to
compose music comes from outside of the task --  you see composing as a means
to fame, fortune and women -- you are extrinsically motivated.  Your
motivation to compose music lies outside the task.

Students could learn how to calculate acceleration because they want an "A"
or because they are genuinely fascinated with free-fall.  Who do you think is
more likely to "play" with the concepts enough to arrive at an innovative
approach?

Repetitive jobs can easily be shipped overseas to be worked by Vietnamese
making $30 a week.  So, we need to ask ourselves whether we're using work and
learning methods that lead to expected or innovative results.  Repeatedly,
research concludes that extrinsic motivation stifles innovation and intrinsic
motivation stimulates it.   Someday this simple concept with be both widely
known and widely accounted for in the design of work and learning.  Until
then, we'll continue to wonder why our median incomes rarely rise and our
trade deficit rarely falls.

--
Ron Davison (RonDavison@aol.com)

=========================================================================

Deming Question #18

The.TQM.BBS@tqm.permanet.org (The TQM BBS)

02 Jul 96 10:05:00 -0500

   From: John Hunter
     To: All
Subject: Deming Questions # 18

>>> Dr Yoshio Kondo, Professor emeritus, Kyoto University, said that
there are three concepts of product quality: (1) conformance to
specifications, (2) satisfaction of the customer, (3) amazement of
the customer. What is the best one to keep faithful customer?
Jean-Marie Gogue <<<

Amazing the customer is the best way to keep repeat customers.  Satisfying
them is good but if they see an alternative that is cheaper or may be
better
or if you make a mistake (introduce a new product they don't like, raise
prices at the wrong time) they may well leave.  If they have been amazed
by
your past service they are unlikely to leave for a slightly lower price,
or a
new competitor or a new product/servcie similar to yours or if you make a
mistake.

    John Hunter
    wdsg@aol.com
    http://pages.prodigy.com/john/hunter.html
--
|PerMaNet       : The TQM BBS 11:202/299
|Internet       : The.TQM.BBS@tqm.permanet.org
|
| Standard disclaimer: The views of this user are strictly his own.

=========================================================================

Re: "joy in Work"

ParetoKid@aol.com

Wed, 3 Jul 1996 07:22:22 -0400

Walter: Thanks for the added explanation on intrinsic. The point you make is
right on target with the added explaination.

Fear, of course, not the only roadblock to joy/pride in work/task. Everyday
hassles caused by out of control "little" processes, as well as the main or
big processes also sap energy. I prefer to use the word "task" for things I
must do in my part of the process of serving cuustomers. Acheiving joy in
task is not difficult. When big and/or little processes are out of control
parameters, my tasks become "work." 

Cheers, Ned Hamson, editor: The Journal for Quality and Participation -
Association for Quality and Participation, 801-B W. 8th St., Suite 501,
Cincinnati, OH 45203
Tel: 513-381-1979 Fax: 513-381-0070 - e-mail: ParetoKid@aol.com
"This is the time... We are the people... Let's work together... Now!" Ned
Hamson
"Always do right. This will gratify some persons and astonish the rest." --
 Mark Twain
http://newciv.org/worldtrans/comqual.html
European representative: Peter Beerten, Belgium Consultancy Group: Tel:
011-32-02-569-0222/Fax: 011-32-02-569-7480 - e-mail: Peter.Beerten@ping.be
Woe, 03 jul 1996 

========================================================================

Svar: Agressive Response to new ideas

Marie-Louise.ThorsenLind@NOTIS.postnet.se

Wed, 3 Jul 96 03:33:24 EDT

        I doubt that people will change an accepted point of wiew
         unless they believe that they will benefit from it.
        So you have to be persistent. Try to make him see
         the results that his company could make with another point of =
wiew.
        You might need examples. Many executives don't accept theory on=
ly
        ecpecially when it goes against "common knowledge".
        Marie-Louise Thors=E9n Lind
        The Swedish Employers Confederation
        marie-louise.thorsenlind@notis.postnet.se

______________________________Svarsavskiljare _________________________=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=

========================================================================

Over reliance on data?

agc@vnet.IBM.COM

Wed, 03 Jul 96 10:17:24 EDT

*** Reply to note of 07/02/96 21:57
Alan Crudden         INET: agc@vnet.ibm.com |  (203) 973-7143  |
290 Harbor Dr.    IBMMAIL: USIB49ZM         |    T/L 342-7143  |
Stamford, CT  06904                         |FAX T/L 342-5465  |
Subject: Robert McNamara

Re: Steve Prevett's post on over-reliance on data

The problems of McNamara, et al seemed to be (from my readings) not so
much a reliance on data but in using measurement as a "management system"
and not as a *tool* for helping to understand the management system
itself.

His methods seemed to encouraged deceptive practices so that quotas and
targets could be achieved.  At one point in Vietnam, I read, enemy body
count quotas weren't being met so a new method was instituted: count an
abandoned rifles as the equivalent of a "kill".  The rationale(?) behind
this was that no good, live soldier would leave his gun behind.

I recommend David Halberstom's "The Reckoning" for more on McNamara's
work and its specific impact on the automotive industry.

Alan

====================================================================

Visit to US

Peter Rose Associates

Wed, 3 Jul 96 18:58:30 EDT

I am on vacation with my wife this July and wondered if any of your Deming 
circles were having meetings when we were in your area.  I see so many 
interesting topics being discussed and would love to attend a session.
We will be in Seattle 19th and 20th July then travelling down to San Fransico 
for 26th and 27th.
Please let me know if there are any meetings on those dates that we may be able 
to attend.
-- 
---------------------------------------------------------------------------
| Peter Rose Associates    EMail From Peter Rose peter@proseass.demon.co.uk     
                              |
| Mail sent via Demon Internet - Tel(44) 01565 651159 or fax 01565 634948  |
---------------------------------------------------------------------------

============================================================================

Intrinsic Motivation/Owings

"Patrick Owings"

Fri, 5 Jul 96 13:25:18 UT

In regard to the posts on intrinsic motivation, I wonder where "the love of 
music" that is "intrinsic to the task" comes from.  What I am getting at here 
is the need for Deming practitioners to gain knowledge of how one may go about 
identifying the most appropriate job assignment for a particular individual in 
the workplace.  Obviously, being placed in (or selecting) a vocation in which 
one finds such intrinsic motivation would be highly desirable.  I know many 
lawyers who are miserable in the practice of law and wish they had chosen 
something else as a profession.  These are bright people who plunged into a 
lifelong pursuit, perhaps without full knowledge of what it was really like to 
be an attorney (how could know?).  Many now feel that they have invested far 
too much of their lives in law practice to change now.  I also know engineers 
and physicians with similar intrinsic motivation problems.  I am aware of 
Briggs-Meyers and other means of choosing a vocation. Do tests these enable 
one to select a job that will have inherent intrinsic motivation for the 
individual or is this an area of psychology requiring more research before we 
have a useful tool?

These are, I'm sure, elementary questions for many on the DEN, but I wonder 
if, when discussing intrinsic motivation versus extrinsic motivation, we ask 
the question, "Is it possible to find intrinsic motivation in life itself and, 
therefore, find intrinsic motivation in almost any job we do?"  In other 
words, Dr. Deming's "joy in work" - is it more from offsetting (or 
eliminating) the "forces of destruction" in one's life or is it the result of 
some unexplained natural proclivity toward "a love of music", or law, or 
medicine, or whatever we love to do?  In my opinion, this discussion has great 
implications for the implementation of The Fourteen Points and is but one of 
many reasons for SPK.  Without a knowledge of psychology, one cannot 
successfully implement The Fourteen Points.  

An example one could give is that of an engineer desiring to design an 
airplane.  In order to come up with a successful design, the engineer must 
possess a knowledge of geometry, algebra, calculus, statics, dynamics, 
mechanics of materials, fluid mechanics, aerodynamics, and a host of other 
things.  These could, I suppose, be an equivalent to SPK for designing an 
airplane.  In other words, SPK is something akin to what knowledge an engineer 
would need to design an airplane, but involves what a manager would need in 
order to implement The Fourteen Points.  If any piece of the required 
knowledge is missing, the engineer does not come up with a successful aircraft 
design.  Correspondingly, if any piece of SPK is missing, successful 
implementation of The Fourteen Points is unlikely for the manager.

The need to get at the depths of the origin of intrinsic motivation is great.  
Any thoughts?

Patrick Owings   

==========================================================================

Deming stamp

"Jack R. Stokvis"

Fri, 5 Jul 96 09:05:25 -0400

Bill:
I heard about the proposed Deming stamp. Can you send info on status, 
timing, addresses, etc.
Thanks,
Jack R. Stokvis

Intrinsic motivation/O'Donoghue

smodonoghue@cbs.com

Fri, 5 Jul 1996 11:53:16 -0400

     The posting by Patrick Owings makes some very important points.
     I believe there is a balance to be maintained within everyone between
     the intrinsic and extrinsic motivation. The extrinsic motivation for 
     me and many of my coworkers is money (we need to make a living) and 
     career advancement ( to assure that we continue to make sufficient 
     amount of money to live and support our families). However if these
     were the only motivations present then why not attempt to achieve 
     these through illegal activities or some other means that simply 
     provides lots of money to support our lifestyle.
     
     The intrinsic motivation which provide joy, and satisfaction in work 
     keep us doing what we like to do. It must be sheer drudgery to have to 
     work at a job (lawyer, doctor) and not find some intrinsic motivation 
     in the tasks performed. I'm sure doctors must derive some great joy 
     and satisfaction at saving lives and healing the sick. Lawyers also 
     must find satisfaction and joy in helping people and businesses 
     resolve issues and conflicts within the legal system.
     
     Regards,
     
     Shane O'Donoghue (smodonoghue@cbs.com)

=========================================================================

DEN RACR Plan (fwd)

clauson@deming.ces.clemson.edu (List Moderator Account)

Fri, 5 Jul 1996 18:08:25 -0400 (EDT)

Linda & Bill,

The following message describes the 'Newt Project'  I thought the Board
would like to see what we can and are doing via the DEN.

This is becoming a more formalized project and I was wondering is a project
sheet needs to be done and forwarded to the board (?)

I just sent a copy to Gipsie, then remembered she is moving and may not
have net access yet.

