「華人戴明學院」是戴明哲學的學習共同體 ,致力於淵博型智識系統的研究、推廣和運用。 The purpose of this blog is to advance the ideas and ideals of W. Edwards Deming.

2013年5月17日 星期五

Graduate School USA recognizes HUD’s excellence in workforce development and training 2013


Graduate School USA recognizes HUD’s excellence in workforce development and training
WASHINGTON, DC – May 17, 2013 – (RealEstateRama) — This week, the U.S. Department of Housing and Urban Development’s Office of Field Policy and Management (FPM) was presented with the Graduate School USA’s annual W. Edwards Deming Award for a staff training and orientation program to reinforce the use of next generation cloud computing.  The W. Edwards Deming Award recognizes impressive workforce development and training initiatives that measurably improve an organization’s performance and impacts its mission.
FPM was selected for implementing its training and orientation program to reinforce the use of cloud computing for tracking and reporting project performance in order to improve its ongoing reporting to Congress and other key constituents.
“This cloud based system has transformed HUD into a 21st century agency – through the use of this cloud system we can now see how our strategic goals are being achieved throughout this great nation and in the unique localities we serve” said Pat Hoban-Moore, Director, Office of Field Policy and Management. “Our staff in Field Policy and Management can now align their daily work to the agency’s strategic plan and understand how they directly impact HUD’s mission.”
The program improved the ability of field-level staff members to track their projects, demonstrate alignment to HUD strategic goals and report on their progress. The complexity of the organizational mission and the geographic distribution of the target populations presented unique challenges, with hundreds of employees in 80 field offices throughout the United States and oversight by 10 regional headquarters. The reporting application is used on a daily basis by FPM staff in the field and at HUD headquarters in Washington, DC.
The cloud based application is used by 100 percent of FPM staff, with 25-30 percent logging in daily to perform project-related work. Numerous field recommendations have resulted in improved reporting to Congress and other critical information recipients.
The award’s namesake is Dr. W. Edwards Deming, who taught at Graduate School USA for 20 years and is considered the father of Total Quality Management.  His work incorporated the idea that employees and managers at all levels of private industry and government have a responsibility to cooperate to improve products and services.
HUD’s mission is to create strong, sustainable, inclusive communities and quality affordable homes for all.
HUD is working to
strengthen the housing market to bolster the economy and protect consumers; meet the
need for quality affordable rental homes: utilize housing as a platform for improving quality of life; build
inclusive and sustainable communities free from discrimination; and transform the way HUD does business.
More information about HUD and its programs is available on the Internet at
www.hud.gov and
. You can also follow HUD on twitter @HUDnews, on facebook at
www.facebook.com/HUD, or sign up for news alerts on HUD’s News Listserv.

2013年5月13日 星期一

Remembering Prof. David Kerridge

Thank you for preparing this piece of David's story for us. The Funeral service will be on Monday 20th May at Kings College Chapel, Aberdeen University. Anyone who is able to attend is welcome.

If we could have your story of how you knew David and the input he had on your life before then, we would be grateful.

Thank You

Deborah Armstrong

Remembering Prof. David Kerridge
                                                             By Hanching Chung

Ever since I wrote to Prof. David Kerridge and received a long reply to my questions in 1995, I think David was my mentor. Although we did meet, I knew we are friends. Let me take  advantage of this note to offer our sincere condolences of Taiwanese Deming Circle to his family.

Indeed, I am one of members of David’s worldwide Deming Philosophy Circle. David put the idea of it to some friends in email, but I laughed at the idea of it, cited many historical cases failed.Nevertheless, David continued to be my never-ending resource on Deming/Shewhart philosophy. Nearly every year, David and his daughter Sarah got some well-thought papers. In his late years, he even shared with us his papers in his Apple computer disk.

David Wrote Dr. Deming’s obituary in European Quality (  "The official journal of the European Organization for Quality"), Circa 1995. He involved in Dr. Deming’s seminars since late 1980s. In Wikipedia’s “W. Edwards Deming” item, a note tells us that

 "Of the four experts, Deming, who can be the harshest as a teacher, seems the most humanistic, insisting that it is every person's right to have "joy in work." He used to say "pride" until David Kerridge, a professor at the University of Aberdeen, pointed out that the Book of Ecclesiastes says "joy" in two different verses. Deming, whose one known hobby is writing liturgical masses, switched to joy. He estimates that no more than two in a hundred managers and  ten in a hundred workers now have joy in their work. In Quality or else: the revolution in world business Lloyd Dobyns, Clare Crawford-Mason – 1991.

I joined British Deming Association (BDA) several years since 1996.  In each issue of the Variations newsletter, I enjoyed David and Sarah’s short essays very much, for example, “Guilty by Association” tells a typical Dr. Deming’s reaction to so-called Total Quality Management (TQM). Their English are good and I hoped someday to publish bilingual version of their collection of essays. Unfortunate, this idea was only partially completed yet. David was the chairman of BDA’s Research Committee. His annual reports for the committee were models of any research organization.  

Thanks to David’s kind permissions, In 1998 I put two essays in my Chinese web siteTransformation not Tampering and “A Model of Transformation” which I thought it is very inspiring. But David was learning all the time, when I proposed to put it in the coming book edited by me about 2008, David thought it is not very good and suggested some other article. In 1999 Feb. I put their essay “Operational Meaning” translation in my website.  (Deming Electronic Network website used to have a collection of David & Sarah Kerridge’s essays at that time.)

David and Sarah’s contributions to some professional journals are worth mentioned. For example, “ Applying the DEMING PHILOSOPHY to the Safety System “ , Professional Safety; Aug 2006, Vol. 51 Issue 8, p.52. Their contributions to The Journal for Quality & Participation are very good: Dr. Deming's cure for a sick system (;Dec 96, Vol. 19 Issue 7, p.24) and “Managing complexity” (March 1, 1997).

