Professional Documents
Culture Documents
Daniel Sloan is an internationally recognized Six Russell Boyles earned his PhD in Statistics at the
Sigma Master Black Belt and an ASQ certified Black University of California, Davis. He subsequently
Belt. His 16 years of experience have been spent two years in the Applied Mathematics Group
distinguished by Six Sigma seminars in Mexico, at Lawrence Livermore National Laboratory, two
Uruguay, Brazil, Australia, and 47 of the United States. years as Director of Statistical Analysis for NERCO
McGraw Hill and Quality Press published five of his Minerals Company, and eight years as Statistical
7 books. As a Senior Vice President of Applied Process Control Manager at Precision Castparts
Business Science for a $500 million company, he led Corporation. As a trainer and a consultant, Russell
their Six Sigma initiative. With "factory floor" Six specializes in Six Sigma Master Black Belt and
Sigma successes ranging from non-woven fabrics, Black Belt certification courses, Design of
extruded products, medical equipment, aerospace Experiments, Gage Studies, Reliability and
engineering, automotive parts, to Internet router Statistical Process Control. A few of his recent
production and health care, Daniel has a proven track papers have appeared in ASQ publications
record in helping companies produce bottom line results. Technometrics and Journal of Quality Technology.
Master Black Belt, Black Belt, Green Belt, Data mining, strategic Information Systems design.
Champion, and Senior Executive certification
training for all industries including manufacturing, Bottom-line business results project coaching.
financial services, and health care.
Consulting support to private industry, government
Consortium Six Sigma events for small companies and academic institutions that are implementing
who wish to pool resources. evidence-based decision systems.
Custom designed training events and multi-media, Custom designed training events and multi-media,
evidence-based Six Sigma materials. evidence-based education and training materials.
http://creativecommons.org/licenses/by-nc-nd/2.5/
http://creativecommons.org/licenses/by-nc-nd/2.5/legalcode
Corporate Headquarters
Creative Commons
543 Howard Street
5th Floor
San Francisco, CA 94105-3013
United States
Profit Signals
How Evidence-based Decisions
Power Six Sigma Breakthroughs
By
All rights reserved. No part of this book may be reproduced in any form or by any means,
electronic, mechanical; photocopying, recording, or otherwise, without the prior written
permission of the publisher. Your support of author’s rights is appreciated. For permissions the
authors can be contacted directly.
Trademark Acknowledgements
Profit Signals® and the phrase “Vector Analysis Applied to a Data Matrix®” and the Profit
Signals tetrahedron on the book’s cover are registered trademarks of Sloan Consulting. LLC,
Six Sigma® is a registered trademark and service mark of Motorola, Incorporated. Sculpey
Clay® is a registered trademark of Polyform Products Co. Excel® is a registered trademark of
Microsoft. Other copyright notices are listed in the production notes at the end of the book.
Illustrations: Cover, Robin Hing. Tables and illustrations, Robin Hing, Russell A. Boyles, M.
Daniel Sloan, John Pendleton, Austin Sloan, and Alan Tomko. Netter illustrations used with
permission from Icon Learning Systems, a division of MediMedia USA, Inc. All rights reserved.
The book’s design and layout, using Adobe InDesign 2.0.2, were completed by M. Daniel
Sloan. Printed in the United States of America.
Premise ................................................................... 9
The Parable of the Paper Bags ........................... 14
The Dollar Value of Evidence............................ 16
Six Sigma ......................................................... 18
How to Read This Book .................................. 20
Endnotes .................................................................24
Index ...................................................................255
P
rofit Signals is a guide for using evidence to make
better, more profitable business decisions. Face
value judgments, opinions, gut feelings, suspicions,
circumstance, and superstitions are not pathways to evidence.
Measurements are. This book will show you how to turn
measurements into evidence and evidence into profit.
���� ���������������
�� �����������������
��� ged
�� era e)
v
(A com ����������������
In
Negotiate and get to “Yes” with your peers. Get to yes with
your executives and those you lead so that your company
is not using brown paper bags to compete against a more
powerful, efficient, effective, and profitable way of doing
work.9 That way of doing work is a vector analysis applied to a
data matrix.