Jim Clauson
DEN Moderator


Forwarded message:
> From Donald.Kerr@alliedsignal.com Fri Jul  5 17:28:17 1996
> Message-Id: 
> From: "Kerr, Donald" 
> To: John Seddon <101454.573@compuserve.com>,
>         Myron Tribus
> 	 <104055.2663@compuserve.com>,
>         Anton Tolman
> 	 ,
>         Bill Cooper , David Kerridge ,
>         Ian Bradbury
> 	 
> To: Jonathan Siegel ,
>         John Paul Fullerton ,
>         Wayne Levin
> 	 ,
>         Matt Barkley , Patrick Owings ,
>         Rip Stauffer 
> To: Stan Adams , john hunter 
> Cc: "clauson@deming.ces.clemson.edu" 
> Subject: DEN RACR Plan
> Date: Fri, 5 Jul 1996 14:26:00 -0700
> X-Mailer:  Microsoft Exchange Server Internet Mail Connector Version 4.0.838.14
> Encoding: 106 TEXT
> 
> Dear DEN RACRs,
> 
> I wrote up a quick proposed plan (prediction, theory) to get us going
> around the first cycle of the PDSA...check it out and give me any
> comments.
> 
> Thanks,
> Don
> 
> 
> Team: Renewing American Civilization Review (RACR) Project Team
> 
> Sponsor:  W. Edwards Deming Institute- Deming Electronic Network
> Customer: Newt Gingrich and The Progress and Freedom Foundation
> 
> Purpose: To cooperate with Mr. Newt Gingrich and The Progress and
> Freedom Foundation in the advancement of the teaching's of W.Edwards
> Deming toward the renewal of  American Civilization.
> 
> Product: Complete electronic review and report of the 1995 The Renewing
> American Civilization lectures by Newt Gingrich, based on the
> foundational principles of W. Edwards Deming.   The electronic report
> will delivered on 1/97 to PFF at RENEWAM@aol.com.
> 
> Other Potential Services (Voluntary): A follow up oral presentation,
> consultation, and/or  facilitation support for future the "Quality As
> Defined By Deming" classes.
> 
> Team Objectives:
>  - To analyze and synthesize the complete series of 10 lectures found at
> http://www.pff.org to avoid out-of-context misunderstandings
>  - To provide non-partisan feedback to Mr. Gingrich based on the
> foundational principles of W. Edwards Deming.
>  - To encourage and support the continuous improvement of the Renewing
> American Civilization course, especially the "Quality as Defined by
> Deming" class.
>  - To engage in dialogue with Mr. Gingrich with the hope of continuously
> improving his understanding of The Deming Philosophy and The System of
> Profound Knowledge.
>  - To engage in dialogue with Mr. Gingrich to continuously improve our
> understanding of the foundational principles for Renewing American
> Civilization.
>  - To make a collective and constructive contribution to  Renewing
> American
>  - To open up the possibility of future cooperative projects
> (Presentations, Consultations, class facilitation, etc.) between the W.
> Edwards Deming Institute and the Progress and Freedom Foundation.
> 
> The Plan:
> Each DEN-RACR Team member will be responsible to read all 10 lectures
> (minimum of 5) over a six-month period (July-Dec96).  Emphasis will be
> placed on the Five Pillars (lectures 1-5) and their interdependencies.
> Lectures 7-10 will be optional and addressed as time permits.
> 
> Schedule-
> July 96
> Lecture 1-  American Civilization
> Lecture 6- Pillar 5- Quality as Defined By Deming
> August 96
> Lecture 2- Pillar 1-  The Historic Lessons of American History
> September 96
> Lecture 3- Pillar 2-  Personal Strength
> October 96
> Lecture 4- Pillar 3- Entrepreneurial Free Enterprise
> November 96
> Lecture 5- Pillar 4- Spirit of Invention and Discovery
> December 96 (Optional)
> Lecture 7- The Third Wave and American Civilization
> Lecture 8- Creating American Jobs in The World Market
> Lecture 9- Replacing The Culture of Violence and Poverty with a Culture
> of Productivity and Safety
> Lecture 10- Citizenship And Community In  21st Century America
> January 97
> Final Review and Synthesis
> Final Report delivered to Progress and Freedom Foundation (PFF)
> 
> 
> The Five-Step Process:
> 
> 1)  You will read the Pillar-Of-The-Month, according to the above
> schedule. 
> 2) You will provide one individual report per Month to me documenting
> your initial comments, concerns, strengths, and weaknesses for each
> Pillar-Of-The Month.  DUE-on or before the last Friday of each month.
> 3) Any time during the month,as I receive each of your reports,  I will
> post them to the rest of the team for comments,dialogue and team
> learning.
> 4) The last week of each month I will summarize the collected reports
> and comments and post it for your review.  I will also post it to the
> DEN as a team status.
> 5)  I will synthesize each Pillar-Of-The Month summary into a Final
> Report and post it for your final review and comments.  This Report will
> then be E-Mailed to Mr. Gingrich.
> 
> Next Steps:
> 1) Provide feedback to me on this plan
> 2) Read Lectures 1 and 6 and send me your individual report no later
> than July 26...hopefully sooner.
> 
> Have a Great Adventure!
> Don Kerr
> 
> p.s.  Ian Bradbury's address  is the only one that bounces.  Does anyone
> have his address or know what's wrong?
> 
>  
> 


-- 

Deming Questions # 19/Gogue

gogue@chailly.ensmp.fr (GOGUE J.M. Societe MAST 39 50 99 67)

Sat, 6 Jul 96 10:20:43 +0200

--------------------------  DEMING QUESTIONS  -------------------------- 

A weekly paper including a set of questions copied in the Deming book 
OUT OF THE CRISIS   Chapter 5.  Questions to Help Managers.

By Jean-Marie Gogue

-----
# 19
-----

1.   (No 37 in the Book) - Exploring the future business

a)    Who makes the decision about whether to buy your product?

b)    What new design would serve better four years from now?

COMMENTS
These questions are the most important and the most difficult to answer. 
If a company can predict that a new design will result in tremendous 
sales, they are nearly invincible. But neither in 'Out of the Crisis' nor 
in 'The New Economics' Deming provides a solution.

Let's come back to the C. I. Lewis' statement : *Knowing begins and ends 
in experience; but it does not end in the experience in which it begins*.
Knowledge is prediction. When a company has made the Deming transformation 
the capital of knowledge is cultivated with great care, for everybody has 
understood that it's the most important resources. When knowledge improves, 
the ability to predict the future business improves.

-----

2.   (No 38 in the Book) - The inspection status in your company

a)   What inspection or verification are you carrying out 
     on incoming materials?

b)   What inspection or verification are you carrying out 
     in process?

c)   What inspection or verification are you carrying out 
     on final product?

COMMENTS
Inspection is necessary, inspection must be documented by procedures, 
but inspection must not be installed for ever. Remember point 3 : 
*Cease dependence on mass inspection*. 

Some organizations suffocate to death under layers of inspections that 
settled year after year, like geological layers. 
  
----------------------------------------------------------------------------

Jean-Marie Gogue
President
The French Deming Association
Versailles   France
gogue@ensmp.fr

=========================================================================

Quiet on the DEN?/Clauson

"James Clauson [Jim Clauson] Deming Net Moderator"

Fri, 12 Jul 1996 00:34:58 -0400 (EDT)

Fellow DEN'ers,

WHY SO QUIET?

Yes, things have been quiet for the last week.  Special cause: I had 
to make an unplanned road trip to Chicago and was unable to access the 
net from 05JUL-11JUL.

I will clear the message queue tonite.

For those of you who sent me back-channel messages, I will respond to 
those as well.

RE: MESSAGE VOLUME:

Over the last month, the DEN has averaged about 140 messages/month or 
4-5/day.  Unless there is a special cause (like the one noted above) 
you should receive a fairly steady, daily flow of messages.

RE: BOUNCED MESSAGES (or those the net cannot deliver and are returned 
to me):

I will usually not manually unsub anyone, unless messages to them 
bounce chronically.  However, the system here at Clemson *does* have 
an "auto-delete" feature which *can* automatically delete your address 
for mail bounces.

WHY AM I TELLING YOU ALL OF THIS?  

If you see a drop in messages for a while, you may have been 
auto-deleted.  If you stop getting messages, send me an e-mail at 
clauson@deming.eng.clemson.edu and I will look into it for you.

Thanks,

**************> Moderator, Deming Electronic Network <***************
* Jim Clauson, Director              Phone     (W): +1 423 882 4611 *
* Quality Training Programs          Phone/FAX (H): +1 423 717 0250 *
* Center for Quality & Innovation    clauson@deming.ces.clemson.edu *
* Roane State Community College      den_moderator@a1.rscc.cc.tn.us *
* Harriman, TN 37748-5011            clauson_jr@a1.rscc.cc.tn.us    *
******************> Primary Listowner, TQM-L <***********************

=====================================================================

Choosing control charts/Stauffer

Ripstaur@aol.com

Mon, 8 Jul 1996 13:20:02 -0400

I have been following this thread with considerable interest. Especially
interesting to me are the answers from Bill Latzko, and several dealing with
Don Wheeler's advanced book.

As I understand it, p- and np- charts are used for data characterized by the
binomial model. Wheeler and Chambers, in Understanding Statistical Process
Control , point out that there are four conditions which must be met in order
for the count to be characterized by the binomial model:
   "1. The area of opportunity for count Y must consist of n distinct items.
   "2.  Each of the n distinct items must be classified as possessing, or not
possessing, some attribute. This attribute is usually some type of
nonconformance to specifications.
    "3. Let p denote the probability that an item has the attribute being
counted. The value of p must be the same for all n items in any one sample.
While the chart will check if p changes from sample to sample, the value of p
must be constant within each sample.
     "4. The likelihood of an item possessing the attribute will not be
affected by whether or not the preceding item possessed the attribute.
(Nonconforming items do not naturally occur together in clusters, and counts
are independent of each other.)"

       The book goes on to say, "When these four conditions are satisfied by
the count data, it is feasible to use a Binomial Probability Model to compute
limits for the sequence of observed values Y1, Y2, Y3,... . That is, we may
use the known relationship between the Mean of the Binomial Distribution and
the Standard Deviation of the Binomial Distribution in our computations...
       "The reader should notice that the symbol n is used here to denote the
Area of Opportunity for a Binomial Count rather than the subgroup size (the
number of values used to compute an average). The Binomial Count, Y, is an
Individual Value, and there is exactly one count per 'subgroup' ."
       As I understand the way count charts work, the assumption of the
distribution is very important. While Shewhart, Deming and Wheeler have all
shown that the shape of a distribution for an averages chart with three-sigma
limits doesn't usually matter (and process knowledge is of utmost importance
in any case), count chart control limits are based on the assumption that one
has a coherent probability distribution (binomial for p- and np-; poisson for
c- and u-). 
       This seems to be an advantage in some cases, and a disadvantage in
others. In the case of the TNE data, I got the following results (results
rounded to the nearest 1/10 of a bead):

                       Newport Beach              Nashville
Xbar(npbar)             9.5                             9.4
mRbar                    2.2                             3.8
UNPL                    15.2                           19.4
LNPL                      3.7                             0 (-.7)
UCLmR                   7.1                            2.4
UCLp                     17.8                           17.7
LCLp                       1.2                             1.1

      I got slightly different results from Bill Latzko--I'm not sure why
(note that I used Wheeler's convention of UNPL/LNPL for Upper/Lower Natural
Process Limit for the individuals chart). The overall message is the same,
however: in both sets of calculations, the XmR chart shows a difference in
dispersion whereas the np chart shows about the same dispersion. This is
because the np chart is based on the average value, essentially the same in
each case. 
      In this case, of course, the results from the np-chart are probably not
suspect. The Bead Bowl is, certainly, about as close to a binomial model as
we are likely to get in the real world (therein lies its immense power to
convince the skeptic). The question is, what is the source of the difference?
       Shewhart said if an average and range chart with subgroups of size n=4
has at least 25 consecutive subgroups which do not indicate any lack of
control, then the process may be said to display a reasonable degree of
statistical control. Wheeler has an extensive discussion of degrees of
freedom for various charts in his Advanced Topics book (he gives an
approximation for degrees of freedom thus: DoF is approximately equal to
.9k(n-1)). Studying these two references leads me to beleive that we don't
have enough data with 24 to provide "solid" control limits.
         We know, however that we have a repeatable process. We know, by
design, it is highly unlikely to display a lack of control (empirical
evidence bears this out, also).
Given the same paddle, bowl, instructions, etc., two teams should give us
equivalent results. Using these assumptions, I appended the Nashville data to
the Newport Beach data set and came up with the following:

Xbar(npbar)             9.4
mRbar                    2.9
UNPL                    17.2
LNPL                      1.7
UCLmR                   9.5
UCLp                     17.7
LCLp                       1.2

In this case, using 48 data from essentially the same process, the XmR limits
come within half a bead of the np limits. I believe if we had 100 data, we
could probably come within 1/10th of a bead (not that ".4 beads" has any
meaning in the real world, other than a measure of discrimination for
determining on which side of the line an integer will fall).
      The reason we can make these statements, though, is WE HAVE PROCESS
KNOWLEDGE. We know the Red Bead experiment. We would be astonished to find a
lack of control in it. After seeing it over and over, we would be surprized
to see control limits other than 1and 18.
       So, which chart is better? It seems that, given enough data and a
process which fits the criteria for the probability model, either one will
do. However, lacking in-depth process knowledge, I would go with the XmR, as
it seems to better reveal the dispersion in the system. If the process is not
characterized by a binomial or poisson model, the dispersion is coming from
something other than a theoretical curve, and reduction of the dispersion is
important. If the process is characterized by a binomial or poisson model,
reduction of dispersion will only come about through reduction of the process
average.
         I would also like to suggest using variables data vice attribute
data wherever possible. Variables data lend themselves to continual
jmprovement much more than do attribute data. They are more discriminate,
less aggregate and contain more information.
        Finally, I would like to add my voice to the rest of those
recommending Dr. Wheeler's books. He has done a lot of research to further
Dr. Shewhart's work.
        
Rip Stauffer
Naval Leader Training Unit
Ripstaur@aol.com

(P.S. Sorry for the long post.  This issue seemed too complex for a simplier
response than this.)
 
===========================================================================
 

Intrinsic motivation/Tribus

"Myron. Tribus" <104055.2663@CompuServe.COM>

08 Jul 96 10:38:24 EDT

Patrick Owings raises an interesting question regarding internal motivation with
regard to love of music.  Paul Robertson has authored an interesting little
booklet "Music and the Mind" which reviews data on how the human mind responds
to music.  In it he discusses findings about neural activity related to musical
tones,  the ability to discriminate music from noise, the need for rhythmic
beats, etc.  A new organization has been formed,  The Music Research Institute,
which has an e-mail address,  
100534.476@compuserve.com
He says that there isn't much published except on the internet!

Internal motivation is,  indeed,  a complex issue and we should be grateful to
Patrick Owings for raising the right question. 

More and more I am coming to believe that education is less a matter of knowing
answers to stock questions but more the ability to phrase useful questions and
then knowing how to go about finding answers.

