We are very pleased to be able to announce a rather special series of articles especially written for the IQA by Professor Henry Neave entitled "SPC - Back to the Future". Henry Neave is the W Edwards Deming Professor of Leadership and Management in the Business School of the Nottingham Trent University, having previously lectured in Statistics at the University of Nottingham. Last year Henry received the American Society for Quality's Deming Medal and was also made one of the first Companions of the IQA. In November he was the IQA's John Loxham lecturer. During the final nine years of Dr Deming's life, Henry was his British assistant at all of his four-day seminars and other events held in the UK and elsewhere in Europe. Henry learned about SPC directly from Dr Deming, who in turn had learned it straight from its creator, Dr Walter Shewhart- and you can't get much closer to the "horse's mouth" than that! Henry has written the essential introductory volume on Dr Deming's work entitled The Deming Dimension.
The eight articles will be released on a monthly basis. Most of them are for everybody - from novice to expert, and everyone in between. Just a couple near the end of the series are more for the experts - whom we'd better warn: watch out for some shocks! There are also a couple of the best case studies that we've ever seen.
To set the scene, here are some comments written by Bernard Patmore FIQA after attending one of Henry's courses.
SPC. Statistical Process Control, or Confusion? SPC is an acronym that brings fear to the minds of most people. For it involves maths - and many people have a mental block about maths. So who could possibly like SPC other than statisticians - and sadists?
During my 40 years in quality and 19 years as a consultant, one thing I've learned is that "simplicity is best". Everyone understands simple words and numbers. But SPC isn't simple, is it? Is it?
While studying at Nottingham Trent University, I attended a four-day seminar as part of my MSc in Quality Management. The seminar was called "The Deming Management Approach", and the presenter was Professor Henry Neave - "Our Henry" as I now call him when talking with other colleagues.
Before he began, I told Henry that I wasn't sure whether I would get much out of the seminar: I was a "Juran man". That didn't phase Our Henry: he just smiled - knowingly. Yes, I was wrong.
And then he started talking about SPC. OK, I thought, time for a nap - I know all that. Wrong again. He told us that we experienced people had probably not been doing SPC as Shewhart (who invented it) had intended. Not being a fan of SPC, I wondered what was coming.
One of the first things he showed us was that I (and most of the others) had been calculating control limits wrongly. And he showed us why—in about 10 minutes! I'd been doing it wrong for over 30 years.
There was much more. He showed us that SPC is not just for the "experts" or the mathematicians. It's kind of easy to understand and to do - it's something for everybody, not the elite.
This led to some lively debate, which I enjoyed. So did Our Henry.
Being Bernard, I thought "Let's throw a boulder in the pond and see how far the ripples roll." So I asked Our Henry to write some articles for the IQA. He obliged. And how! I hope you enjoy them as much as I have.
Bernard Patmore FIQA
Finally, we asked Henry why he has used the title "SPC - Back to the Future":
"At least three answers to that question. First, there's the obvious purpose of SPC, which is going Back to the past to find and interpret information to help us in the Future. Second, going Back to better understand why Shewhart invented SPC, so that we can gain greater benefit from it in the Future. And third, also going Back to discover how and why both Shewhart and Deming rejected much of the complexity with which most "experts" surround SPC, so that many more people can really understand it and enjoy using it and communicating with it in the Future."
1. An 11-year-old can do it!
SPC?
"SPC? Oh, Statistical Process Control. Nothing to do with us. We're not in manufacturing."
"SPC? Oh yes, I think the shop-floor does some of that. Nothing to do with me: I'm a manager."
"SPC? Oh, I don't understand anything about that. I just collect the numbers and pass them on to Quality Control."
"SPC? Oh, that's not for me. I'm no mathematician."
"SPC? Oh, no way. I don't trust statistics."
"SPC? Oh yes, of course. I'm a professional statistician. It's quite simple really. But you have to be sure that your data are normally distributed, else it's not valid."
Six responses. Six sad responses.
11-year-old Patrick
A story which the American management teacher Dr W Edwards Deming was fond of relating during his celebrated four-day seminars concerned 11-year-old Patrick Nolan (Neave, 1990a: 393-395). Day by day, Patrick recorded the time of arrival of his school-bus and plotted it on a chart.
Figure 1 shows Patrick's chart as drawn by Dr Deming (reproduced from a roll of overhead projector transparency after one of his seminars). The times of arrival varied, of course (else the points would just have formed a straight line). But most of the variation was effectively random - "chance"- within certain limits. Deming had drawn in such limits. He described that variation as being due to common causes.
Figure 1
There were however two exceptionally late arrival times, well beyond those limits, which Patrick had circled and explained. Deming referred to such occurrences as special causes. Interestingly, he also circled the earliest and latest times which lay between the limits, and pointed out that, quite rightly, Patrick had made no effort to explain those. Chance variation also has its low and high points, but who could "explain" them?
"If 11-year-old Patrick could understand this, what's your excuse?"