You are probably reading this book because you have made
business decisions. Some of those decisions were good. Some
were bad. Some were based on evidence; others were not. We
ask you to contrast the profit related to good decisions with
the loss related to bad ones. The difference between these two
numbers forecasts the initial Return on Investment (ROI) you
can expect from reading this book.
Six Sigma
The iterative nature of the Six Sigma project cycle has taught
us which parts of Six Sigma are essential. We, and other
experienced professionals in the field, also have learned
which parts are extraneous. The demand for additional,
rapid, dramatic breakthroughs can be satisfied only if we trim
fat from Six Sigma’s middle-aged spread. Six Sigma made
Profit Signals possible. We return the favor by showing how
to flex the evidence muscle without carrying the weight of
bureaucracy.
Anyone and everyone can learn this unifying theory. They can
learn it quickly. Anyone can master what is called the Black
Belt Body of Knowledge (BOK). Based on our experience,
and with the support of senior management leadership, this
process can be accomplished in 10 to 16 days.
You can speed read this book in about a week. To get the “big
picture” quickly, skim the illustrations. Read the captions to
these exhibits. “Closing arguments” at the end of each chapter
summarize the key content.
Are these new and different? No. They are just other ways
of using the New Management Equation. They are vector
analysis applied to a data matrix.
Endnotes
1
Box, Joan Fisher. R.A. Fisher, Life of a Scientist. John
Wiley & Sons, New York. 1978.
2
Box, Joan Fisher. R.A. Fisher, Life of a Scientist. John
Wiley & Sons, New York. 1978.
3
Harrison, G. Charter. Cost Accounting to Aid Production
– I, Standards and Standard Costs, Industrial Management,
The Engineering Magazine, Volume LVI, No. 5, October,
1918.
4
Harrison, G. Charter. Cost Accounting to Aid Production –
II, Standards and Standard Costs, Industrial Management,
The Engineering Magazine, Volume LVI, No. 5,
November, 1918.
5
Harrison, G. Charter. Cost Accounting to Aid Production –
II, Standards and Standard Costs, Industrial Management,
The Engineering Magazine, Volume LVI, No. 5,
December, 1918.
6
Johnson, H. Thomas, and Kaplan, Robert S. Relevance
Lost, The Rise and Fall of Management Accounting.
Boston, Harvard Business School Press, 1987.
7
Garrison, Ray H. and Noreen, Eric W. Managerial
Accounting, 10th Edition. Boston, McGraw-Hill Irwin,
2003. Page 431.
8
Anthony, Robert N., and Reece, James S., Accounting:
Text and Cases, Eighth Edition. Homewood, Irwin, 1989.
Page 941.
9
Fisher, Roger, and Ury, William, Getting to Yes,
Negotiating Agreement Without Giving In. New York,
Penguin Books. 1981.
10
Royall, Richard. Statistical Evidence, A likelihood
paradigm. New York, Chapman & Hall. 1997.
11
Shewhart, Walter A. Nature and Origin of Standards
of Quality. The Bell System Technical Journal. Volume
xxxvii, number 1, January, 1958.
12
Six Sigma is a registered trademark and service mark
of Motorola Incorporated. The Motorola web site is a
recommended resource for researching this history of Six
Sigma. For a summary overview please read: Barney, Matt,
“Motorola’s Second Generation,” Six Sigma Forum Magazine,
May 23, 2002, pages 13-16. http://mu.motorola.com/pdfs/Mot_
Six_Sigma.pdf
14
Sloan, M. Daniel and Torpey, Jodi B. Success Stories
on Lowering Health Care Costs by Improving Health Care
Quality. Milwaukee, ASQ Quality Press. 1995.
The Five-
Minute PhD
P
hD’s, medical doctors, scientists, engineers,
mathematicians, statisticians, economists, managers,
and executives don’t own the lock and key to data
analysis. Anyone can learn to do a vector analysis. What was
once the high water mark of postgraduate study is now as
simple as a Google web search. You don’t need a certificate on
your wall to analyze data.