Myron Tribus

============================================================================

SoPK as a theory/Stauffer

Ripstaur@aol.com

Sun, 7 Jul 1996 11:26:43 -0400

In a message dated 96-06-18, Ian Griffiths wrote:

<< Dr Deming insisted on the importance of evidence -- without which you
cannot
 test a theory. A theory incapable of refutation has no value. What I think
Dr
 Deming meant by 'not looking for examples' is that a few supporting 'case
 histories' are not a good enough test of a theory. A few such cases provide
 plausibility. That should encourage deeper exploration -- not premature
 relaxation of our efforts to probe deeper. In practice this is really tough,
 given the pressures that managers are under.
  >>

This strikes me as a superb bit of explanation for this point. Imagine what
might have happened at the Hawthorne plant had the researchers relied only on
the first results in testing their theories, and no one bothered to say,
"Hey, let's check out the other side of this thing--lets turn the lights back
down and see if productivity goes back down." Thanks, Ian.

Rip Stauffer
TQL Division, Naval Leader Training Unit
Ripstaur@aol.com

========================================================================

Quality tool useage/Lamb

Robert Lamb

Thu, 11 Jul 1996 16:22:23 +1000 (EST)

I have not been able to find any answers in 
the general literature & hope that some of the assembled expertise might 
be brought to bear.
 
1. Skewed distributions and the Empirical Rule
 
When using an individuals chart with a distribution having a long tail, 
one can come up with a negative lower control limit. For example, one 
might be tracking sales per day, where there is a large proportion of low 
values and a few relatively high values. In this case, is the control 
chart of any use in identifying special causes on the low side? If one 
recorded three consecutive zeroes, would one advise the manager not to 
intervene? How can one use a control chart to detect a meaningful falling 
off of sales? Should one look at e.g. elapsed time between sales?
 
2. Pareto charts and population distributions
 
It is common practice to use Pareto charts to stratify data to determine 
an area in which intervention may be beneficial where variation is all 
common cause. There is no mention of doing a control chart to see whether 
the Pareto item is a special cause relative to the underlying process. On 
the other hand, Deming advises us not to take action [on people] based on 
a ranking within a common cause distribution. What is the fundamental 
difference between the two scenarios? 
 
3. Alternative tools for problem solving
 
Intellectually, I find the fishbone diagram to be the least satisfactory 
part of the quality improvement toolbox. I appreciate the need to focus 
the analysis in the context of a ÒQC StoryÓ or similar methodology. 
However, having got to the point of identifying the problem area by 
statistical means, is there any reason why one should not substitute an 
alternative problem solving framework, for example causal loop diagrams, 
systems dynamics models, or even a Kepner Tregoe Òis / is notÓ problem 
analysis? 
 
Regards

Robert

-- 
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 
Robert Lamb        rlamb@netspace.net.au       Melbourne, Australia 
 
 Life isn't meaningless: it just has a poor signal to noise ratio.

======================================================================

Intrinsic Motivation/Tolman

"Anton O. Tolman, Ph.D."

Tue, 09 Jul 1996 12:06:04 -0700 (MST)

On July 5th, Patrick Owings continued the discussion on intrinsic 
motivation and asked some questions:

> I am aware of 
> Briggs-Meyers and other means of choosing a vocation. Do tests these enable 
> one to select a job that will have inherent intrinsic motivation for the 
> individual or is this an area of psychology requiring more research before we 
> have a useful tool?

Patrick, the Meyers-Briggs is a test based mostly upon Jungian 
personality theory; it is not a test specifically designed to help 
match individuals to work assignments or vocations, although it is 
widely used in industrial circles in applications for which it was 
not originally intended.  (I am a little out on a limb here because I 
am not a specialist in vocational counseling and am not up to date on 
the most recent literature).  There are other tests, such as the 
Strong-Campbell interest inventory which _were_ designed specifically 
to try and match people's reported natural interests with vocational 
possiblities.  There are several other newer tests out which purport 
to achieve the same end.  I can get names of these other tests for 
you if you like.  

I think that these types of tests are useful tools for helping 
someone narrow down their options and examine what they consider to 
be their interests and try to match those interests with available 
options.  However, I also believe that there may be times when the 
person thinks that they would enjoy X, only to find after doing X for 
a while that they really don't like it.  As far as I know, there is 
no test to determine what someone might really like that they are 
unaware of.  However, personality tests like the Rorschach, which 
are again designed for personality assessment not job-matching, 
provide information about personality styles that taken with a 
vocational interest test might prove very useful indeed.  

Intrinsic motivation, to me, comes from a combination of natural, 
perhaps genetic or biological "action tendencies" or preferences (for 
degree of stimulation, amount of people interaction, etc.) which are 
shaped by environment and exposure, mostly by parents' own 
preferences and opportunities provided to the child, and conscious 
decisions.  When we are engaged in tasks which match our preferences, 
but which we also find stimulating, rewarding (by whatever 
definition), and worthwhile, I would say that we experience intrinsic 
motivation.  

Studies on "flow" or maximum work performance indicate 
that we know we are flowing when time seems to go by very quickly and 
we are "lost" in what we are doing.  Studies indicate that the flow 
state is perceived to be highly motivating and that mood is at a peak 
when we are in a flow state.  When we are engaged in non-flow 
activities, we tend to feel frustrated, irritated, or that we are 
performing routine drudgery.  In fact, if you consider the effect of 
teaching line workers how to relate to their work in a new way -- via 
control charts, understanding of variation, etc., you may be creating 
new opportunities for individuals who had looked at their work 
perhaps as drudgery, to become more stimulated, to become more "lost" 
in their work and improve their morale and intrinsic motivation.  (By 
the way, the lowest mood ratings were obtained when people were 
watching television).  

> "Is it possible to find intrinsic motivation in life itself and, 
> therefore, find intrinsic motivation in almost any job we do?"  In other 
> words, Dr. Deming's "joy in work" - is it more from offsetting (or 
> eliminating) the "forces of destruction" in one's life or is it the result of 
> some unexplained natural proclivity toward "a love of music", or law, or 
> medicine, or whatever we love to do?  

I think that there are those individuals who are predisposed or whose 
natural preferences are so broad that they find pleasure in many 
different things and can find intrinsic motivation in almost any task 
they perform.  These individuals, as long as they are able to manage 
or eliminate the irrational thinking and some of the other thought 
processes that interfere with adaptation to life that I've described 
on the DEN before, might be described as finding intrinsic motivation 
in life itself.  For others, who by temperament, genetics, or lack of 
early exposure to novel situations and a wide range of possibilities, 
may have limited preferences, I think the job of becoming 
intrinsically motivated is more difficult.  Nothing is impossible, 
but I believe that people range on a continuum for this element.

Perhaps part of understanding the psychology section of SPK and 
applying it to the system is to reach out and assess where our 
people fall on this and on other continua so that we can better match 
their skills and preferences to their work assignments, maximizing 
their potential and joy in work.  What do others think?
________________________________
Anton O. Tolman, Ph.D.
Director, Psychological Services
Director, Program Services
Wyoming State Hospital
P.O. Box 177
Evanston, WY  82931-0177
Anton@wsh.state.wy.us
(307) 789-3464
---------------    
=========================================================================

Control chart assistance/Tolman

"Anton O. Tolman, Ph.D."

Thu, 11 Jul 1996 08:31:33 -0700 (MST)

I am constructing some control charts to examine patterns in our 
overtime usage during two different phases of how we have done 
scheduling -- centralized scheduling and unit-based scheduling.  We 
have 16 data points (representing a week's worth of data) for 5 units 
across each time period.  This is how I am going about it:

Use subgroups of n=5 (we have 5 treatment units) and look for 
variation across the weeks (since we are interested in the time 
comparison).  The first control chart was on all 32 data points.  It 
showed,remarkably enough, that there were no (other than maybe the 
very first point) out of control points.  

My interpretation of this initial data is that the system appears 
stable regardless of which scheduling pattern we are using.  Would 
that be a reasonable conclusion?  My next step is to create mini-
charts for each time period, mostly to see the size of the sigma 
limits under the two conditions.  Comments?

I am also starting to look at the data itself to look for patterns in 
the run using Wheeler's 8 points on one side of the line suggestion.  
My last question is should I also use some of the more complicated 
run tests (e.g. 2 out of 3 successive points more than 2 sigma units 
from the central line)?  

I would appreciate any advice you could give.  

________________________________
Anton O. Tolman, Ph.D.
Director, Psychological Services
Director, Program Services
Wyoming State Hospital
P.O. Box 177
Evanston, WY  82931-0177
Anton@wsh.state.wy.us
(307) 789-3464
---------------    

========================================================================

Control chart assistance/Rajagopalan

"Rajagopalan R"

Fri, 12 Jul 1996 15:18:10 +0500 EST

I do not understand how you can define n=5 as the sub-group size. 
The overtime usage in different sections may respond differently to 
centalisation/ decentralisation differently.

I have this suggestion:

Why not conduct a standard t-test testing for difference in the mean 
overtime usage in each of the 5 sections- under each system- you have 
16 data points on each section-under the two alternative systems. You 
may in fact find that in some sections the overtime might have gone 
up and in others it might have come down. The same might have 
happened about the variability in usage. 

R.Rajagopalan
raja@fac.irm.ernet.in

============================================================================

Chaos vs. Randomness-Help!/Stauffer

Ripstaur@aol.com

Fri, 12 Jul 1996 07:49:10 -0400

     I am considering writing a paper or article on the relationship between
chaos and randomness, the thesis of which would be something like "chaos" is
to "randomness" as a histogram is to a distribution curve, as the real world
is to the mathematical plane. The abundance of literature is huge, though, so
I may be just repeating someone else's work. Does anyone know of a paper or
article which already does this?
     I have read Dr. Shewhart and some of Dr. Deming's thoughts on the nature
of randomness. I have read several books on chaos, and some papers by Wheeler
and others which touch on the relationship between the two, but I haven't had
a chance to read (and might not understand if I did read) any of Fiegenbaum,
et al--the original technical papers on Chaos Theory.
     So, anyway, I'll defer to those on the net whose experience and wisdom
are so much greater than mine--is this an article which would be useful, or
should I concentrate my efforts elsewhere?
     Your thoughts would be greatly appreciated.

Rip Stauffer
Naval Leader Training Unit
Ripstaur@aol.com

=======================================================================

Tolman's control charts/Stauffer

Ripstaur@aol.com

Fri, 12 Jul 1996 07:49:12 -0400

      I'm not sure I understand your subgroups. You have 5 treatment centers;
are you taking one data point from each treatment center and averaging across
them? If that is what you are doing, one thing you want to check for is lack
of stratification. If a lot of your data (one rule of thumb Wheeler uses--I
think he got it from Nelson--is 15 consecutive points) is within one sigma of
the centerline (especially on the range chart), it is an indication that you
may be assuming a parallel process where there is none. In this case, you
usually want to reconsider your subgrouping scheme. 

      This technique is counterintuitive and most of my students have a hard
time seeing the value in it. They see a chart with lots of data right near
the centerline and well within the control limits and "think" they see an
indication of a very stable system. The "specification limit" mentality is so
engrained they can't see what could be wrong with a process "so far inside
the limits." That paradigm has to shift before they can understand that the
empirical rule cuts both ways--if 60 to 75 percent of the data should be
within one sigma, then 95 percent should NOT be.

       Wheeler and Chambers, in chapter six of Understanding Statistical
Process Control, have an excellent discussion of rational subgrouping. It
shows the same data from the same process organized three different ways. The
new insights available from the different subgroupings are eye-opening. 

       Regarding the other rules, I like the idea of using Wheeler's rules
1-4, the Western Electric zone tests. The more rules you add, the shorter
your Average Run Length (ARL) between false signals will be; but the ARL
drops dramatically as you add rules after Rule 4. In a recent seminar, Don
said to use rule 1 first, of course, rule 4 (8 consecutive on one side of the
centerline) second, then use the other two. It depends on what's economical
for you.

      Hope this helps.


Rip Stauffer
Naval Leader Training Unit
Ripstaur@aol.com

==============================================================================

Re: Quality tool useage/Hacquebord

heeroh@mailbag.com (Heero Hacquebord)

Fri, 12 Jul 1996 12:26:52 -0500

Skew Distributions:

That is the problem with the chart you are using. It is full of holes.  Use
an X-bar  R chart with n=4 and be careful that you select rational
subgroups.

Pareto Diagram:

The pareto diagram analyses the common causes of a stable process to see
which cause occurs the most over time.  This is not ranking and rating as
Deming referred to. Those few causes that occur most often can be further
analysed seperately in terms of control charts, or whatever makes sense in
order to reduce them.  Note that the purpose here is to improve the system.
Improving the system takes more profound knowledge than mere tools

Alternative tools for problem solving:

What I do't like about the cause and effect diagram, (fishbone diagram) is
that it is linear.  Most problems are not linear.  Any tool that works can
be used.  Kepner Tregoe is great for going after special causes, which is
what it was designed for.  I invent most of the tools I use, on the job, as
I face the problem with my clients.  Tools come from theory. It is the
theory that counts.