I read  “Managing complexity” in The Quality Yearbook ( Cortada and Woods (editors) , McGraw-Hill , 1998) and wrote to David to congratulate them. A Chinese version of it was put in our book System and Variations (2010 ). David also wrote a foreword for our book The Essential Deming (2009). David also helped me to translate Dr. Deming’s paper On Probability As a Basis For Action (American Statistician, 29, 1975, pp.146-52).

David was an authority on Walter Shewhart’s philosophy. He helped the paper
“W. Edwards Deming’s mentor and others who made a significant impact on his views during the 1920s and 1930s” (Beth Blankenship and Peter B. Petersen, Journal of Management History, Vol. 5 Iss: 8, pp.454 – 467). I remember how happy he was when I discovered a "Tribute" for Dr. Shewhart in U.S.A..His reply to my mail listed for your reference: “Thanks for this web site. I have extracted the photographs from the html pages, and enlarged them. This makes all the writing in the pictures,
even the handwriting, easy to read.  Yes, the main tribute is from ASQC.”

He was also pleased to know that Dr. Deming’s name was mentioned in the note of one of Sir Ronald Aylmer, Fisher (1890-1962) paper in Fisher’s Archieve. I knew his profound knowledge of statistical thinking so I asked him many “Big Questions”. One of his answer is “I think that R A Fisher saw further than Neyman and Pearson, but not as far as Shewhart.” In 2011, I had several mails with him about the Scientific Inference by Sir Harold Jeffreys. He was always very helpful.

In 2009 David had a sense of urgency and replied to one of my mails to him:

 “I am afraid that the generation of us who knew and worked with Dr Deming is getting older. We badly need a new generation of scientists who understand the Shewhart/Deming ideas and will take them forward. So much research remains to be done...”

                                                     May 13, 2013, Taipei, Taiwan


鍾漢清加入英國戴明協會(BDA)。認識 David Kerridge 教授,深受關愛。獲創始會長 Henry R. Neave 博士贈其佳作 The Deming Dimension


Prof. David Kerridge贈論文『The Theory of Knowledge』等

; 前英國戴明協會的「統計研究小組」(仍續由 Dave Kerridge領導),在著名的醫學刊物上登一Shewhart管制圖的應用:六個英國醫院和醫師的績效(包括去年震驚全世界的醫師兼凶手的高死亡 率),利用二項式Shewhart管制圖來區分共同因誤為特殊因,來學習科學的改善。文章資料如下:Mohammed MA, Cheng KK, Rouse A, Marshall T."Bristol, Shipman and clinical governance: Shewhart's forgotten lessons." Lancet 2001;357:463-67. (to be published on the internet 9th of Feb http://www.thelancet.com/ and on the 10th in paper form)Lancet 手術中用的柳葉刀或稱刺血針,常用它來代表醫學這一專業,例如The Lancet為創始於1823年的英文醫學、醫物刊物,現在為周刊。中國青島有「柳葉刀叢書」。

Dear Hanching,
Thanks for your thoughts. I especially appreciate your pointing out Professor's Johnson excellent speech, containing similar themes, which I found at:



HC/ David Hsu 2006-10-27 08:35:15



在「明目」與兩位留德的老師談德國政治和文化。周四買四本書,兩本與英國文化相關,讀趙珩{彀外譚屑──近五十年聞見摭憶}(北京:三聯書店) 中之數篇,還有些古風範,也讀到很間接的"朋友"…….

中華民國品質學會QKC每月聚會(地點: 台北市羅斯福路二段75號九樓)。我與林公孚老師談昨天之學習:包括"時空 spread"…..
十月10/26主講人陳善德 題目為 教育品質:「系統思考」與「心」教育【會前讀David KERRIDGE教授寫 Deming and Learning……..

主持人陳寬仁老師剛從台中的「英雄館勤益」回來:請吃太陽餅。他最近"驛馬星動",剛從廈門歸又要……「去北京要交文章一篇 前言 結語 請看看 不必客氣 請修正 :民生建設過程中『貨惡其棄於地也』之體認」

草草回復:但憑記憶:中山先生著作(含三民主義)似乎著重infrastructures基礎建設,所以有蔣中正先生之{民生育樂補篇}。我認為過去50 年台灣剛好建立民生產業多樣產品之世界級能力,而這正是大陸最短缺的,也是將中山思想提到滿足全世界要求之水平。所以大陸之改革開放,其實是這領域的迎頭 趕上。而台灣貢獻頗大,這當然包括台灣掌握民生產品的品質與生產力訣竅,這,或許加上大陸的改革可開創第二春。大陸Juran quality handbook 一印即10萬本或可當作一明證。


 Dear Hanching,

>We encountered an issue in the Essential Deming proof-reading. Perhaps
>can help .
>We don't know how to type statistical symbols.
>Perhaps you can forward them to us.
>First, the formula of Shewhart's optimal sample size (which Tippett
>independently discovered it).

Yes, I am afraid that mathematical and statistical symbols are a
terrible nuisance. They were designed in the days of pencil and paper,
so they could be anything you like. But they don't fit in with modern

Unfortunately I don't have the Shewhart paper you are referring to, so I
don't know what formula you want. The other problem is that I work on an
Apple computer (made in Taiwan), and know nothing about how to type the
symbols on a Microsoft based machine. There is, unfortunately not
complete standardisation of type-faces.

Anyway, if you show me what formula you want, I can try to set it up on
my machine, and send you the result.

Henry Neave told me recently that he had problems of the same kind, in
producing a new edition of his statistical tables. But he found all the
answers by searching various articles on the internet. He also has an
Apple machine.

>Second help , if you can forwards "Probability as Basis of Action"
(textform not PDF form.