My school-bus
Dr Deming's story about Patrick rang a bell*. Throughout my schooldays, I lived out in the countryside, several miles from where I went to school. School started at 9.00 each morning.
* I am indebted to David Young for reminding me of this thought. He has given a similar account in his guide for Rolls-Royce Aerospace: Simple Process Control.
I also travelled to school by bus. The bus would usually arrive at the bus-stop near my home some time between 8.25 and 8.35, though occasionally a little earlier or later. Now and again, like Patrick's, it was very late (due to a special cause), e.g. because it had broken down or had been delayed by an accident. Except on such rare occasions, it would always get me to school on time. But I was not happy!
To be fairly sure of catching the bus, I had to be at the bus-stop by 8.25. To be really sure of catching it, I had to be there by 8.20. But much of the time I'd then be waiting 10 to 15 minutes, occasionally longer—often when the weather was cold and wet. (Sympathy, please: this was a bus-stop, not a bus-shelter!) So I would often be soaking wet and/or freezing cold by the time the bus arrived. Not a good start to the day!
If only the (common-cause) variation in those bus-arrival times could have been smaller! If only the bus could have arrived within, say, one minute either side of 8.30, rather than within 5 or even 10 minutes. Or, indeed, one minute either side of 8.25, or one minute either side of 8.35 - or one minute either side of any suitable average time of arrival. Then I could have arranged my mornings much more efficiently - and, with rare exceptions, suffered no more than a two-minute soaking!
Thus I learned at an early age that variation affected my quality of life. The variation was actually more important than the average time of arrival. The greater the variation, the more I risked either missing the bus altogether or getting wet through.
Variation is the enemy of quality.
How it all began
In the early 1920s, people in the Western Electric Company were hard at work trying to improve telephone technology and associated equipment. For a while they made great progress. But then the rate of progress slowed. They were still trying as hard, if not harder than before. They were still pouring time and money - and probably emotion - into the improvement effort, but somehow it just wasn't working any more. They were experimenting and analysing and trying to interpret data in just the same ways as before. Those ways had previously reaped great rewards. But no longer. Increasingly, not only were they failing to improve: they were beginning to make things worse rather than better! That is when they invited Dr Walter Shewhart to help them.
Now we'll let Dr Deming take up the story (transcribed directly from a presentation to an audience in Versailles, France on 6 July 1989) (Neave, 1990b: 2-3):
"Part of Western Electric's business involved making equipment for telephone systems. The aim was, of course, reliability: to make things alike so that people could depend on them. But they found that the harder they tried to achieve consistency and uniformity, the worse were the effects. The more they tried to shrink variation, the larger it got. When any kind of error, mistake or accident occurred, they went to work on it to try to correct it. It was a noble aim. There was only one little trouble. Things got worse.
Eventually the problem went to Walter Shewhart at the Bell Laboratories. Dr Shewhart worked on the problem. He became aware of two kinds of mistakes:
1. Treating a fault, complaint, mistake, accident as if it came from a special cause when in fact there was nothing special at all, i.e. it came from the system: from random variation due to common causes.
2. Treating any of the above as if it came from common causes when in fact it was due to a special cause.
What difference does it make? All the difference between failure and success.
Dr Shewhart decided that this was the root of Western Electric's problems. They were failing to understand the difference between common causes and special causes, and that mixing them up makes things worse. It is pretty important that we understand those two kinds of mistakes. Sure we don't like mistakes, complaints from customers, accidents; but, if we weigh in at them without understanding, we only make things worse. This is easy to prove."
How did Mr Deming (as he then was) learn about this? By great good luck! At the time, he was studying for his PhD in Mathematical Physics at Yale. Just like most students these days, he was having to "work his way through college", i.e. earn money to support himself. For this reason, he took summer vacation jobs in 1925 and 1926 - at the Western Electric Company. He just happened to be there at the right time. How fortunate! For it was Shewhart's breakthrough in this new understanding of the types and causes of variation that proved to be the launchpad for W Edwards Deming's extraordinary life's work.
Apart from introducing some of the basic concepts, this early piece of history is important in emphasising that the environment and purpose in and for which SPC was created was one of improvement. Shewhart invented the control chart to provide guidance on the types of action most likely to bring about improvement and warnings on the types likely to do harm.
Interestingly, Dr Deming first used the terms "common cause" and "special cause" not in connection with control charts but whilst discussing prison riots (Deming, 1986: 314-315)! Did something special occur to spark off a riot? Or was it due to the procedures, the environment, the morale of both the prisoners and the prison staff, the way the staff treated the prisoners, etc.? That is, was the common state of affairs (which Deming would refer to as the system) in the prison such that riots would be bound to occur from time to time? Or would it take something special?
Learning And Unlearning
We're already two-thirds through this article. So I can probably now risk confessing my origins without too much fear of frightening off those readers who have come this far!