Regards,

Heero Hacquebord
Consultant in Continual Improvement
heeroh@mailbag.com

=========================================================================

Re: Choosing control charts/Hacquebord

heeroh@mailbag.com (Heero Hacquebord)

Fri, 12 Jul 1996 12:00:55 -0500

 I have not got the time to study in detail all the responses about control
charts, but let me say this to try and help!

The limits that Dr. Deming uses for his red beads are the same as what is
shown in most text books.  Use them and you will not go wrong .  In fact,
in 1980, Dr. Deming was using the limits as applied to a poisson
distribution. I pointed this out to him and suggested he use the one's he
had currently been using.  He has used these limits since 1980 ever since I
pointed this out to him. He gave me recognition for those limits based on
the binomial distribution in his seminars, at least once that I was there.

The attributes data that applies to the binomial distribution has to be
used on a p-chart.
If you have attributes data as you describe, but it does not meet the
binomial distribution, then there are special causes, or you have some
other kind of a distribution, but not a normal distribution.  This is
precisely what the p-chart does for attributes data, to see whether we have
a stable process, which means a binomial distribution.  Without a binomial
distribution for attributes data you can not predict, which is the overall
purpose.

 If you use variables data charts for attribute data you could get false
signals.  Consider a p-chart. The limits are not influenced by the point to
point variation, but by the mean.  Consider an X-MR chart. The limits
depend on point to point variation. Every special cause will be included in
the calculation of the limits, except special causes due to trends.  DO the
following experiment:  Find a process for attributes data that signals
several special causes on a p-chart, i.e. points outside the control
limits.  Use now an X- MR chart for this same data.  You will probably find
no special causes, Why?

Therefore, if you had a process for attribute data that has  special causes
you will find these special causes on a p-chart, but often will miss them
on a chart for variables data.  This is why Shewhart spoke about a
"Rational subgroup". The question is not how the binomial or other
distribution approximates a normal distribution, but what do you see when
there are special causes. Parametric statistical theory assumes a
distribution.  This is often not the world that we are in.

The purpose of the control chart is to see whether the process is stable,
which implies a distribution close enough to the known ones. The question
is not what distribution is approximated by a stable process, but whether
there is a better one than what we know?  If a p-chart shows special causes
we do not have a (binomial) distribution, and therefore we cannot predict.
If an X bar R chart, or an X -MR chart shows special causes we do not have
a (normal) distribution and therefore have no prediction.  If a c-chart
shows special causes we do not have a (poisson) distribution and therefore
cannot predict.

 Don't mix different charts with different types of data or you will get
false signals, and crazy control limits, and don't mix different
distributions with one another just because statisticians say that they
approximate one another, (under certain conditions).  Now it is true that
IF a process is stable (but how would we know) you can mix different charts
and get similar answers. But that is not the pupose of the chart.  The
purpose of the chart is to see WHETHER the process is stable, and for this
we should not mix the different charts, and different distributions.

For attributes data use a p-chart or np chart (same thing, because p =
np/n). For count data, (rare events, where you can count occurrences but
can not count non-occurrences, e.g. accidents, # of lighting strikes etc,)
use c-charts or u-charts (same thing because u=c/n). For variables data use
X-bar R charts, or X- MR charts if you have few data. Where possible use
variables data, i.e. gauges that give readings instead of go-no-go gauges,
because with variables  data we do not require large sub groups). There are
other charts for other distributions, such as exponential, but the charts
mentioned for the data as applied will work for 99% of all cases.

Heero Hacquebord
Consultant in Continual Improvement
heeroh@mailbag.com

===========================================================================

Re: Chaos vs. Randomness/Prevette

Steven_S_Prevette@RL.gov

Fri, 12 Jul 1996 12:00:23 -0700

--IMA.Boundary.631891738
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Description: cc:Mail note part



     >I am considering writing a paper or article on the relationship between
>chaos and randomness, the thesis of which would be something like "chaos" is 
>>> snip <<<     
>Your thoughts would be greatly appreciated.
     
>Rip Stauffer
>Naval Leader Training Unit
>Ripstaur@aol.com


Here are my ideas on appropriate definitions:

        CHAOS = Future events CANNOT be predicted whatsoever on the basis of 
past events.  There are no coherent patterns nor controls.  In Statistical 
Process Control parlance, the process is "out of control".

        RANDOMNESS =  I do not know exactly what the next event will be, but I 
can make a prediction based upon past events.  For example, I know that past 
data fits a normal distribution with an average of 10 and a standard deviation 
of 2.  Therefore, I can make predications about future events, and (by making 
use of the Central Limit Theorm) can be accurate over a large population.

An assumption is that we are talking about a stable ("in control") random 
process.  Further interesting points to discuss would be INDEPENDENCE from event
to event vs. some DEPENDENCE between events.
     
     - Steve Prevette
       steven_s_prevette@rl.gov
       Westinghouse-Hanford Corp.
--IMA.Boundary.631891738
Content-Type: text/plain; charset=US-ASCII; name="RFC822 message headers"
Content-Transfer-Encoding: 7bit
Content-Description: cc:Mail note part

Received: from [130.97.153.199] by ccmail.rl.gov with SMTP
  (IMA Internet Exchange 1.04b) id 1e67d691; Fri, 12 Jul 96 09:29:29 -0700
Received: by fep1.rl.gov (5.51/5.17.rl-1)
	id AA25462; Fri, 12 Jul 96 09:29:43 PDT
Received: from deming.eng.clemson.edu.eng by eng.clemson.edu (SMI-8.6/SMI-4.1)
	id MAA17304; Fri, 12 Jul 1996 12:26:24 -0400
Received: by deming.eng.clemson.edu.eng (4.1/SMI-4.1)
	id AA25743; Fri, 12 Jul 96 12:26:44 EDT
Resent-Date: Fri, 12 Jul 1996 07:49:10 -0400
Old-Return-Path: 
Old-X-Envelope-From: Ripstaur@aol.com  Fri Jul 12 07:47:17 1996
From: Ripstaur@aol.com
Date: Fri, 12 Jul 1996 07:49:10 -0400
Message-Id: <960712074908_154364764@emout15.mail.aol.com>
To: den.list@deming.ces.clemson.edu
Subject: Chaos vs. Randomness-Help!/Stauffer
Approved: den_moderator@deming.ces.clemson.edu
Sender: clauson@deming.ces.clemson.edu
Resent-Message-Id: <"H01jTC.A.CKG._wn5x"@deming>
Resent-From: den.list@deming.ces.clemson.edu
Reply-To: den.list@deming.ces.clemson.edu
X-Mailing-List:  archive/latest/9
X-Loop: den.list@deming.eng.clemson.edu
Precedence: list
Resent-Sender: den.list-request@deming.ces.clemson.edu
--IMA.Boundary.631891738--

=====================================================================

Re: Quality tool useage/Prevette

Steven_S_Prevette@RL.gov

Fri, 12 Jul 1996 12:23:35 -0700

     
>1. Skewed distributions and the Empirical Rule
     
What I am currently doing in my job in similar situations is invoking the 
7 points in a row below average, or 4 of 5 in a row below one standard 
deviation below average (if the average minus 1 standard deviation is > 
0), and 2 of 3 in a row below 2 standard deviations below average (again, 
if average minus 2 standard deviations is greater than 0) rules for 
detecting significant change.  So three zeroes in a row would only be 
significant if the average minus two standard deviations is greater than 
zero.

I know this causes some inaccuracies.  If this is a real fear, you could 
always invoke a more sophisticated run theory (average run length) 
approach.  Or even change your time interval to longer so more sales 
occur per time interval and the lcl will be greater than 0.  But over the 
past 3 years I have found the simpler approach (above) to be sufficient 
in the long run.

     
>2. Pareto charts and population distributions
     
     I would say (based upon how I have been doing this) is the difference 
     in approach is the time interval over which you do a Pareto.  Say that 
     I just had seven months in a row above average and declared a 
     significant increase occurred.  I might do a Pareto over those seven 
     months to learn what special cause has occurred.
     
     However, if the process has been "in control" over a long period of 
     time, I want to try to get some ideas how to improve (ie change the 
     process), I might do a Pareto breakdown over the life of the process 
     (hopefully at least 25 time intervals).  This is less likely to be 
     worthwhile, but I am having good use of it.  For example (this is 
     real), our accident case rates are "in control" and stable.  The 
     company wants to figure out how to improve.  I routinely do long time 
     interval Pareto charts (over years of accidents) showing the most 
     common causes (in our case, Overexertion: Lifting, Pushing, Pulling).  
     Now I know I can tell management - "If you want to reduce accident 
     rates, we need to change the process by which with lift objects".
     
>3. Alternative tools for problem solving
     
     I like flow charting out the process, myself.
     
Steve Prevette
steven_s_prevette@rl.gov
Westinghouse Hanford  - performance measures
    
======================================================================
 
     
     

VARIATION : Pareto charts/Kromkowski

Kromkowski@aol.com

Fri, 12 Jul 1996 17:23:19 -0400

In a message dated 96-07-12 01:26:05 EDT, Lamb writes:

<< 2. Pareto charts and population distributions
  
 It is common practice to use Pareto charts to stratify data to determine 
 an area in which intervention may be beneficial where variation is all 
 common cause. There is no mention of doing a control chart to see whether 
 the Pareto item is a special cause relative to the underlying process. On 
 the other hand, Deming advises us not to take action [on people] based on 
 a ranking within a common cause distribution. What is the fundamental 
 difference between the two scenarios?  >>

In exploring this relationship, I have taken data from a Pareto (Juran) chart
and run a control chart (IndX/moving R). I always rearrange data into
"random" order (either by drawing lots or by alphabetizing). Invariably, I
have found that the "significant few" fall outside limits.  However, two
things have struck me.  First, what is the operational definition for a
Pareto/Juran distribution.   Could in fact, a control chart provide the
means? However, n.b., that the Significant Few plus the trivial many, often
are not large enough to use a C.C. without trepidation.

Second, does it matter whether or not there is an operation definition for a
Pareto/Juran distribution?  Implicit behind Juran's developement of Pareto
method (as I recall) is the underlying assumption that one needs to do
something to begin cycle of improvement.  Second, it provides a method to get
the biggest bang for the quality buck.  Even if the line of defects or causes
in a Pareto chart produce no actual significant few (out of control) and
trivial many, what they then do represent is a system in control (e.g.
consistantly turning out crappy stuff).  Improvement of this system is the
responsible of management.  But by what method does management decide to
attempt to act.  The PSDA gives us a general process but it can not make the
decision for management on what to begin.  I suggest that in this sense the
Pareto Chart is less a stastical process tool than a management
decision-making instrument.  

John D. Kromkowski
Attorney at Law
The Jefferson Building -- Suite 103
105 W. Chesapeake Ave.
Towson, MD  21204

Tel. 410-821-6116
Fax 410-828-0705
kromkowski@aol.com 

=========================================================================

PSYCHOLOGY: Intrinsic Motivation/Kromkowski

Kromkowski@aol.com

Fri, 12 Jul 1996 17:21:31 -0400

Re: discussion of Intrinsic.  Wouldn't Deci shed some light?

===========================================================================

OTHER: Chaos vs. Randomness/Kromkowski

Kromkowski@aol.com

Fri, 12 Jul 1996 17:23:29 -0400

In a message dated 96-07-12 12:27:49 EDT, Stauffer writes:

<< I am considering writing a paper or article on the relationship between
 chaos and randomness, the thesis of which would be something like "chaos" is
 to "randomness" as a histogram is to a distribution curve, as the real world
 is to the mathematical plane. The abundance of literature is huge, though,
so
 I may be just repeating someone else's work.  >>

I'd comment that "chaos" is a pretty loaded term with lots of different
meanings.

For example, it might be used 

to describe "almost-periodic" data;

to describe data that ostensibly meets "tests" of randomness but actual
contains some underlying order, which may be described by non-linear
equations;

to describe a dynamic systems producing ordered (i.e., with order or
predictability) outcomes (e.g., outcomes in statiscal control) which however
may when perturbed even slightly (i.e. by that which ostensibly is
insignificant) may produce unordered (i.e., not predictable) outcomes.   (Or
visa versa.  Or where slight perterbations may alter system from one state of
control into another.);

or to describe data that is self-similar across scales (fractalness).

"Random" also has definitions associated with it coming from a bit longer
tradition of mathematicians and statisticians.

Hence, it light of your paper:  do you intend to decide first what you mean
by randomness and chaos and _then_  explain and test your thesis? Or do you
plan to shape the definition of randomness and chaos to fit your thesis,
regardless of the tradition and history (longer for random, shorter and
less-well defined for chaos)?