Yes, I have attached it in Rich Text Format, so that it keeps some of
the formatting, but unfortunately the tables get messed up.

 Dear Hanching,

I have just checked back over my correspondence with henry Neave. He
used the "Equation Editor" that comes free with Appleworks (but not with
the more advanced iWork package that I use now.

If you have much of this sort of thing to do, I believe that MathMagic,
which is much more powerful than the free Equation Editor has versions
for all operating systems.

But as I said, if you let me know what equation you want, I will see if
I can set it up.

Best wishes

 Dear Hanching,

>I can not open the file. I checked with my draft, only symbol of
>"Hypothesis" needed. Other placed in my book seems solved.

Sorry, I thought that Rich Text Format was universal. Here it is in
plain text.

In Deming's ideal world of cooperation between competitors, there would
be none of this deliberate incompatibility. But that seems a long way

Best wishes

(The Essential Deming)

雜之管理 David & Sarah Kerridge
系統與變異: 淵博知識與理想設計法 (2010)

David Kerridge 教授的兩封信都貼在台灣戴明圈

 Dear Hanching Chung

 Dear Hanching,

>Actually I am thinking about the "complexity" paper.

Yes, that's probably the most suitable one. The others assume knowledge
of Shewhart and Deming concepts, which is (sadly) rare even among Deming

>By the way, I just suggest the sponsor to use our educational
>resourse to invite you to join the conference. Is it possible for you
>visit Taiwan next year?

I am afraid I have had to give up all travel in recent years. Its a
great pity, as I was thinking, before you wrote, how interested I would
be in such a conference.

Anyway, thank you for the kind thought. I hope the conference is a great
success. 11/20/09
 >Thanks for the information.
>It seems quite difficult to get members of Deming Philosophy Circle

I am afraid that the generation of us who knew and worked with Dr Deming
is getting older. We badly need a new generation of scientists who
understand the Shewhart/Deming ideas and will take them forward.

So much research remains to be done.

>I wonder the "Tribute" (bottom picture)  was by ASQC?

Thanks for this web site. I have extracted the photographs from the html
pages, and enlarged them. This makes all the wrinting in the pictures,
even the handwriting, easy to read.

Yes, the main tribute is from ASQC.

 Dear Hanching Chung

>The book mentioned about George Barnard and yesterday I knew he died in
> 2002 and first time saw his portrait.

Yes, I knew George Barnard. He was our external examiner for statistics
when I was at Sheffield University, and I often saw him at meetings in

>In 2008 I read a draft d a paper by Deming with a reference to Fisher's
>Statistical Inference book but later Deming decide to drop it for
>reason. (I knew many statisticians prefer Fisher not to write the

Fisher was a remarkable man, but nobody could argue with him. I expect
Deming did not want to give the impression that he approved of Fisher's
theory of inference. Fisher recognised the weakness of inference based
on repeated sampling from the same population - something that is rarely
possible in practice, and equally rarely relevant to the problems of
science. But Fisher's solution was to refer to sampling from a
"hypothetical infinite population." This idea seems too indefinite for
science, since it is not subject to operational definition. But only
Shewhart saw his way through this.

I think that the problem was that Fisher began as a mathematician, and
then became a scientist. That's the wrong way round. Deming and Shewhart
were scientists who became statisticians.

For many mathematicians their abstract models are more real than the
real world itself.

>Yesterday I suggested David Hsu here to write System of Profound
>for Beginners, He asked me advice for his two years book project here,

At a research meeting in New York, Deming sid to me:

"We are all beginners here."

 Dear Hanching Chung

>If it not a bothering for you, how about help me understand " Fisher vs
>Shewhart" their relationship to the concept of  analytic study.
>Or my question is wrongly put. Sorry, I try to pick up some of my
>understanding of your commentary,

I will do my best, but this is a big question, and will take time to
answer properly. It is to do with the whole nature of scientific method,
on which even the philosophers are confused.

I think that R A Fisher saw further than Neyman and Pearson, but not as
far as Shewhart.

Anyway, I will try to put some thoughts in writing, but will not hurry
it, as this is a difficult one to explain.

Best wishes Dear David,
Thanks. I think it is worthwhile to write down what you think and later expand it into an more professional article to share with more more people,
Then I am glad to be a catalyst for learning.

I spent all last two weeks to correct or rewrite this year;s book on System Thinking and Variations, It runs to about 500 pages (my part is about 300 pages, You & Sarah's one rewrite more than four times,) I realised next one shout set limit to no more than 300 pages altogether.

Today I went to National University to read two revised books by George Box One is 1978's Statistics for Experimenters and the other is You can nearly improve everything around 2005 while the first edition in Box on Quality and Discovery (2000) . He come Shewhart-Deming Cycle with his model in an article wrote in 1999.

Best Regards,

昨天接受英國 KERRIDGE 教授的信。我兩周前向他提「科學推論」的大哉問,可能讓他忙翻了。現在畢業了,真好!因為我以前(1978)在導師課(一對一)考老師的問題,他們都在學為考卷上回敬我。

 請教David Kerridge 關於 ' epistemological probability".


Scientific Inference by Sir Harold Jeffreys

Dear David,

I came across a Chinese translation of Scientific Inference by Sir Harold Jeffreys. Unfortunately it is a second edition one .
I can browse it in Google Books, it is a third edition.
I don't know exactly the meaning of his ' epistemological probability".
And what is the main difference between  his view of probability and dr. Deming's.
Perhaps you might be of help to explain it to us.


Dear Hanching,

That's a really difficult question. I will do what I can, but I had
better think first how to put it.

It happens that I spent years trying to solve to problems of "What is

Scientific Inference was one of the books that I found very inspiring.
But so were a lot of others.

Anyway, I will try to make some useful comments next week. But I must
find a way to reduce it to a few words.