For convenience, each factor was set at only two levels. The
high setting is coded +1. The low setting is coded -1. Table
1 contains all eight possible combinations of a three-factor
experiment with two levels for each factor. This is called a 23
(two raised to the third power) experimental design. “Two
raised to the third power” is a mouthful, so it is usually
pronounced, “two cubed.” Think of tea with two cubes of ice,
rather than an equation, and the idea will be more refreshing.
��
�����
The differences in appearance
�����
���
illustrations were created using
vectors. ������ ������ �������
Activity
�
RAW DATA
Figure 6 A long, strong Noise
vector and a short, weak Profit
N
NOISE
IATI
IT
R OF AL
P GN
SI
Profit Signals
Only a few years before Fisher used the cube and higher-
dimensional experiments to dramatically increase profitable
crop yields in England, Pablo Picasso and George Braque
created a new art form called Analytic Cubism. The analogies
between Picasso’s and Fisher’s cubes are intriguing.
Figure 10 Perpendicular
planes representing the eight-
dimensional profit signal vector for
the interactive effect of factors X
and Z.
Figure 11 Perpendicular
planes representing the eight-
dimensional profit signal
vector for the interactive
effect of factors X and Y.
Figure 12 Perpendicular
planes representing the eight-
dimensional profit signal vector
for the interactive effect of
factors Y and Z.
Data Recycling
For example, have you noticed that all the corner values are
used repeatedly? Every data point appears six different times,
once in each of the six profit signal vectors shown above! This
is a lot of work for only eight little numbers to do.
The larger the number of factors, the greater the savings. This
bottom-line result is enhanced by the fact that vector analysis
gives you the right answers to your most pressing business
problems.
Table 6 Like any true the principles of evidence-based decisions. Yes. This is yet
Generalizaion, the cube another paradox in evidence-based decisions.
experiment is a Law of the
Universe. Have some fun
practicing with your new PhD in After you have completed your experiments with family,
universes of your own. friends, and colleagues at work, discuss the implications
of these analogies. Since you are now a PhD, feel free to
throw around phrases like “Hegelian Dialectic” during your
conversations. This crowd-pleaser will let other doctors of
philosophy know that you know what you are talking about.
Closing Arguments
Endnotes
1
These factors are also commonly known as independent
variables.
2
Box, Joan, Fisher. R.A. Fisher, Life of a Scientist. New York,
John Wiley and Sons, 1978.
3
Netter, Frank. The CIBA Collection of Medical Illustration,
Volume 5, Heart. Commissioned by CIBA. 1969.
4
Netter, Frank. The CIBA Collection of Medical Illustration,
Volume 5, Heart. Commissioned by CIBA. 1969.
5
Dubin, Dale. Rapid Interpretation of EKG’s, Edition V.
Tampa, Cover Inc., 1996. Page 4.
6
http://www.rothamsted.bbsrc.ac.uk/pie/sadie/reprints/
perry_97b_greenwich.pdf
7
http://www-gap.dcs.st-and.ac.uk/~history/
Mathematicians/Descartes.html
8
http://www.ibiblio.org/wm/paint/tl/20th/cubism.html
9
http://www.artchive.com/artchive/P/picasso_
analyticalcubism.html
10
Inscription on the southern ceiling of the rotunda leading
to a James Turrell Skyspace installation at the Henry Art
Gallery on the University of Washington campus.
11
Einstein, Albert. Relativity, The Special and General Theory,
A Clear Explanation that Anyone can Understand. New York.
Crown Publishers, 1952. Page 32.
18
Einstein, Albert. Relativity, The Special and General
Theory, A Simple Explanation that Anyone Can
Understand. New York, Crown Publishers, 1952. Page 90.
19
http://www.pbs.org/art21/artists/turrell/
http://goflorida.about.com/library/bls/bl_wdw_
20
waltdisney_quotes.htm
Standards of
Evidence
W
hat are the objective standards of evidence your
business uses to make decisions? We ask all new
clients this question. Too often the answer is an
uncomfortable silence or, “We’ve never asked ourselves that
question before.”
Stories are not laws. They do not, and are not intended to,
reliably describe historical facts or physical realities. Story
telling does have its place, but it can be at odds with science.
Story telling often involves tales of trial and error.