I am also not sure that the mathematicians involved in the study of "chaotic"
systems would suggest that randomness is just some ideal contraposed to
chaos.  Of course, I am not one of those mathematicians and I am certainly
not wise.  However, it was my understanding that these fellows are perfectly
willing to say that some sets of data are chaotic and some are random and
some are ordered (neither chaotic nor random, but satisfying some linear
equation) and some are of a kind that it cannot be determined whether or not
they are random or chaotic.

Good luck on your enterprise.

J. D. Kromkowski

===========================================================================

Re: Chaos vs. Randomness/Kaliste

jkaliste@skipjack.bluecrab.org

Fri, 12 Jul 1996 19:10:14 -0400 (EDT)

>     I am considering writing a paper or article on the relationship between
>>> snip <<<
>     Your thoughts would be greatly appreciated.
>
>Rip Stauffer
>Naval Leader Training Unit
>Ripstaur@aol.com
>
Mathimatica, which studies fluid dynamics and other things I do not understand
may point out that just because something is beyond our range of knowledge
at the present it may not be random or chaotic.  Try a search engine.
Wisdom?
                                            Thanks,
                                              John

jkaliste@skipjack.bluecrab.org

========================================================================

FWD: Photo of Dr. D.

Jim Clauson

Sat, 13 Jul 1996 03:12:35 -0400 (EDT)

--Boundary (ID hgXYYFEIskvXkNl+MDuMwA)
Content-type: TEXT/PLAIN; CHARSET=US-ASCII

Tom & Del,

RE: attached note from Bill Ratcliff.

Either of you remember where the Deming GIF came from?

CyberQ-Guy out.

--Boundary (ID hgXYYFEIskvXkNl+MDuMwA)
Content-type: MESSAGE/RFC822

Date: Thu, 11 Jul 1996 12:34:48 EDT
From: Bill Ratcliff 
Subject: Photo of Dr. D.
To: Jim Clauson 
Message-id: <31E52D28.1C4C@radix.net>
MIME-version: 1.0
Content-type: TEXT/PLAIN; CHARSET=US-ASCII
Delivery-date: Thu, 11 Jul 1996 09:33:00 EDT
Posting-date: Thu, 11 Jul 1996 08:33:41 EDT
Importance: normal
A1-type: MAIL

Jim

Do you know the origin of the photo that is the downloadable file on DEN? 
Variuos people have been looking for photos which have owners who will 
give formal permission to use in publicatios.

I see the Web page is up and looking good.

Regards Bill

--Boundary (ID hgXYYFEIskvXkNl+MDuMwA)--

Chaos and randomness/Tribus

"Myron. Tribus" <104055.2663@CompuServe.COM>

Tue, 16 Jul 96 00:09:17 EDT

Don Wheeler has written a nice little paper on the relation  between chaos and
probability.

The concept of randomness is treated in my book,  "Rational Descriptions,
Decisions and Designs" (Pergamon,  1969).  The treatment is not necessarily
consistent with what you will read elsewhere,  but is consistent with Don
wheeler's views,  as far as I know.

My working definition of randomness is:  A table of numbers is RANDOM if,
knowing the first N entries in the table,  there is no way to predict the value
of the (N+1) entry.  A number taken from such a table is called a RANDOM NUMBER.


Generally a table of random numbers is generated to be uniformly distributed in
a specified interval,  say 0 to 1.   Then,  after the person generating the
table is convinced (??) that the distribution is uniform,  a transformation may
be applied to make the distribution "normal" or some other distribution.  

Numbers generated via chaos theory do not fit this definition.   However,  if
you do not know the source of the numbers,  you might well conclude that the
numbers are random. 

Your thesis sounds like a lot of fun.  Enjoy it.  Good luck.

Myron Tribus

========================================================================

Report #1 from Jerusalem/Tribus

"Myron. Tribus" <104055.2663@CompuServe.COM>

Mon, 15 Jul 96 13:22:35 EDT

This note is written from Jerusalem where I am in the middle of a two week
exposure to the teachings of Dr. Reuven Feuerstein.  His theory of cognitive
modifiability,  that is,  that the cognitive structures of human beings can be
modified,  regardless of their condition,  has strong implications for the
quality movement.   He has made applications to children with Down Syndrome,  to
the aged,  to people who have had brain damage and in all these cases his
successes have attracted world wide attention.  There are some,  who like
myself,  who ask,  "If these methods work with those who have learning
disabilities,  how will they work with the so-called 'normal' people." The
results here, too, have been impressive and that is why I came.

For most of us,  not trained in psychology,  the language is a barrier.  In this
progress report I shall do my best to translate and I apologize in advance to
those who know much more than I do and do not require any translation.  Indeed,
they may wish to rush forward to correct my imprecision and for that I shall be
grateful.

To cause someone to change their cognitive structure means to change their
paradigm -- to change the way they understand the world and their place in it.
It means to change the way they think and,  in particular,  how they approach
the task of adapting to new situations.  Cognitive structures govern the way we
approach problem solving.

I am learning that there are many things in Feuerstein's theories which are the
same as Deming's approaches.   One aspect is beautifully illustrated in the book
by Feuerstein,  Rand and Rynders, Plenum Press, 1988,  ISBN 0-306-42964-0,
"Don't Accept Me as I Am: Helping 'Retarded' People to Excel".  In this book the
authors describe their successes with children and adults with severe handicaps,
such as Down syndrome,  brain damage caused by accidents,  etc.  The book does
not discuss the successes which are now being achieved in improving the
performance of so-called 'normal' children,  the issue which brought me to
Israel to study in the first place, but for those interested in education in
general,  there will be many hints on what to do and what not to do even with
adults and so-called 'normal' students.

The book discusses the belief system required for success and I was struck by
how much it applies to quality management.  Here are the statements lifted from
the first chapter of the book.   Each statement requires a few paragraphs of
explanation,  but those familiar with quality management will quickly fill in
the missing material.  In parenthesis I put the equivalent statement for
quality)

Belief #1: Human Beings are Modifiable 

(Organizations can be changed)
If a manager does not believe this,  then the next step will be to marshall the
information on which he or she may change their mind.

Belief #2: The Individual I am Educating is Modifiable 

(My organization,  this one,  can be changed.)
If your client says,  "Oh yes,  but those organizations were different" you have
to show him or her that they were equally difficult at first.  This is where
networking with other CEO's is important.  It is a kind of benchmarking not
normally recognized.

Belief #3: I am capable of modifying the individual 

(I, the manager, can, or will acquire the ability to, change my organization)
If the client thinks of himself as uneducable,  then you really have a problem.
This may require you to ask a number of searching questions to uncover why.  In
fact,  such a client is a ripe candidate for Feuerstein's method of Mediated
Learning.  More on that in another report.

Belief #4: I, Myself,  am a Person Who May - And Has To- Be Modified.

(If I undertake to change my organization,  I, too,  will be modified.  That is
a good outcome,  for I probably need it.)
This is a delicate issue to raise.   It is essential that the manager accept
that he will have to change.  Later we can discuss how to bring about changes in
managers.  What is required is the belief that changes in the person will come
and they will be beneficial.  

Belief #5: Society - And Public Opinion - Are Modifiable and Have to Be
Modified.  (The enterprises with which I do business and the business community
can be and have to be changed to make quality the way we do busin ess.)
The real leaders in our enterprises,  such as Ford,  Xerox,  Procter and Gamble
have, in fact,  formed a consortium and put up real money ($9 million to be
precise) just to change the scene.  You can refer to this action as evidence
that others believe so.

Managers who do not believe these things,   will not succeed,  for they will not
attempt the required changes.  However,  you should not accept them the way they
are,  but you should try to modify them.   Read the Feuerstein book to see how.

This set of beliefs,  usually not articulated clearly, is why  meetings about
quality seem to take on a religious tone.  Alas,  this is also why quality
meetings often generate zealots.

I commend this book to anyone wanting to get a start on understanding human
development.  

When I learn more,  I shall write more.  MYRON TRIBUS

=====================================================================

Deming in Education/Butler

raymond butler

Sun, 14 Jul 1996 10:39:54 -0500 (EST)

Thanks.  Before I leave, 
there is one question that I would like to pose: as new technology is 
having a tremendous impact on the various ways education and educational 
concepts are delivered to today's students, how can the principles of 
Deming be applied to strengthen academic of success to insure a quality 
program of deliverance?

Thanks, for the support, the information, and the guidance through the 
Deming principles of quality application.  I am learning a great deal 
more than I ever thought that I am ready to learn.

Again...Thanks!

 Raymond Butler                          EMAIL: rbutler@indiana.edu 
 ---------------------------------------------------------------------------
 URL: http://copper.ucs.indiana.edu/~rbutler

=======================================================================

Re: Randomness & Control Charts/Bradbury

LNUSPTH1.RZYJ8C@gmeds.com

Tue, 16 Jul 96 23:16:30 EDT

I have some difficulty with Myron's definition of randomness:

>My working definition of randomness is:  A table of numbers is RANDOM if,
>knowing the first N entries in the table,  there is no way to predict the value
>of the (N+1) entry.  A number taken from such a table is called a RANDOM 
NUMBER.

According to this definition, numbers generated by any pseudo random number 
generator, as used in all
computer random number generators, would not satisfy the definition. Many (not 
all for sure) of these are well
regarded with respect to their behavior.

Also, a highly autocorrelated time series would meet the definition of random 
proposed here, but would
not possess the properties of independence one would typically like to be 
exhibited.

We never know in practice whether the process that lies behind the generation 
of a series of numbers is
random. Randomness is a concept. We may agree to criteria which may be applied 
in practice to make
a judgment, and we may assess the results of repeated application of such 
criteria in use. In doing so we
provide operational meaning to the use of the term [random]. Dr. Deming used to 
say that a table of
random numbers becomes useful when we know what is wrong with it.

The criteria which we agree to use for judgment of whether or not the concept 
of randomness is applicable
in a given instance do in fact have their basis in the mathematical theory 
containing the concept of
randomness. This does not, however, mean that such theory is subsequently 
assumed in the application
of the agreed upon criteria. The criteria are what provide operational meaning 
to the concept, once agreed
upon. Judgment of the appropriateness or usefulness of the criteria will 
subsequently not be relative to
the theory from which they were derived, but their effectiveness in use.

This brings me to a point of disagreement with some of the control chart 
discussions that have been
occuring recently. How can it be argued that, on the one hand, the Normal 
distribution is not a required
assumption for use of X-bar and R charts or I and MR charts but, on the other, 
that the Binomial
distribution is a necessary assumption behind use of p or np charts (or 
equivalently Poisson behind
c or u charts). I believe this is clearly logically inconsistent. It would 
appear that I am in disagreement
with Don Wheeler on this point, so I'd better put forward my argument:

The Normal distribution was used by Shewhart in his derivation of formulae to 
compute control limits
for X-bar and R charts. One does not need to assume that data are Normally 
distributed in order to use
X-bar and R charts. If one needed to make such an assumption one would already 
have made the
judgment, prior to use of the chart, that a particularly restricted form of 
state of statistical control exists.
Since one of the uses intended by Shewhart for the control chart was that of 
judgment of whether a
state of statistical control exists, it clearly does not depend upon the 
assumption of statistical control!
This is the sense in which the control chart provides an operational definition 
for the concept of
statistical control, and thus puts communicable meaning into the concept. 
Judgment of the usefulness
of the control chart is not relative to Normal theory, but rather to whether or 
not the control chart is found
in use to provide signals that are a reliable basis of learning for improvement.

In the same vein, the Binomial distribution provided a basis for deriving the 
control limits for p and np
charts. That does not mean that one needs to meet the conditions that provide 
the basis for the Biinomial
distibution in order that one may use a p or np chart. One should ensure that 
the measurements are
of agreed upon form - groupings of events which may be classified in one of two 
consistent ways.
Beyond this however, the p or np chart may be considered as an operational 
definition of whether or
not a state of statistical control exists in the process which generates such 
data. Let the control chart
provide signals of causes which might be manifested as non-constant p, lack of 
independence
and so on.

I would like to illustrate this through the following example in which I was 
involved. Some data had
been collected by a casting facility on the % of their transmission cases that 
were being rejected
by their customer on a daily basis. The data were plotted on a p-chart which 
showed a lot of out of
control signals - the control limits were probably only capturing about 25% of 
the overall process
variation. Rather than immediately concluding that this was a problem with the 
choice of control
chart, the potential causes were investigated through thinking about how one 
might see far more
variation in results than would be expected for Binomial type data. When the 
inspection method
was investigated at the transmission plant, it was found that it was a 
sample-lot inspection scheme.
The top 4 transmission cases were being inspected in each batch, and if 1 or 
more defective
case was found, the whole batch was rejected. This introduces an additional 
source of variation
into the data beyond what one would have observed had the data been generated 
from inspecting
every casting. One could look at it as introducing a lack of independence 
within each batch between
the results of inspecting the first 4 cases and the results for the remainder 
of the batch. It seems to me
that the Binomial assumption would not have been justified. This judgment is 
based on knowledge
gained through investigating the potential causes of the signals on the chart, 
so in this case I would
have been faced with the perplexing problem of needing to have knowledge for 
use of a method
without whose use I would not [necessarily] generated the necessary knowledge.