Dear Hanching

Just to let you know that I am still working on answering your question.
I am including in the comparison the commonly taught frequency view of
probability as well, because I can't see how to compare Jeffreys and
Deming without it.

I don't have a copy of Scientific Inference, though I read it many
times. But I do have copy of Jeffreys other book "The Theory of
Probability" and I attach a copy of the preface, which is relevant.

Best wishes
Dear David,

Actually we can read most parts of the Scientific Inference - Harold Jeffreys  (The Google Books)

Thanks again.
Dear Hanching,

Thanks for the reference to Scientific Inference on Google books. My
out-of date system won't let me access it at present, but I will note
the address.

However, I thought you were asking about probability, rather than
inference. Perhaps I misunderstood the question.

Anyway, I will do my best, and then you can ask further questions.

Dear Hanching

I am sorry to have been such a very long time in answering your
interesting question. There have been two reasons. I have been ill, and
found it hard to concentrate. But that would not have stopped me if the
question did not require a lot of concentration.

Both Dr Deming and Sir Harold Jeffreys held views of probability that
are different from those usually taught. They are, in fact extreme
cases. Sir Harold Jeffreys held views that are regarded by most people
as very out-of-date, being the same as those held by LaPlace a hundred
years before.

Deming (following Shewhart) based his views on the new philosophy of
science that came in with Relativity and Quantum Theory. This is so
advanced that few other statisticians are even aware of it yet. Shewhart
assuned that it would be universally adopted in time: but there is
little sign of it yet. (John W Tukey is the one exception that comes to

That's *why* there is a difference, and why both differ from the views
of probability in most textbooks. But to explain what the difference is,
is not easy, without explaining the different philosophical viewpoints.

That's what I have spent a lot of time trying to do. It makes it no
easier that Deming did not explain his view of probability explicitly. I
had to deduce it from remarks he made in research meetings, and in
papers like "On Probability as a Basis for Action."

Note that for Jeffreys, and those before him, probability is simply a
matter of logic: action may result from probability, but the nature of
probability is not defined in terms of action. For Shewhart and Deming,
everything scientific is defined in terms of action, rather than
thought, because actions can be observed, while thoughts can not.

This really requires a book to explain it - a pity Shewhart didn't write
one on this topic. But he probably felt that the problem of defining
probability was still not completely solved.

I spent years on this problem using the same advanced philosophical
viewpoint as Shewhart, long before I met Deming. I couldn't find any
other statistician in the UK who understood what I was saying. And I
heard Deming discussing his ideas with others, who couldn't see what he
meant. So he had the same problem.

This may be enough to be going on with. The rest may take a long time to
express in simple words. I intend to try, but it can't be hurried.

Best wishes

Dear David,

Thanks very much. I am very surprised. I thought you gave up to popularize the subject. 

I like to keep you inform that Joyce is editing a book of Dr. Deming's papers and speeches. The book will publish by McGraw-Hill Next year.

Last week I read a book review from New York Times. I wrote a note in  the blog that you told me about 10 years ago that while Dr. Deming was in England, he helped to redistribute the papers by Mr Bayes.

The article was listed for your reference. Thanks again.

Best Wishes,

The Mathematics of Changing Your Mind


Dear Hanching,

Thanks for the information about Joyce Orsini's forthcoming book, and
the review of the book on Bayes Theorem.

I think that there is a lot more to be discovered about Bayes Theorem.
It is certainly a very important practical tool. For example, my email
is scanned for spam by a programme based on Bayes Theorem. But I rather
suspect it is not being used correctly - and still works.

In 1964 I was a Research Fellow, and two of us were investigating
methods of medical diagnosis. In other words, trying to develop
statistical rules for guessing what illness a patient is suffering from,
based on a limited number of tests or symptoms.

We found a paper, written by a computer scientist, that claimed to use
Bayes Theorem in medical diagnosis. But his method did not allow for the
obvious fact that different symptoms are not statistically independent.

We were shocked at such ignorance, and misuse of statistical theory. So
we set out to compare all the best methods we could find, including new
ones we developed ourselves, based on multivariate logistic analysis,
and another based on comparing each case with all the data in a

Sad to say, the "wrong" method, based on bad theory, worked at least as
well as the other methods, though which worked best, for a particular
disease, depended on the sample size available. Ours was better for
large samples.

It seems that a simple, even a wrong model, can equal or sometimes beat
a better model with fewer parameters. I later found out that Norman
Bailey, at Oxford, had discovered the same thing in a different problem
(multiple regression), but found it so shocking that he did not publish

The point I am making is that "correct theory" does not guarantee better
results in practice, and vice versa. Statistical theory hasn't caught up
with this fact yet, as far as I know.

I will try to explain the difference between the way Sir Harold Jeffreys
and W Edwards Deming *used* probability, which tells you more than what
they say. But it will take time, as I like to be thorough.

Title:        W. Edwards Deming’s mentor and others who made a significant impact on his views during the 1920s and 1930s
Author(s):        Beth Blankenship, (Deming Scholar, Hunt Valley, Maryland, USA and), Peter B. Petersen, (Professor of Management and Organization Theory, Johns Hopkins University, Baltimore, Maryland, USA)
Citation:  Beth Blankenship, Peter B. Petersen, (1999) "W. Edwards Deming’s mentor and others who made a significant impact on his views during the 1920s and 1930s", Journal of Management History (Archive), Vol. 5 Iss: 8, pp.454 - 467
David Kerridge  to hc/ Oct 16 (2 days ago)
Dear Hanching Chung

Thanks, the two copies have arrived safely. I just wish I could read

david kerridge/wed
Not that the story need be long, but it will take a long while to make it short.                Thoreau
And we must never forget Walter Shewhart's fundamental yet
neglected discovery. If a process is under statistical control,
trying harder cannot produce better results, and often makes things
worse. So if rewards *do* make people try harder, and the results
are better, the process was not under control. Why persist with
a bad system? David Kerridge
Shewhart wrote "Progress in modifying our concept of control
has been *and will be* comparatively slow." (My emphasis).
By control in this context he meant, I believe, his whole
approach, not just control charting, though that is a good
place to start.