“Scientific” Management
� � � �
Table 2 A spreadsheet consists of �� �� ��
rows and columns. The cells can �
contain text, numbers, graphics, � � � �
symbols or formulas. There are no �� ������ ������ ������
rules governing the interpretation �
of rows and columns. There are � � � �
no laws for arranging or analyzing �� ������ ������ ������
data. �
� � � �
�� ������ ������ ������
�
�
Ironically, Fisher developed the Analysis of Variance in a farm
near London right around the same time Taylor and Harrison
were promoting their “scientific” management and cost-
accounting principles.
� � �
Table 3 Like a spreadsheet, a ��������� ���������
data matrix consists of rows and
columns. The rows of a data matrix
����������� �����������
represent records—the individual ����� �����
objects or events we have data on. � �
The columns represent fields—the � � �
variables for which we have data. �������������� �� ���
Each stack of numbers in a data
matrix column is a vector.
�
� � �
�������������� �� ���
�
�
For example, the data vectors in Table 3 are (3, 4) for Variable
1 and (5, 2) for Variable 2. These vectors are plotted in Figure
2 . The two coordinate axes correspond to the two objects.
It does seem coincidental that (3, 4) and (5, 2) have the same
average, but we did this on purpose. So, how do these vectors
differ?
(D1)2 = A2 + (V1)2
25 = 24.5 + (V1)2
This final number, 0.71, the length of the variation vector for
Variable 1, is called the sample standard deviation for Variable
1.
4.55 = (V2)2
Degrees of Freedom
Now let n stand for the number of objects in your data set.
This is the same as the number of rows in your data matrix. It
is your sample size. All the vectors are now n-dimensional.
Table 5 lays out the basic vector calculations for the sample
standard deviation, s. In this case s = 0.14. This is a vector
analysis in four-dimensional hyperspace, because there are
four data points.
= SUMSQ(cell range)
All four data points lie within two standard deviations of the
mean. We must conclude that the deviations from the mean
© M. Daniel Sloan and Russell A. Boyles, All Rights Reserved, 2003
Standards of Evidence 71
value are a result of natural, or Chance, variation. There is
certainly no evidence of significant differences among these
totals.
���������������������
�������
Figure 9 The monthly revenue
numbers plotted in time sequence. ������� �������
�������
�������
�������
�������
�������
�
�
�
��
��
��
��
��
��
��
��
��
������
����������� �����������
��� ���� ��������� ���� ���������
��� ���� ����� ����� �����
����� ���� ���� ����� ����
��� ����� ����� ���� �����
������ ����� ����� ����� �����
������� ����� ����� ���� �����
Figure 10 Wear rate data as arrayed ���� ���� ���� ����� ����
��� ���� ���� ����� ����
in Excel. ����� ����� ����� ����� �����
������ ����� ���� ���� ����
���� ���� ����� ����� �����
��������������������� ������ ������
����������������������� ����� �����
���������� ����
������������������
���������������������������
���������� �����
Ten boys were enlisted for the test. Each boy wore one shoe
made from Material A and one from Material B. Coin tosses
were used to randomly assign Material A to the left or right
foot for each boy.
The average wear rate for Material B comes out 0.41 units
higher than for Material A, an increase of 3.86%. Given
the price difference between the two materials, the manager
concludes that the difference in durability is irrelevant.
��������������������
Figure 11 Wear rate data as analyzed
by a spreadsheet bar graph. �����
�����
����
����������
����
� � � � � ����������
����
������������
�������������
Figure 12 Frequency histogram of
differences in wear rate (B minus A).
The next Black Belt, Green Belt, Yellow Belt and Champion
courses are filled to capacity. The waiting lists for the
following sessions are long. The company takes the next step
forward by implementing Six Sigma across all projects and
functional responsibilities in the corporate matrix. Their first-
wave Black Belts are now in Master Black Belt training using
their own case studies.
= FDIST(11.215, 1, 9)
= FDIST(2.843, 1, 2)
Closing Arguments
1
American Heritage Dictionary of the English Language, Third
Edition. Boston. Houghton Mifflin Company. 1992.
2
Dawkins, Richard. Unweaving the Rainbow, Science Delusion
and the Appetite for Wonder. Boston, Houghton Mifflin
Company, 1998.