The next question was what next? One option would have been to revise the 
method of inspection
or the method of recording results to remove the cause of variation identified. 
A p-chart might then
be used to test the theory that this was the cause, and provide a basis for 
further learning for
improvement. If the method of inspection were to remain, an alternative here 
would be to consider
use of an individuals and moving range chart. The reason for such consideration 
is the deliberate
intent to incorporate the identified cause of variation into the calculation of 
control limits. This is a
rational basis for changing the type of control chart used. Whether or not it 
turns out to be useful
will depend on whether or not one is able to use it to learn and improve. It 
would appear to me that
changing the control chart one uses on such a basis is a very different thing 
to changing based
upon getting "too many" or "too few" signals alone.

I shall be very interested to see what you all think of such thinking.

Cheers, Ian Bradbury
LNUSPTH1.RZYJ8C@GMEDS.COM

===========================================================================

Re: Randomness & Control Charts/Kimbler

"D. L. Kimbler"

Wed, 17 Jul 1996 08:36:55 -0400

At 11:16 PM 7/16/96 EDT, you wrote:
>I have some difficulty with Myron's definition of randomness:
>
>>My working definition of randomness is:  A table of numbers is RANDOM if,
>>knowing the first N entries in the table,  there is no way to predict the
value
>>of the (N+1) entry.  A number taken from such a table is called a RANDOM 
>NUMBER.
>
>According to this definition, numbers generated by any pseudo random number 
>generator, as used in all
>computer random number generators, would not satisfy the definition. Many (not 
>all for sure) of these are well
>regarded with respect to their behavior.
>
If we add the words "without knowledge of the mathematical rule governing
the table" at the end of the first sentence, then we have a definition of
randomness that should satisfy you and Myron.  Pseudo-random number
generators have the appearance of randomness because they use a rule that we
generally don't know, and generate numbers that appear statistically random.
>Also, a highly autocorrelated time series would meet the definition of random 
>proposed here, but would
>not possess the properties of independence one would typically like to be 
>exhibited.
>

Random does not necessarily imply independence.  The rule that governs the
sequence may include autocorrelation.  That's how we could use a random
number generator to simulate an autocorrelated time series.  Important point
here:  randomness is not completely without pattern; randomness is a pattern
that obeys statistical and mathematical rules, but the next number in the
sequence cannot be predicted _with_certainty_ without knowing in detail the
rule that governs the sequence.

>We never know in practice whether the process that lies behind the generation 
>of a series of numbers is
>random. Randomness is a concept. We may agree to criteria which may be applied 
>in practice to make
>a judgment, and we may assess the results of repeated application of such 
>criteria in use. In doing so we
>provide operational meaning to the use of the term [random]. Dr. Deming
used to 
>say that a table of
>random numbers becomes useful when we know what is wrong with it.
>
>The criteria which we agree to use for judgment of whether or not the concept 
>of randomness is applicable
>in a given instance do in fact have their basis in the mathematical theory 
>containing the concept of
>randomness. This does not, however, mean that such theory is subsequently 
>assumed in the application
>of the agreed upon criteria. The criteria are what provide operational meaning 
>to the concept, once agreed
>upon. Judgment of the appropriateness or usefulness of the criteria will 
>subsequently not be relative to
>the theory from which they were derived, but their effectiveness in use.
>

You have to have both.  If you divorce the theory from which they were
derived (the "mathematical rule") from the effectiveness of use, you have no
basis for objective judgment; effectiveness becomes a matter of opinion.
Understanding the pattern of randomness (its probability distribution, other
mathematical rules that govern it) helps us analyze its effect.

>This brings me to a point of disagreement with some of the control chart 
>discussions that have been
>occuring recently. How can it be argued that, on the one hand, the Normal 
>distribution is not a required
>assumption for use of X-bar and R charts or I and MR charts but, on the other, 
>that the Binomial
>distribution is a necessary assumption behind use of p or np charts (or 
>equivalently Poisson behind
>c or u charts). I believe this is clearly logically inconsistent. It would

X-bar and R charts take advantage of the central limit theorem.  In effect,
the individual data points are not required to be normal because we use them
in subgroups in a way that allows us to use the normal distribution to
compute limits.  I and MR charts are much more sensitive to the distribution
of the data; if it departs substantially from normal, I and MR charts may
give surprising results leading to erroneous conclusions.  This is because
the central limit theorem does not apply, since we are not subgrouping.
 
>appear that I am in disagreement
>with Don Wheeler on this point, so I'd better put forward my argument:
>
>The Normal distribution was used by Shewhart in his derivation of formulae to 
>compute control limits
>for X-bar and R charts. One does not need to assume that data are Normally 
>distributed in order to use
>X-bar and R charts. If one needed to make such an assumption one would already 
>have made the
>judgment, prior to use of the chart, that a particularly restricted form of 
>state of statistical control exists.
>Since one of the uses intended by Shewhart for the control chart was that of 
>judgment of whether a
>state of statistical control exists, it clearly does not depend upon the 
>assumption of statistical control!
>This is the sense in which the control chart provides an operational
definition 
>for the concept of
>statistical control, and thus puts communicable meaning into the concept. 
>Judgment of the usefulness
>of the control chart is not relative to Normal theory, but rather to
whether or 
>not the control chart is found
>in use to provide signals that are a reliable basis of learning for
improvement.
>

Without correct use of the theory governing the control limits and their
associated probabilities, how can you judge effectiveness?  There are
probably ways, based on long-term observation of the process, but they are
not as efficient as control charts, nor are they as reliable.

>In the same vein, the Binomial distribution provided a basis for deriving the 
>control limits for p and np
>charts. That does not mean that one needs to meet the conditions that provide 
>the basis for the Biinomial
>distibution in order that one may use a p or np chart. One should ensure that 
>the measurements are
>of agreed upon form - groupings of events which may be classified in one of
two 
>consistent ways.
>Beyond this however, the p or np chart may be considered as an operational 
>definition of whether or
>not a state of statistical control exists in the process which generates such 
>data. Let the control chart
>provide signals of causes which might be manifested as non-constant p, lack of 
>independence
>and so on.
>
>I would like to illustrate this through the following example in which I was 
>involved. Some data had
>been collected by a casting facility on the % of their transmission cases that 
>were being rejected
>by their customer on a daily basis. The data were plotted on a p-chart which 
>showed a lot of out of
>control signals - the control limits were probably only capturing about 25% of 
>the overall process
>variation. Rather than immediately concluding that this was a problem with the 
>choice of control
>chart, the potential causes were investigated through thinking about how one 
>might see far more
>variation in results than would be expected for Binomial type data. When the 
>inspection method
>was investigated at the transmission plant, it was found that it was a 
>sample-lot inspection scheme.
>The top 4 transmission cases were being inspected in each batch, and if 1 or 
>more defective
>case was found, the whole batch was rejected. This introduces an additional 
>source of variation
>into the data beyond what one would have observed had the data been generated 
>from inspecting
>every casting. One could look at it as introducing a lack of independence 
>within each batch between
>the results of inspecting the first 4 cases and the results for the remainder 
>of the batch. It seems to me
>that the Binomial assumption would not have been justified. This judgment is 
>based on knowledge
>gained through investigating the potential causes of the signals on the chart, 
>so in this case I would
>have been faced with the perplexing problem of needing to have knowledge for 
>use of a method
>without whose use I would not [necessarily] generated the necessary knowledge.
>

The values of p from lot to log can vary widely.  Major point is that you
have a p value that is affected by the inspection process.  You are right,
in that you need the knowledge of  how the inspection is done.  However, the
binomial assumption is still justified.  An interesting exercise would be to
compute the value of p for accepted lots and p for the overall process.  You
don't say whether rejected lots are screened and defectives replaced.  This
would further complicate the process, and lead to large differences in p
from lot to lot.  Should you have knowledge of the process that feeds your
control chart?  Definitely yes.  Control chart decisions are not made in a
statistical vacuum, they are made with process knowledge, aided by
statistics.  But this process is still binomial, as you describe it.

>The next question was what next? One option would have been to revise the 
>method of inspection
>or the method of recording results to remove the cause of variation
identified. 
>A p-chart might then
>be used to test the theory that this was the cause, and provide a basis for 
>further learning for
>improvement. If the method of inspection were to remain, an alternative here 
>would be to consider
>use of an individuals and moving range chart. The reason for such
consideration 
>is the deliberate
>intent to incorporate the identified cause of variation into the
calculation of 
>control limits. This is a
>rational basis for changing the type of control chart used. Whether or not it 
>turns out to be useful
>will depend on whether or not one is able to use it to learn and improve. It 
>would appear to me that
>changing the control chart one uses on such a basis is a very different thing 
>to changing based
>upon getting "too many" or "too few" signals alone.
>

Changing control charts to find the one most appropriate for both learning
and correctly matching the underlying process is important, and right, it
should not be done simply because of the number of signals.  Be careful,
though, of replacing a p or c with I and MR.  You will then have the
problems associated with departure from normality to worry about.  Shewhart
came up with workable rules based on 3-sigma limits, based on normal theory,
that led to predictable errors.  When we throw out the theory, we lose the
predictability.

>I shall be very interested to see what you all think of such thinking.
>
>Cheers, Ian Bradbury
>LNUSPTH1.RZYJ8C@GMEDS.COM
>
D. L. Kimbler, Ph.D., P.E.
Chair, Dept of Industrial Engineering
Clemson University
kimbler@ces.clemson.edu

===========================================================================

Control Charts/Siegel

Jonathan Siegel

Wed, 17 Jul 1996 13:19:39 -0400

In the control charting question Dr. Tolman presented, there was 
discussion about collecting data from five different sources (one point 
for each source) on a single control chart.

I would think carefully about this approach. A key purpose of a 
control chart is to reduce differences among the products a process 
produces. Reducing unnecessary differences helps us. Reducing 
necessary differences may hurt. 

Before we use a single control chart on what may or may not be different 
processes, we need to ask whether it makes management sense to treat the 
processes as if they were the same. To answer this question, I might 
suggest questions like: Do the five centers serve the same types of 
clients? Do they perform the same kinds of therapeutic activities? Do 
they have the same treatment goals? Are they trying to accomplish the 
same results? Not being a psychologist, I really don't know exactly what 
questions to ask, but the basic thrust is to ask if the five different 
areas really should be treated as the same for management purposes, 
and if it makes sense to expect them to behave the same way, given the 
organization's overall goals.

If the five sources serve different purposes, I would suggest different 
control charts on each at least to start, understanding that this will 
require more time and patience. 

The reason is that if the areas serve different purposes, differences 
among them may be useful, even necessary, to the organization's 
well-being. We may find later on that we want to increase the 
differences to better each area's contribution to the whole. We may find 
the opposite: differences in the charactistics we want to measure may be 
completely irrelevant to the five sources' differences in purpose and 
function and be nothing more than a nuisance. But the fact of the matter 
is, we don't know yet. By thinking of all five as one process from the 
start, we not only predecide the necessary inquiry, but, by acting on 
the chart and eliminating what may be legitimate and necessary 
differences between different processes as "special causes," we may be 
tampering with the system and making things worse.

If the five all serve the same purpose and are functionally the same, 
and hence it makes management sense to want to try to make them as 
similar as possible, I would then go ahead and attempt a single control 
chart. Of course, the five may prove to be "out of control" when thought 
of as a single process, but by making the management decision that all 
parts should be regarded similarly, we will have legitimate reason to 
treat such a result as a problem to be solved, rather than as a 
potential source of organizational strength.

The questions to ask in the inquiry are generally qualitative ones, not 
quantitative ones. It requires organizational and subject-matter 
expertise. A statisician can help here only by suggesting possible lines 
of inquiry.

Jonathan Siegel
jmsiegel.info.research@worldnet.att.net

=======================================================================

Re: Randomness & Control Charts/Siegel

Jonathan Siegel

Wed, 17 Jul 1996 12:35:26 -0400

The reason why a X-R chart works well with nonnormal (but reasonably 
symmetric) distributions but a p-chart doesn't is that the X-R model is 
more robust (less sensitive) to deviations from normality than the p 
model is to deviations from the binomial. This is partly because the X-R 
chart is working with a mean (which tends to smooth out individual 
nonnormal observations and lessen the effect of outliers) while the the 
p-chart relies on individual observations and gives outliers their full 
effect.

I remember Dr. Deming saying that there is no logical connection between 
the concept of randomness and the theory of probablity -- it just seems 
to happen to work out that way! 