Since 2008, I published one or two books annually on Deming Philosophy. So I think it might be a good idea to send you 3 books now available (a small part of them are in English). I can send you several set if you think it is acceptable. Prof. David Kerridge has a set and you might like his opinions.

David and Sarah Kerridge, „Aristotle's Mistake“. 31. May 2002. Aristotle's Mistake or the ...
  • "joy in work" the phrase, originally "pride in work" was amended to "joy" by Deming in 1988, after David Kerridge, professor of statistics at Aberdeen, pointed out that "joy" in labour was found twice in the Book of Ecclesiastes.[34][35]

The Wisdom of David Kerridge, Part 1

By Davis Balestracci on Jun 30, 2009 | Categorized under: General
Back to basics
I discovered a wonderful unpublished paper by David and Sarah Kerridge several years ago. Its influence on my thinking has been nothing short of profound. As statistical methods get more and more embedded in everyday organizational quality improvements, I feel that now is the time to get us “back to basics”—but a set of basics that is woefully misunderstood, if taught at all. Professor Kerridge is an academic at the University of Aberdeen in Scotland, and I consider him one of the leading Deming thinkers in the world today.
Deming distinguished between two types of statistical study, which he called “enumerative” and “analytic.” The key connection for quality improvement is about the way that statistics relates to reality and lays the foundation for a theory of using statistics.
Because everyday processes are usually not static “populations,” the question becomes, “What other knowledge, beyond probability theory, is needed to form a basis for action in the real world?” The perspective from which virtually all college courses are taught—population based—invalidates many of its techniques in a work environment, as opposed to a strictly research environment.
To translate to medicine, there are three kinds of statistics:
  • Descriptive . What can I say about this specific patient?
  • Enumerative. What can I say about this specific group of patients?
  • Analytic. What can I say about the process that produced this specific group of patients and its results?
Let’s suppose there is a claim that, as a result of a new infection-control policy, acquired-MRSA (methicillin-resistant Staphylococcus aureus, a strain of staph that is resistant to the broad-spectrum antibiotics commonly used to treat infections) in a particular hospital has been reduced by 27 percent—a result that would be extremely desirable if that kind of reduction could be produced in other hospitals, or in public health communities, by using the same methods. However, there are a great many questions to ask before we can act, even if the action is to design an experiment to find out more.
Counting the number of infections in different years is an enumerative problem (defining “acquired infection” and counting them for this specific hospital). Interpreting the change is an analytic problem.
Could the 27-percent reduction be due to chance? If we imagine a set of constant conditions, which would lead, on average, to 100 infections, we can, on the simplest mathematical model (Poisson counts), expect the number we actually see to be anything between 70 and 130. If there were 130 infections one year, and 70 infections the next year, people would think that there had been a great improvement—but this could be just chance. This is the least of our problems.
Some of the infections may be related, as in a temporary outbreak or pandemic. If so, the model is wrong, because it assumes that infections are independent; or the methods of counting might have changed from one year to the next (Are you counting all suspicious infections, or only confirmed cases?). Without knowing about such things we cannot predict from these figures what will happen next year. So if we want to draw the conclusion that the 27-percent reduction is a “real” one, that is, one which will continue in the future, we must use knowledge about the problem that is not given by those figures alone.
Even less can we predict accurately what would happen in a different hospital, or a different country. The causes of infection, or the effect of a change in infection control methods, may be completely different.
So this is the context of the distinction between enumerative and analytic uses of statistics. Some things can be determined by calculation alone, others require the use of judgment or knowledge of the subject, others are almost unknowable. Luckily, your actions to get more information inherently improve the situation, because when you understand the sources of uncertainty, you understand how to reduce it.
Most mathematical statisticians state statistical problems in terms of repeated sampling from the same population. This leads to a very simple mathematical theory, but does not relate to the real needs of the statistical user. You cannot take repeated samples from the exact same population, except in rare cases. It’s a different kind of problem—sampling from an imaginary population.
In every application of statistics we have to decide how far we can trust results obtained at one time, and under one set of circumstances, as a guide to what will happen at some other time, and under new circumstances. Statistical theory, as it is stated in most textbooks, simply analyzes what would happen if we took repeated, strictly random samples, from the same population, under circumstances in which nothing changes with time.
This does tell us something. It tells us what would happen under the most favorable imaginable circumstances. In almost all applications, we do not want a description of the past but a prediction of the future. For this we must rely on theoretical knowledge of the subject, at least as much as on the theory of probability.
So, get your head around these concepts and I’ll give you more of Kerridge’s wisdom in my next column that relates to everyday work.
As you see, it is totally different from the clinical trial mindset in which most physicians have been taught. There is an additional problem, since you shouldn’t have acquired infections (or medical errors, or pressure ulcers), epidemiologists have a tendency to treat any infection as a special cause and want to determine the root cause. This would be helpful in an outbreak, but in terms of everyday work, one usually has to take the view that you are perfectly designed to have infections—you must at least consider the possibility of using a common cause strategy, and “plotting the dots” will tell you which.
For more information on using strategies, see Chapter 8 of my book, Data Sanity: A Quantum Leap to Unprecedented Results (Medical Group Management Association, 2009).
A review of this book is available here: www.qualitydigest.com/inside/quality-insider-news/books-data-sanity-statistics-are-doable.html.
You can order Data Sanity at www5.mgma.com/ecom/Default.aspx?action=INVProductDetails&args=3785&tabid=138.