I began my career life as a conventional mathematical statistician. I learned conventional mathematical statistics as a student, and then, as a Lecturer, I taught what I had learned.
In my own defence, I had some reservations. These led me to dabble in areas of the subject regarded by the purists as slightly unconventional. But I had neither the wit nor the courage to dip more than my toe in the water.
So then came my stroke of great good luck! I was singularly fortunate in the early 1980s to become involved with the British subsidiaries of the first American company to start taking Dr Deming's work at all seriously. As I soon discovered, this was some 30 years after the Japanese began to learn from him, and instituted their famous national award for quality in his name (Figure 2).
FIGURE 2 The Deming Medal
But I was puzzled. I was told that Dr Deming's work was based on statistics - my chosen career path. Yet, search as I might, I could find little in his work that had any relationship with the subject on which I had been building my career and reputation.
Despite that continuing perplexity, in 1985 I received an invitation to assist Dr Deming at his first four-day seminar to be held in Britain. I enjoyed that same privilege and responsibility during all of his visits to Europe throughout the remaining nine years of his life.
I looked forward eagerly to that first four-day seminar. Now I would at last learn the truth about where my great knowledge of mathematical statistics would fit into it all!
Wrong again! Granted, there was some stuff about collecting and analysing data. But it really was rather disappointing. The only technique he ever seemed to use was the control chart - and he didn't even do that right! Where were the probabilities, the normal distributions, the Central Limit Theorem, the action and warning limits? You see, for years I had covered control charts in my 60-lecture first-year course at the University: I'd probably spend one lecture, maybe even two, on it. Seemed rather dull, really - nothing more than a slightly glorified significance test. And this was all he was using - and, worse still, without any of the clever mathematics?
Yes, unlearning is much more difficult—and painful—than learning.
Words of wisdom…
…from Dr Shewhart, who created the subject…
"The fact that the criterion which we happen to use has a fine ancestry of highbrow statistical theorems does not justify its use. Such justification must come from empirical evidence that it works. As a practical engineer might say, the proof of the pudding is in the eating." (Shewhart, 1931: 18)
"Some of the earliest attempts to characterise a state of statistical control were inspired by the belief that the normal law characterised such a state. The normal law was found to be inadequate: all hopes [for such an approach] are blasted." (Shewhart, 1939, p12)
…and from Dr Deming—surely Shewhart's most famous protégé!…
"It would be wrong to attach any particular figure to the probability that a statistical signal for detection of a special cause could be wrong, or that the chart could fail to send a signal when a special cause exists. The reason is that no process is steady, unwavering."(Deming, 1986: 334)
"It is true that some books on the statistical control of quality and many training manuals for teaching control charts show a graph of the normal curve and proportions of area thereunder. Such tables and charts are misleading and derail effective study and use of control charts." (Deming, 1986: 335)
"It is nothing to do with probabilities. No, no, no, no: not at all. What we need is a rule which guides us when to search in order to try to identify and remove a specific cause, and when not to. It is not a matter of probability. It is nothing to do with how many errors we make on average in 500 trials or 1000 trials. No, no, no - it can't be done that way. We need a definition of when to act, and which way to act. Shewhart provided us with a communicable definition: the control chart. Shewhart contrived and published the rules in 1924. Nobody has done a better job since." (Neave, 1990b: 4)
…and from the Japanese, who both learned and unlearned…
"The ease with which [Dr Deming] was able to speak in simple terms was admirable. He showed that quality control is not exclusively for those who are strong in mathematics. The usual diffidence of technicians who lack mathematical knowledge, but should be the ones actually in charge of quality control, has been completely wiped away." (JUSE, 1950)
"Prior to Deming's visits in the early 1950s, Japanese quality control had been butting its head against a wall created by adherence to difficult statistics theories. With Deming's help, this wall was torn down." (Noguchi, 1995: 35-37)
…and finally, from Dr Deming's lecture-notes in Japan in 1950:
"The control chart is no substitute for the brain."
References
(Some of the quotations are abbreviated, but without losing their original sense.)
Deming, W Edwards (1951), Elementary Principles of the Statistical Control of Quality, Nippon Kagaku Gijutsu Renmei, Tokyo.
Deming, W Edwards (1986), Out of the Crisis, Massachusetts Institute of Technology, Center for Advanced Engineering Study.
Neave, Henry R (1990a), The Deming Dimension, SPC Press, Knoxville, Tennessee.
Neave, Henry R (1990b), Profound Knowledge, British Deming Association Booklet A6.
Noguchi, Junji (December 1995), "The Legacy of W Edwards Deming", Quality Progress, vol 28, no 12.
Shewhart, Walter A (1931), Economic Control of Quality of Manufactured Product, van Nostrand, New York.
Shewhart, Walter A (1939), Statistical Method from the Viewpoint of Quality Control, Graduate School of the Department of Agriculture, Washington.
Statistical Quality Control (August 1950), JUSE, Tokyo, vol 1, no 6.