Jonathan Siegel
jmsiegel.info.research@worldnet.att.net

========================================================================

Re: Randomness & Control Charts/Prevette

Steven_S_Prevette@RL.gov

Wed, 17 Jul 1996 06:43:30 -0700

     
     From Ian Bradbury:
     
>The Normal distribution was used by Shewhart in his derivation of formulae to 
>compute control limits
>for X-bar and R charts. One does not need to assume that data are Normally 
>distributed in order to use
>X-bar and R charts.
     
     In the performance measurement work I have been doing, I would agree 
     with Ian's writeup.  I find it worthwhile to think of a control chart 
     as a test of the hypothesis that the process is creating stable, 
     random data that is predictable and "in control".  A failure of the 
     control chart (points out of control, detectable trends) rejects the 
     hypothesis that the process is "in control".
     
     One concept worth discussing is "robustness".  Several statistical 
     tools (not only xbar-R control charts, but also Analysis of Variance 
     and Linear Regression) were based upon an assumption of Normal 
     distribution.  However, many of these tools are "robust" in that they 
     also work well for non-Normal data distributions.
     
     A specific case where I have found a c-chart worth while is in 
     graphing the number of "occurrence reports" at Hanford.  There is a 
     specific criteria by which a formal occurrence report must be 
     submitted following an incident or accident.  Originally, I graphed 
     these using x-charts.  However, I kept noticing that the calculated 
     standard deviations were very close to the square root of the average, 
     suggesting empirically that a c-chart might be appropriate.  I now use 
     c-charts to plot the number of occurrence reports per time interval.  
     It (in retrospect) makes sense that occurrence reports could follow 
     the independent, random arrival process (Poisson).  
     
     The discovery that the Poisson and c-chart fits actually helps to 
     understand the occurrence reporting process and how to work to 
     decrease the number of accidents and incidents.  I now can tell if 
     tampering were to occur in the process of writing occurrence reports 
     (such as "if you write another occurrence report, you're fired").  
     Thankfully, no such tampering is evident in the c-chart data.
     
     - Steve Prevette
       steven_s_prevette@rl.gov

=======================================================================

Need icebreaker: subject intro to variation/Adams

SAdamsTimD@aol.com

Wed, 17 Jul 1996 15:52:35 -0400

We (Jim McKinley, Carole and I) are starting a new public offering (free)
thru the chamber on SoPK.  We know that there is a very heavy 7 habits
influence among many who may attend so we have elected to begin with a focus
on variation by using the Deming library tapes with a facilitated discussion
afterwards.   
Does anyone have any constructive comments on the approach?
Does anyone have a simple icebreaker for this first meeting? 
ptlstan 

============================================================================

Re: Randomness & Control Charts/Tribus

"Myron. Tribus" <104055.2663@CompuServe.COM>

17 Jul 96 14:22:16 EDT

Points well taken.  I was briefer in my remarks than I should have been.

I cannot cite the source now (here in Jerusalem) so I am relying on memory.
Once upon a time a set of random numbers was generated in the following way.  A
spinning disk was used with numbers from 00 to 99 marked on its circumference.
This disk was photographed through the use of strobe light.  The strobe light
was triggered by the passage of a cosmic ray through an ionization chamber.
Since cosmic rays arrive in a random way,  so it is presumed,  so ought to be
the numbers thus generated.  

The professor in whose laboratories the apparatus was located went on a vacation
and his assistant produced another set of random numbers.  On inspection of the
table of numbers thus produced it was decided that they were not random!
Further investigation that the professor had actually censored the tables before
they were released but the assistant had not.   The reason for the censoring is
that in a long list of numbers it is certain that there will occur some kind of
recognizable sequence,  i.e.,  a string of even numbers,  a string of numbers in
ascending or descending order,  as sequence of prime numbers,  etc.   Indeed,
it is not difficulty to imagine an unending sequence of numbers which might be
generated by some scheme which sequence could be found.  

So,  based on this story,  it is not far fetched to speak of finding an error in
a table of random numbers.

All this is to say that Bradbury is right on the mark in his comments about
randomness.  

CAn we get Don Wheeler to chime in?
MYRON TRIBUS

[Moderator's note:  Don has not yet subbed to the DEN]
=======================================================================

Re: Randomness & Control Charts/Hacquebord

heeroh@mailbag.com (Heero Hacquebord)

Wed, 17 Jul 1996 23:00:18 -0500

Thank you Ian for your contribution.  I agree with you.  It is sometimes so
difficult to understand that a distribution only exists when the process is
in a state od statistical control, and therefore we need to use
distributional theory as the basis, or operational definition, as to
whether a state of statistical control exists.  Using the normal model,
i.e. X- MR chart for attribute data violates this basis, and will tend to
hide special causes should they exist.

There are so many books on SPC that teach the wrong things.  Dr. Deming was
concerned about this ten years ago, but had given up arguing with people.

Heero Hacquebord
heeroh@mailbag.com

========================================================================

Control charts - data independence/Burns

"A.D.Burns"

Fri, 19 Jul 1996 07:17:31 -0700

One of the basic assumptions in all types of control charts is the 
independence of each data point. If this were not true, randomness could 
not be assumed. The difficulty is that data independence does not happen 
in the real world. Particularly with attribute charts, with low p, 
defects will tend to occur in groups. The process will have a 
predisposition to defects when a special cause has arisen or is evolving. 
I'm sure most people have experienced how everything seems to go wrong at 
once.
 
The closer points are together in time, the less independence there will 
be. This is a problem for subgroups in X bar R charts.

Can anyone comment on how lack of independence of data is handled?

Dr Tony Burns
mmedia@ozemail.com.au

============================================================================

MBTI Source/Walker

Walker Associates

Wed, 17 Jul 1996 17:04:04 +0000

On 5th July, Patrick Owings asks if Myers-Briggs may be used to select 
a vocation with intrinsic motivation.
A data base exists showing the occupations actually chosen by 
different Myers-Briggs personality types.   Career counsellors can 
use this to provide advice on careers empirically attractive to 
different personality types.
This is covered in the manual "A Guide to the Development and Use of 
the Myers-Briggs Type Indicator", ISBN 0-89106-027-8.

Regards, Peter Walker.
-------------------------------
Walker Associates
Tel:+44 1908 510593 Fax:511772

==========================================================================

Random number generator?/Jordan

LNUSPTH1.WZ33XG@gmeds.com

Thu, 18 Jul 96 09:36:39 EDT

A point to consider about Dr. Tribus example.

>A spinning disk was used with numbers from 00 to 99 marked on its 
circumference.
>This disk was photographed through the use of strobe light.  The strobe light
>was triggered by the passage of a cosmic ray through an ionization chamber.
>Since cosmic rays arrive in a random way,  so it is presumed,  so ought to be
>the numbers thus generated.

This mechanical method may not generate random numbers! If the rate of
 turning of the wheel is not uniform some numbers will spend longer 
under the strobe than others. The random input of light cannot correct
for the unequal opportunity of selection. This will not yield equal coverage.
This is much like the paddle that prefers one color of bead over
another.  A simple random sample should have EQUAL and COMPLETE coverage.
Equal all possible outcomes have the same probability of occurrence.
Complete all outcomes of interest are possible.

Jack Jordan
LNUSPTH1.WZ33XG@GMEDS.COM

======================================================================

Which SPC books recommended?/Kerr

"Kerr, Donald"

Thu, 18 Jul 1996 05:54:00 -0700

[Next of thread]             
[Next by author]             
[First of thread]            
[First by author]             

Re: Randomness & Control Charts/Hacquebord
heeroh@mailbag.com (Heero Hacquebord)
Thu, 18 Jul 1996 12:02:49 -0500
[Next of thread]             
[Next by author]             
[First of thread]            
[First by author]              
Jonathan Siegel Wrote:

>The reason why a X-R chart works well with nonnormal (but reasonably
>symmetric) distributions but a p-chart doesn't is that the X-R model is
>more robust (less sensitive) to deviations from normality than the p
>model is to deviations from the binomial. This is partly because the X-R
>chart is working with a mean (which tends to smooth out individual
>nonnormal observations and lessen the effect of outliers)

My Response:

Yes, the distribution of averages of subgroups larger than four tend
towards normality.  This was one of the great contributions that Shewhart
made.  But note that here we are dealing with variables data.

Jonathan Siegel wrote further: > while the the p-chart relies on individual
observations and gives outliers their full effect.

My Response:

 I disagree that the p-chart gives "outliers" their full effect.   The
limits for the p-chart depend on n and p-bar.  They have nothing to do with
"outliers", only in that "outliers" will affect p-bar.  Which is less than
what they would affect the control limits of an X-MR chart, for instance.
Therefore an X-MR chart will gives "outliers" a fuller effect than a
p-chart. So the p-chart is less sensitive to "outliers" than an X-MR chart.
Therefore the danger in using charts for variables data instead of a
p-chart, when we have attributes data. The p-chart does depend on
normality, which is actually the binomial approximation of the normal
distribution. Therefore when using a p-chart, np-bar must be greater than 5
for p-bar less than 0.5 or n(1-pbar) must be greater than 5 for p-bar
greater or equal to 0.5.

(The word "outliers" was used by Jonathan, and I am not sure whether this
meant "special causes"  My response implies that Jonathan's "outliers" are
taken to mean "Special Causes")

Thanks,

Heero Hacquebord
heeroh@mailbag.com

============================================================================





[Next of thread]             
[Next by author]             
[First of thread]            
[First by author]              

Hypothesis testing and control charts/Hacquebord 
heeroh@mailbag.com (Heero Hacquebord)
Thu, 18 Jul 1996 12:14:22 -0500
[Next of thread]             
 [Author List]                        
[First of thread]            
[First by author]              
>
>     From Ian Bradbury:
>
>>The Normal distribution was used by Shewhart in his derivation of formulae to
>>compute control limits
>>for X-bar and R charts. One does not need to assume that data are Normally
>>distributed in order to use
>>X-bar and R charts.
>
>     In the performance measurement work I have been doing, I would agree
>     with Ian's writeup.  I find it worthwhile to think of a control chart
>     as a test of the hypothesis that the process is creating stable,
>     random data that is predictable and "in control".  A failure of the
>     control chart (points out of control, detectable trends) rejects the
>     hypothesis that the process is "in control".
>
>

You would have had Dr. Deming doing flips if he heard anyone relate
hypothesis tests to the control chart.  The underlying theories and
assumptions between these two are vastly different, as are the purposes.

I agree with Dr. Deming's approach that we need to be precise in our
language if we want to reduce misunderstanding.

Regards,

Heero Hacquebord
heeroh@mailbag.com

======================================================================





[Next of thread]             
 [Author List]                        
[First of thread]            
[First by author]              

Re: Myron's Report on Fauerstein/Tolman 
"Anton O. Tolman, Ph.D." 
Thu, 18 Jul 1996 11:32:00 -0700 (MST)
 [Topic List]                          
[Next by author]             
[First of thread]            
[First by author]              
On Monday, July 15th, Myron sent his first report from Jerusalem.  
All I can say is, I bet you're having more fun than we are! :)

I appreciate the information and thoughts, and especially the bridge 
between the language systems.  The concept of paradigm changes is 
essential in achieving change in people and in systems.  This refers 
partly back to my previous posts on the DEN about the effect of 
irrational thought patterns which tend to inhibit learning and growth 
or which actively produce withdrawal from new situations and 
information, or in many cases, produce symptoms of mental illness.

I would bet that Fauerstein may use the term "schema" at times to 
describe some of these processes.  This is also a very useful concept 
for understanding cognitive development and change in people.  A 
schema is a complex brain organization/function which serves not only 
to process information in a specific way, but which actually affects 
the type of information we take in the first place through our 
senses.  For example, a police officer may have a "break" schema 
which organizes the way he takes his morning breaks; perhaps it 
involves eating donuts in a 7-11 store, scanning the headlines of the 
papers, or chatting with the clerk.  What information he attends to, 
how his behaviors are shaped, and what he thinks about are governed 
by his schema or set.  However, if he gets a call to the 7-11 saying 
that it is being held up at gunpoint, he may go back to the same 
environment with a _different_ schema, say a "danger" schema 
activated.  This schema will process information entirely differently 
-- both perceptually and cognitively.  He will be scanning the 
environment differently -- for movement, gunshots, screams, escape 
routes, etc. and will NOT notice the information he usually processes 
when in that same environment.  

The description of how particular schemas are activated and 
deactivated at different times would seem essential for education and 
learning.  I hope that some of what you will post, Myron, will help 
me understand this system better.  I think I will have a look at 
Fauerstein's book as well.  I look forward to your continued posts.  