2013年5月12日 星期日


我忘掉曾經將這篇放在網路上 十幾年之後用速讀方式補充一下

〝A review and a reflection of promoting Dr. W. E. Demings Statistical Thinking〞
本文的目的在說明愛德華‧戴明(W. Edwards Deming)的「淵博知識體系」中的根本,即統計思考(變異觀)的大綱及意義,以及它對其他如「系統觀」、「知識理論」、「心理學」的一些應用,尤其注 重他極強調的「分析性研究」(作業定義(衡量與溝通)、透過SPC及DOE等來從事持續改善品質、降低變異),並舉些例子說明何以他的統計學觀,可以對企 業界有深遠的影響。
我們並介紹台灣近年如何推行「戴明學派」,包括 出版叢書、從事各種教育訓練等活動,並給予推動上的一些省思及展望。

Purpose Statement:
To have an overview of late Dr. W. E. Demings profound knowledge system , and focused on his thinking on variations and analytic studies.
In the ways of:
Summarizing his basic specifications for understanding variations and its reduction methods;his methodology of common causes vs. special causes strategies for variation reduction;differentiate and illustrate analytic studies in the areas of experimental design and measurement for some performance;a case report of organizational learning for〝Fourth Generation Management〞and〝Four Days with Dr. Deming〞.
So that
We can understand the power and the potential constraints in the promotion of Demings statistical thinking in Chinese-speaking communities and enterprises.


(美)W. E. 戴明(1900-1993)逝世以來,他作為二十世紀偉大的經營管理思想家的地位,日益穩固。舉個例子說,1996年英國的〝Management Today〞雜誌譽其為「本世紀十大管理思想家之一」。許多重要的管理學作家,例如著名的組織學家Charles Handy在其名著《非理性的時代》中,譽其書〝Out of the Crisis(《轉危為安》)〞為每位經理人必讀的書。Intel的A. Grove在《十倍速的時代》中,雖不同意戴明對「恐懼」的看法,然而那是名詞界定的問題而已。這些,不管是正面或辯證的過程,在在皆肯定戴明學院在經營 管理界的影響力。

當然,戴明學說並不只是可應用在「產、官、學」等行業的「經營管理(management)」上,它既是一種擴延鏈(extention chain)上的「新經濟學(The New Economics)」,也是要「促進社會的繁榮、世界的和平」。(我在2009年的"轉型"一書中解釋為什麼取名新經濟學。

在上述這種聲譽下,我們當然更要研究他的哲學。這方面,台灣領先其他非英語系的國家,因為「戴明學派」(即戴明作品及其寫序肯定之作)在1997年內,會 陸續翻譯完成、出版。

然而,正如所謂的「組織學習之悖論(組織本身只是學習工具,甚至是反學習者)」的意義深遠,我們要學習戴明學說,必須深入其核心思想。他的核心思想,當然 是其晚年完成的精義,即「戴明之淵博知識體系(Demings Profound Knowledge System)」。由於其中系統、變異(即統計學)、智識論(持續學習)及心理學等融為一體,環環相扣,所以很有威力。

我這篇文章要指出一些重要論點。第一,淵博知識體系中最具特色的,是變異的理論,這點使戴明有別於其他所謂的「管理思想家或管理大師」。第二,戴明後半輩 子(1948-1993)的名片都印為「統計學顧問(Consultant in Statistical Studies)」。換句話說,他是以統計學的發展與應用為其終生職志,他在組織上的應用,不妨稱之為「發揮統計思考方法的經營管理學之應用」。所以本文 在介紹完戴明的變異理論及分析性研究概念澄清之後,又會回過頭談談企業界的應用。

戴明自己認為所有有志於「轉型(transformation)」的領導者,都要知道、了解、欣賞一些統計學理論之知識,也就是變異的知識。本段特別多談 點他所謂的分析型研究,以符本文之目的。

他的變異(variation)觀為Shewhart-Deming學派中的主角,所以在1990年元月21日的講義中,列舉以下四條為其重點:(講義只 列條文,條文底下為作者之看法。)

一般人沒有這種機率的修養。在管理界,更重要的是,除了變異之外,任何衡量系統要先在「統計的管制狀態」中,才成為系統,而且也要有特性作業定義,才能溝 通,才能談系統之能力,並進而談預測。而「預測」對經營管理是最重要的。在Hand(1996)的論文中,此為統計理論中講《衡量》的非主流派,可是文中 完全不談Shewhart-Deming。

根據Chamber及Wheeler等人的看法,戴明的貢獻是把W. A. Shewhart的「機率原因與非機率原因」概念,用更與經營管理活動相關的話來說明,易名為「共同因vs.特殊因」。雖然戴明說「名稱」不重要,然而他 區分一般系統原因中96%為共同因,該由管理者負責系統的想法等,也是有「深遠影響」的。


前者為求知〝how many?〞,後者為追問〝why?〞,並要預測,是戴明的重要統計方法學起點(詳細說明請參考Deming(1969) 和我2009年出版兩本書)。

戴明博士曾說其終生矢志變異知識的減少。我們根 據上述,以及他的各種講學資料,可以匯總他的「變異觀之基本知識」如下。(除了在分析性研究另談之外,多取材Neave(1990)及 Deming(1993))

這世界總都有變異,而且會持續如此下去─人與人 之間、產出之間、績效之間、產品之間、服務之間都會有變異存在。我們得了解這些變異的意義為何?我們是否該對它採取行動?什麼樣的行動?