3
Huff, Darrell and Geis, Irving. How to Lie with Statistics.
New York, W.W. Norton and Company. 1954.
4
Taylor, Frederick Winslow. Scientific Management, Mineola:
Dover Press, 1998. pages 55-59. The original 1911 version
was published by Harper and Brothers, New York and
London..
5
Oxford English Dictionary, 1989.
6
Garrison, Ray H. and Noreen, Eric W. Managerial
Accounting, 10th Edition. Boston, McGraw-Hill Irwin, 2003.
Page 431.
7
Harrison, G. Charter. Cost Accounting to Aid Production
– I. Application of Scientific Management Principles. Industrial
Management, The Engineering Magazine, Volume LVI, No.
4, October 1918.
8
Harrison, G. Charter. Cost Accounting to Aid Production
– I, Standards and Standard Costs, Industrial Management,
The Engineering Magazine, Volume LVI, No. 5, November,
1918.
9
Harrison, G. Charter. Cost Accounting to Aid Production
– I, The Universal Law System. Industrial Management, The
Engineering Magazine, Volume LVI, No. 6, December, 1918.
10
Johnson, H. Thomas, and Kaplan, Robert S. Relevance
Lost, The Rise and Fall of Management Accounting. Boston:
Harvard Business School Press 1991. Pages 10-12.
11
Anthony, Robert N., and Reece, James S., Accounting: Text
and Cases, Eighth Edition. Homewood, Irwin, 1989. Page 15.
13
MacKay, Charles, Memoirs of Extraordinarily Popular
Delusions, Copyright 2002 eBookMall version available for
$1.75. http://www.ebookmall.com/alpha-authors/m-authors/
Charles-MacKay.htm Page 8.
15
Sagan, Carl. The Demon Haunted World, Science as a Candle
in the Dark. New York, Ballantine Books, 1996. Page 241.
16
Anthony, Robert N., and Reece, James S., Accounting: Text
and Cases, Eighth Edition. Homewood, Irwin, 1989. Page
941.
17
Oxford English Dictionary, 1989.
18
Box, Joan Fisher. R.A. Fisher: The Life of a Scientist. New
York: John Wiley and Sons, 1978. Page 97.
19
Box, Joan Fisher. R.A. Fisher: The Life of a Scientist. New
York: John Wiley and Sons, 1978. Page 100-102.
20
Box, George E.P., Hunter, William G., and Hunter, J.
Stuart. Statistics for Experimenters, An Introduction to Design,
Data Analysis, and Model Building. John Wiley & Sons. New
York. 1978.
21
Dilson, Jesse. The Abacus, The World’s First Computing
System: Where it Comes From, How it Works, and How to Use it
to Perform Mathematical Feats, Large and Small. New York, St.
Marten’s Press. 1968.
22
http://www.wikipedia.org/wiki/Straw_man
23
http://www.commonlaw.com/Justice.html
Evidence-based
Six Sigma
S
ix Sigma (6σ) is a proven, pursuit-of-perfection business
initiative that creates breakthroughs in profitability,
productivity, and quality. It is a highly structured,
project-by-project way to generate bottom line results. It
produces significant dollar value through a never-ending
series of breakthrough projects. Evidence-based decisions
characterize the 18-year, 6σ record of accomplishment.
Figure 5 This flow chart has guided projects toward bottom line business results for years.
�������������������������������������������
����������
��������� ������������
���������
���������� �������������� ����������
������������
������������� ����
��������� ��������������� ��������
��������� ������������
��������������� ��������������
���������
�������������� ����������������
�������
� ���������
������������ ����� ������������
������� ����������� �����������
������������ ������
�������������
�������� ��������
��������� ���������� ��������������
������ ����� �����������
����������
����� ����������� ������������
�������� ����
������� �������������
��������
�����������
���������
������ ��������� ��������
���������� ������������ ��������
��������������
������� ��������� �������������
�������������
������ ������������
������������������������� ���������
���������������������� �������
������������������
�������� �������������
���������������
�������
The series of five steps in the top row and the two final steps
in the bottom row are top-level management and leadership
responsibilities. The middle three levels are Black Belt project
tasks. Each of these steps takes time, so every 6σ project result
needs to be substantial and financial.