________________________________
Anton O. Tolman, Ph.D.
Director, Psychological Services
Director, Program Services
Wyoming State Hospital
P.O. Box 177
Evanston, WY  82931-0177
Anton@wsh.state.wy.us
(307) 789-3464
---------------    

===========================================================================




 [Topic List]                          
[Next by author]             
[First of thread]            
[First by author]              

Wheeler & Control Charts/Stauffer
Ripstaur@aol.com
Thu, 18 Jul 1996 17:10:50 -0400
[Next of thread]             
[Next by author]             
[First of thread]            
[First by author]            
       I have been reading all the comments on XmR vs. count charts with a
lot of interest. A lot of appeals to theory have been made, some more
convincing than others. Since I feel partially responsible for kicking off
some of this controversy (and some of the responses are starting to get
vicious), I now feel compelled to chime in again with an appeal to reason.
Once again, my line of reasoning comes from Wheeler's texts and seminars. I
have read and heard many statisticians on control charts, but his reasoning
and the evidence he presents, while "heretical" to the ears of
traditionalists, is very convincing. It is also interesting to note that he
ran most of his material by Dr. Deming prior to publication. Dr. Deming
apparently approved of what he read, judging from the Forward to
Understanding Statistical Process Control (and from other letters Dr. Deming
wrote to Dr. Wheeler over the years). Don told me, "Dr. Deming was never shy
about telling me when he thought I was misguided about anything." 
       I asked Don specifically about the material presented on using XmR
charts for count data. He said, "Yes, Deming was very interested in it. He
loved the material in Understanding Statistical Process Control." I don't
think it's a matter of whether Dr. Wheeler is credible or not; if Dr. Deming
thought he was a credible resource, who am I to argue? 
       Besides, I've read all his books. I've been to his seminars, and I
have corresponded with him quite a bit. The ideas he presents are not what we
learned in school, and not what we have read in most texts, but the bottom
line is, they work. He has presented a cogent, coherent argument for a very
practical approach which has worked time after time. Further, he has
reexamined some concepts in light of theory, and is able to show that the
theory actually tends to support this approach, rather than the converse. He
has data and experience to back up the theory. 
     Advanced Topics in Statistical Process Control could be the most
complete argument on control charts since Shewhart's Economical Control of
Manufactured Product. I haven't been able to find any better; this isn't to
say that I am not still searching. I have read a lot of arguments, but most
are long on pure theory, "what we learned in basic statistics," and
paradigm-bound. Wheeler makes these statements that seem outrageous, but he
presents very clear, logical, arguments to back them up, then goes on to
present data, REAL WORLD data, to show that the theory, and the argument,
hold up very well.
      I'm probably the last person anyone should listen to in advancing these
theories. I don't have a PhD (yet), and although I read everything I can
find, I obviously can't read everything. I am just a person in search of
knowledge like everyone else on the DEN. The problem I see with most of these
arguments is they appeal to limited aspects of some of the theoretical ideas,
and generally come without conclusive evidence for support. I also sense that
many of those arguing with Wheeler's ideas have not yet read his books,
because instead of directly addressing his arguments, they mostly say, "it
can't be true because it's not what I was taught." 
      
      The two main arguments I've seen recently concerned (a) the shape of
the distribution/central limit theorem and (b) the use of XmR charts for
count data.

       I'd like to quote from section 5.2 of Advanced Topics in SPC to (try
to) answer the first argument. Wheeler talks first about the fact that the
CLT only applies to averages, not to ranges, then he says:

     "The second reason that the central limit theorem cannot be said to be
the basis for control charts is the conservative nature of three-sigma
limits. When you have already bracketed approximately 99% to 100% of the
values by brute force, one does not need to appeal to any theoretical
arguments to understand that a point outside the limits is very likely to
represent a signal. The conservative nature of three-sigma limits makes the
central limit theorem irrelevant.
      "While Walter Shewhart did mention the Central Limit Theorem for
subgroup averages on pages 181-182 of his 1931 book, this was not the
operating basis for the computation of control limits. He merely used the
Central Limit Theorem for averages to support the claim that the Camp-Meidell
Inequality may be used in lieu of the more conservative Tchebycheff
Inequality. Following this discussion of the control chart for Subgroup
Averages, Shewhart discussed control charts for:
       (1) Mid-Ranges,
       (2) Root Mean Square Deviations,
       (3) Subgroup Variances,
       (4) Subgroup Ranges,
       (5) Skewness, and
       (6) Kurtosis.
In each of these cases Shewhart noted, and empirically demonstrated, that the
conditions for the Camp-Meidell Inequality are not satisfied. (There is no
Central Limit Theorem for measures of dispersion or measures of shape.) Thus,
Shewhart returned to the Tchebycheff Inequality as his justification that one
may establish meaningful control limits for these various statistics. Between
page 174 and page 213 of Economic Control of Quality of Manufactured Product,
Shewhart argued that there need be no stronger condition than Tchebycheff's
Inequality in order to establish a reasonable control chart scheme. He then
continued to refine his arguments for another 100 pages before giving the
definitive guidelines of using three-sigma limits.
     "In other words, the Central Limit Theorem for Subgroup Averages is a
true theorem, and it does facilitate our work with sample averages in many
situations. However, when it comes to the use of three-sigma limits, the
existence of a Central Limit Theorem is irrelevant. ^Regardless of the
distribution of the sample statistic, virtually all of the values will fall
within three-sigma limits whenever the process displays statistical control.^
     "Undoubtedly, this myth has been one of the greatest barriers to the
effective use of control charts with management data nad process-industry
data. Whenever data are obtained one-value-per-time-period it will be logical
to use "subgroups of size one." However, if someone believes this myth, they
will feel compelled to average something in order to invoke th blessing of
the central limit theorem, and the rationality of the data analysis will be
sacrificed to superstition."

      Concerning the second argument, one argument against using XmR charts
for counts has been the risk of masking special cause data due to the way the
chart is calculated. Dr. Wheeler freely acknowledges that there are fewer
degrees of freedom for XmRs, so they usually require more data to firm up the
control limits. He also notes that as a general rule, it will be a bad idea
to use an XmR if your count per area of opportunity is less than one. 
      However, he gives a data set in section 11.2, some real-world data, for
a film producing process, counting blemishes in the film. The data for line 2
are:

       12, 9, 19, 6, 13, 15, 8, 11, 9, 11, 9, 10, 10, 13, 13, 11

If you do an XmR and a c-chart for these you find:
     "While the limits found using the XmR chart do differ slightly from
those found using the c-chart, they do not differ appreciably (note: less
than 1/10th of a blemish-RS). Moreover, the Moving Range Chart identifies an
anomaly on line two that is not identified by the c-chart--the drop from 19
to 6. Such a dramatic drop in the number of blemishes could be a very
important clue on how to improve the performance of all the lines, and it
will only be detected if one is using the mR chart."

     These are the short answers, I guess. If anyone wants longer
explanations, my advice is to read the book. My intention in giving these
quotes is not to directly argue the points myself, but to (hopefully) pique
some interest in what is, apparently, a paradigm-breaking publication. I
hadn't realized the immense controversy these ideas would engender.

Rip Stauffer
Naval Leader Training Unit
Ripstaur@aol.com

(Quotations from Advanced Topics in Statistical Process Control are used by
permission of the author.)

=======================================================================




[Next of thread]             
[Next by author]             
[First of thread]            
[First by author]            

Deming Questions # 20/Gogue
gogue@cri.ensmp.fr (GOGUE J.M. Societe MAST 39 50 99 67)
Fri, 19 Jul 96 11:35:47 +0200
[Next of thread]             
[Next by author]             
[First of thread]             
[First by author]             

--------------------------  DEMING QUESTIONS  -------------------------- 

A weekly paper including a set of questions copied in the Deming book 
OUT OF THE CRISIS   Chapter 5.  Questions to Help Managers.

By Jean-Marie Gogue

-----
# 20
-----

1.   (No 39 in the Book) - Reliability of inspection and test

a)    How  reliable is your inspection at each of these points? 
      How do you know?

b)    What data have you to show whether your inspectors are in line 
      with each other?

c)    What  about your test instruments, or rather, your use of them? 
      Can you present evidence of statistical control of the system of 
      measurement or classification? Visually? Or by instrument?

COMMENTS
Perhaps some readers may think they are not concerned, but inspection exists 
everywhere. Administration, police, schools, universities, department stores, 
wine shops, doctors, etc. have inspection. Even families with children have 
inspection. 

The concept of inspection suffers from the Taylor's cliche' of the inspector  
in a factory , sorting out defective products. But Deming was not an 
engineer, and his teachings were not devoted to factories. In the Deming 
philosophy, inspection is one of the three steps of the universal cycle 
proposed by Shewhart in 1938: Specification, Production, Inspection, that 
he transformed into the Deming cycle: Design, Make, Put on the market, Test 
in service. Thus inspection is related to Knowledge.

Misunderstanding about the aim of inspection leads to a poor inspection 
system, that is a big source of waste. This question is a gold mine.

-----

2.   (No 40 in the Book) - Optimizing the cost of inspection

a)    Where is inspection being carried out where no inspection would 
      minimize total cost?

b)    At which  points are you carrying out no inspection where you ought 
      to carry out 100 per cent inspection to minimize cost? (See Ch. 15.)

COMMENTS
 I remember visiting a French manufacturing plant with Dr. Deming in 1980. 
In the inspection and test areas he asked questions about the cost of 
materials, the investments, the time of operations, the percentage of 
defects. Two weeks later I received a letter where he explained that 
their inspection system was far from the optimum. In some places they 
were under control, with almost no failures, but 100 per cent inspection. 
In other places, no inspection, but out of control with a huge cost of 
defects. He made calculations to estimate the possible savings: about 
2 percent of their sales.

----------------------------------------------------------------------------

Jean-Marie Gogue
President
The French Deming Association
Versailles   France
gogue@ensmp.fr

==================================================================



[Next of thread]             
[Next by author]             
[First of thread]             
[First by author]             

Applications for Chaos?/Griffiths
Ian Griffiths <100326.1264@CompuServe.COM>
19 Jul 96 09:37:23 EDT
 [Topic List]                          
[Next by author]             
[First of thread]            
[First by author]             
In response to Steve Prevette's note re Chaos and Randomness:

This is just a comment on points raised in Steve's note.

I am still very unclear about the practical importance for business systems of
work being done on chaotic behaviour. But my guess is that it will be, AT LEAST,
insightful -- if only in helping us better to classify problem types. It may do
much more. We need to find out.

It is useful to know there are business problems (characterised by small numbers
of players) where game theory can be helpful, and others where the number of
inluences (or players) is very high and statistical thinking is fruitful. It is
also helpful to know what are the 'limits or boundaries on the useful'. In
mathematics it is of practical value to be able to prove that some things cannot
be done (so we don't waste time attempting to calculate or prove something that
is unachievable).

A century ago most scientifically educated people believed that the motion of
planetary bodies was rigorously determined by Newton's Laws and (in principle)
precisely calculable. For most practical purposes, we still act in the belief
that this is very nearly true. But Poincare showed nearly a century ago that
this was not really the case. His work laid the foundations of the serious study
of chaotic behaviour, though he did not use modern terminology. The interesting
thing here is that the motion of a planetary body can appear to be extremely
regular for a long time and then (apparently out of the blue) it may change
drastically. This happens without the 'system' changing in any way. It is not as
if two bodies collide, or something of that sort. The STRUCTURE of the system is
the same as before; it's just that the BEHAVIOUR of the system appears to become
very different.

The same sort of behaviour happens also in social and ecological systems. It is
quite easy to provide a mathematical description of a number of interacting
'loops' of influence that exhibits 'ordinary' regular behaviour for long periods
and then 'flips' into a surprising new sort of behaviour. The behaviour may not
even be chaotic in a strict mathematical sense -- but it can be VERY surprising.
The important point is that, to an observer, this looks very much like a change
from 'in control' to 'out of control' behaviour by the system. BUT it would be
futile to look for some new 'external cause'. The 'strange' behaviour is
entirely natural and intrinsic to the system.

In the real world there are a host of problems that would be better tackled if
people had greater awareness of the statistical ideas developed by Shewart and
Deming. We know that simple tools and techniques, coupled with an awareness of
'systems', and a concern for people, can be very fruitful. We have hardly
started down this track of improvement.

But lurking in the background is a set of new issues where we need to improve
our understanding. Some of the problems are of major importance. For example, it
is NOT enough to track the catch of a fish population without some prior
understanding of the population dynamics. We now know this (there are reasonable
theories and lots of experimental evidence) -- and we know there are lots of
problems of this sort. These are certainly not just academic problems, even
though much of the work currently being done in this area appears rather
'academic'. As understanding improves, we can expect the subject to become more
'practical'.

We certainly need to do a lot more work in this area.

Ian Griffiths
email: 100326.1264@compuserve.com
tel: (44) 01383 872937

========================================================================




 [Topic List]                          
[Next by author]             
[First of thread]            
[First by author]