錯誤1:對任何過錯、抱怨、錯失、停頓、事故、 缺料等,看似由於特殊原因造成,而實際上是由系統或制度所造成,即是由共同原因之隨機變異所造成者。
錯誤2:對任何過錯、抱怨、錯失、停頓、事故、 缺料等,判為由共同原因所造成,實際上是由特殊原因所造成。

〝這兩種原因誤判的代價多少呢?沒有數據顯示, 會計師何從知道?〞變異之知識有助於我們了解這兩種錯誤及其帶來的損失。錯誤1為干預;〝最佳努力〞者常是干預,使事情更糟。…


所以我們需要這兩種錯誤之最少經濟損失的程序之 知識。

我們如何達到最少的經濟損失呢?如何判斷該製程 是穩定或不穩定呢?這與錯誤1與錯誤2的機率無關。我們如何計算最少損失?他們如何界定呢?因為戴明曾說:「沒有流程是…固定不變的。」即使知道如何定 義,大部份的成本也不清楚及不可知,又如何找最少成本呢?我們需要的是作業定義,可以溝通的行動準則,告訴我們何時、如何做。管制圖(1924年由 Shewhart發明)就是此方面的利器。大部份的人仍未採用。目前採用者,大部份是不知而行,諸多誤解,例如管制圖是用來從事改善的,而不是作「調整」 之用。業界都在未知系統是否穩定時,即大談或計算「製程能力指標」。


MBO造成的損失是未知數,不可知;不過卻不能 忍受。管制界限告訴我們穩定系統的能力所在。如果我們希望成果超出界限,唯一可行而有意義的辦法是改變系統,這是主管的工作。單單設定目標而已,只是干 預。如果我們要別人的績效超出系統之能力,他只能〝欺騙〞才能達到目標。目標常是可以達到的,但這只會增加變異而損及全系統之績效。
請注意,如系統不穩定(即不可預測),談其能力 就無意義。只有在流程達統計的管制狀態,才能談可界定的能力。

這是他不滿於統計學教育未能提供「產、官、 學」、「淵博智識之士」的看法。積極而言,他提出了另一種統計學在組織中的定位、架構。


我們要了解人、團隊間、事業部間、公司間、產業 間、國家間的依賴與相互依賴。時代不同了。現在世界的事,對我們都是重要的。我們必須了解依賴與相互依賴對我們的工作、產品、服務、品質之影響;不然就難 免淪為次佳化。

所以研究各種專業或作為之間的「交互作用矩陣」 或「整合矩陣」是很重要的。


我們要了解統計資料的不同不確定性來源。資料是 如何獲得的?抽樣過程中是否有誤差?抽樣設計本身是否就有問題?在衡量或洽談中是否有瑕疪或錯誤?回答與不回答的誤差各為何呢?

對列舉研究及它在醫藥、醫療、化學、農業、森林 等行業的實驗測試而言,統計理論都很重要。不過,測試、實驗之結果的解釋,是屬於分析性問題。這涉及對流程或程序是否該有特定改變,或不改變等決策問題。

我們不妨舉下列的一些分析性研究重要面向:下表 綜述一些分析性研究的要點。
(取材R. D .Moen , T. W. Nolan , L. P. Provost (1991)〝Improving Quality Through Planned Experimentation〞p.54,並加以增添後四項)

除了製程的實驗設計之外,下述為分析型研究的一些例子:以福特公司的服務部門為例(Gitlow , et al , 1989 and 1995):服務作業根深柢固,盤根錯節,因為除以服務為主的行業外,其他業別如製造業等,幾乎也要龐大的服務作業來支援。

我們接上表中〝使用磁碟機之變異〞,採用另外一例來說明戴明分析性研究的方法。此個案取材自K. T. Delavigne和J. D. Robertson之〝Demings Profound Changes〞(pp.86-87)


科學家Sid Fernback是戴明在1944年戰時統計研習班的學生。他在30年後成為加州大學羅侖斯‧利威爾莫(Lawrence Livermore)國家實驗室的經理。Fernback記取戴明的原理教示,聘了統計學家佩里‧葛拉克曼(P. Gluckman,按:統計學顧問,本書作者的導師,作品為《每日英雄》)來改善其電腦中心的可靠性。

當時該實驗室擁有世界上三分之一的克雷式 (Cray)超級電腦,其程式設計人員有三千餘人。平均每四十五分鐘會有一次電腦停機,使得許多極機密的國防電腦化工作停擺。該中心忍受此種問題已達四 年,經佩里作停機帕累托(Pareto,按:重要的少數)分析,找出停機主因為儲存器故障。一旦把電腦的運轉及維修過程文書化後,它們就用一台迷你電腦, 把每次記憶器停機時刻,重要硬體的情報紀錄下來。如此他們就能針對停機之問題的運轉時間作管制圖分析,以找出故障的原因是共同因或特殊因。此舉更使他們改 變維修程序及做法,例如減少空氣中的灰塵量。他們也透過實驗了解記憶器對室溫極敏感,雖然他們已把溫度控制在製造商建議的範圍之內。根據此一新知識,他們 做了一些改善。

他們應用戴明的方法,使得該中心的可靠性,是全 國同類電腦中最高的,更顯著地改善了生產力。他們的分析更可協助做停機的預測,並加以防範未來。

成功的公司會持續地發展新的擴延知識,並加以修 正、評價。而組織中常見的舊有的抗拒變革(甚至敵對)方式必須先去除之。


田口損失函數的重要性可分兩方面來談。首先,它 告訴我們,品質不能以符合要求、規格或零缺點等方式來界定,不管是符不符合規格,都會有損失,而只能以降低變異的方式來減少損失。其次,可由田口損失函數 的斜率了解哪一品質特性特別要注意,即管理者的重點所在,先做最緊要的。

戴明學派把損失函數的應用,推廣到另一高度及涵 蓋面,如對室溫、火車等待、對待兒女夜歸問題等等(請參考Deming(1993)、W. W. S.(1991)、Joiner(1992)等)。


以漏斗實現為例。試有某製程,其固定因的變異是 不利的,但尚不嚴重,徒有好意圖而無知識,干預結果就堪虞(規則2,3,4,)。如果固定因的變異本身就很嚴重,那麼系統很快就會變得更糟。

譬如由工人教工人例子所產生的累積偏差,或是一 群不懂「淵博知識」的主管,想做好事,但是反而因干預而未果。同理,政府單位或委員會也是如此,擴大委員會不會有結果,因為它自然不具備〝淵博知識〞。


A公司引入B. L. Joiner所著的《第四代管理》及相關的錄影帶教學(1992)。由於Joiner為戴明學派的主要闡述者之一,本身又是個不錯的統計教師,所以在整個 課程中的「了解變異」及「各種共同因vs.特殊因的問題分析及解決對策」等方面,深入淺出,功力深厚。學習效果不錯。為了補強SPC的不足,該公司循著戴 明學派的建議,採用福特汽車的SPC中譯本為藍圖,由廠內人員負責教學,自行繼續耕耘下去。

不過,戴明很深邃的分析型研究洞識力(管制圖及結合固有智識為改善的根本),卻很難真正落實。此為一般人對SPC的了解不夠深入,而且很少結合固有之知 識。此為何戴明慨嘆Shewhart式管制圖理念,五十年來仍然很管用的,可惜它要深入人心,可能要再花上五十年。

我由於無法深入參與其改善之運作,而又想在戴明的基礎上建立人人可用的一般因策略方法學,所以幾經考慮,考量該公司在田口方法的訓練基礎及限制,以及J. M. Juran優秀的品質改善結構化流程(突破-三部曲),就決定採用中庸之道,即結合Motorola公司的〝應用診斷七大手法(也可稱之為〝實驗計劃七大 手法〞,為Shainin學派所創)〞及戴明學派的實驗計劃,加上Box. Hunter and Hunter發展一套適合一般用的診斷工具。

戴明博士在經營管理學的洞識力及哲學,應用在產官學可成一澎湃的趨勢。例如〝The Quality Yearbook (1997 Edition)〞中,都收有應用在政府、教育界的重要文章。
然而,他的核心思想,即變異觀是否可以進一步發展呢?就我所知,戴明學派在「損失函數」為顧客之聲(WWS, 1991)及個人生活(Joiner , 1992;Deming , 1993)上,都有些小突破。


另外,或許可以有人融合各家之長,寫出一本平衡而有深度、不誤導人的戴明學派實驗計劃課本。華人戴明學院除了預計出版戴明的主要統計論文選之外,也計劃出 版Moen等人的著作,藉以拋磚引玉,並引A公司的實驗計劃七大手法之實施印證。

參考資料(含戴明學派主要著作及出版的或預計的 中譯本):
Cortada and Woods (editors) , The Quality Yearbook , McGraw-Hill , 1997.
Deming , W. E.,Boundaries of Statistical Inference , Chapter 31.Johnson , N. L. and Smith , H (editors)〝New Developments in Survey Sampling〞John Wiley , 1969. 收入戴明博士文選 (2009)

Deming , W. E., Elementary Principles of the Statistical Control of Quality ,Second Printing , June 1952. Tokyo:Nippon Kagaku Gijutsu Remmei , 1951.收入戴明博士文選 (2009)

Deming , W. E.,Some Theory of Sampling , New York:John Wiley&Sons , 1950. Reprint Dover 1984.
Deming , W. E.,Out of the Crisis , Cambridge , MA:Massachusetts Institute of Technology , Center for Advanced Engineering Study , 1986. 轉危為安
Deming , W. E.,The New Economics , Cambridge , MA:Massachusetts Institute of Technology , Center for Advanced Engineering Study , 1993. 新經濟學
Gitlow , et al , Tools and Method for Quality Improvement , Irwin , 1989.
Gitlow , et al , Quality Management , Irwin , 1995.(本書為上述作品改版的另名。)

Joiner , Brian L., Fourth Generation Management:The New Business Consciousness ,McGraw-Hill , 1994.(中譯本名為《第四代管理》)

K. T. Delavigne and J. D. Robertson , Demings Profound Changes , rentice Hall , 1994.

D. J. Hand , Statistics and the Theory of Measurement , Journal of Royal Statistical Society , Vol.159 , Part 3 , 1996.

W. J. Latzko and D. M. Saunders , Four Days with Dr. Deming , Addison-Wesley , 1994.(中譯本名為《戴明博士四日談》)

Moen , Ronald D., Thomas W. Nolan , and Lloyd P. Provost , Improving Quality Through Planned Experimentation , New York:McGraw- Hill , 1991.

Neave , Henry R., The Deming Dimension , Knoxvile , TN:SPC Press , 1990. 1992二版 這本佳作詳細追博士的思想發展 尤其是 joy in the work和 sysyem of profound knowledge的發展 令人動容。

Scherkenbach , William W., The Deming Route to Quality and Productivity, Rockville , MD:CEE Press , 1986.(中譯本名為 《戴明修練I》)

Scherkenbach , William W., Demings Road to Continual Improvement , Knoxville , TN:SPC Press , 1991.(中譯本名為《戴明修練II》)

Shewhart , Walter A., The Economic Control of Quality of Manufactured Product , Princeton , N. J.:D. Van Nostrand Company , 1931. Reprinted by American Society for Quality Control , 1980.
Shewhart , Walter A. and W. E. Deming , Statistical Methods from the Viewpoint of Quality Control , Washington , D. C.:Department of Agriculture , 1939.

Wheeler , D. J. and D. S. Chambers , Understanding Statistical Process Control , Second Edition , Knoxville , TN:SPC Press , 1992.

Whitney , John O., The Trust Factor:Liberating Profits and Restoring Corporate Vitality , New York:McGraw-Hill , 1993.