How many did you count? Pause to write your answer here
before moving on. _______
Process Maps
Figure 9 Black Belts use personal interviews, first hand observations, and measurements to complete
this map. Drawing these maps from the end to the beginning is the best way to produce a meaningful
SIPOC map showing all the relationships.
�������
��������� �������� ��������������������
���������� �����������
���������������
�����������
������ �� ���� ����
�������
����������
������������
����� ����
�������
������� ������������
����������
�������������
�����������
������
�������
�����
��������� ��������� ��������
����������
� �������
��� �������
���������� ��������� �������
Figure 10 The hidden factory of
rework in this map includes Processes �����
4-6 and the related delay.
��
���������
���������
���������
���������
�����������������
������������
�������������������� �������� ���������
����������� ���������
���������� ��������� ���������
With the lean Six Sigma strategy in place a second can be, and
often is, literally worth thousands of dollars. For example, in
one San Jose Internet router factory a 2 foot by 2 foot by2
foot pile of scraped motherboards was time-valued at more
than $6 million dollars.
Like a vector analysis, those who are familiar with lean tools
do not argue against them. To do so would be as foolish as
arguing against the speed of light, the existence of gravity, or
the impact of variation on measurement.
�������������������������
������������������������
������������������
�������
We will use dice rolls for our example. (Chapter 7 will extend
this experiment to include 4 die.) Feel free to roll your set
of dice until you get 5,000 measurements. Or, you can
accurately simulate the outcome of 5,000 rolls in a minute
using software. Comparing both methods will give you a good
feel for the value of Six Sigma vector analysis software.
Let’s game this system and improve our Cpk by setting our
LSL and USL perfection expectations at –10 and 30. Figure
14 shows that our process is a smoking Six Sigma process
fully capable of producing perfect quality outcomes 99.99999
percent of the time! We’re in the money. Note how tightly the
distribution curve is centered on the target of 7.
In real life, Six Sigma companies earn “high” Cpk values not
by lowering their standards, but by raising them relentlessly,
geometrically, and exponentially.
© M. Daniel Sloan and Russell A. Boyles, All Rights Reserved, 2003
114 Evidence-based Six Sigma
�������������������
���� �� ������� ����� �� �������
��� ������������� �������� ����� ����
���������������� �������� ���������
Figure 13 This process is not capable ��������
��������������� �����������
of perfection. Its Cpk value is only
0.640. ����������
������������
��� ���
� ��
�������������������
���� �� ������� ����� �� �������
��� ������������� �������� ����� ����
���������������� �������� ���������
Figure 14 A Six Sigma process will
produce perfection every time.
��������������� �����������
��
�����������
��� ���
� ��
1
Box, George E.P., Hunter, William G., Hunter, J. Stuart.
Statistics for Experimenters, An Introduction to Design, Data
Analysis, and Model Building. New York. John Wiley & Sons,
Inc. 1978. Pages 170- 201.
2
The body of knowledge that is widely regarded as the most
comprehensive is posted by the American Society for Quality
http://www.asq.org/cert/types/sixsigma/bok.html
3
Mikel Harry, a popular leader in the Six Sigma field,
reported this history on a video tape recorded in 1995.
4
Shewhart, Walter. Economic Control of Quality of
Manufactured Product. Brooklyn, D. Van Nostrand Company,
Inc. 1931, page 5.
5
Cohen, J. Bernard. Revolution in Science, Cambridge, 1985,
Belknap Press of Harvard University Press. Page 96.
6
Cohen, J. Bernard. Revolution in Science, Cambridge, 1985,
Belknap Press of Harvard University Press. Page 96.
7
Our bell curve illustrations were inspired by a drawing
originally produced by Control Engineering Online.
8
Deming, W. Edwards. Out of the Crisis. Cambridge.
Massachusetts Institute of Technology, Center for Advanced
Engineering Study. 1982.
9
As a sidebar note, it is interesting to know that Gantt
patented a number of devices in collaboration with Frederick
Taylor when they worked together at the Bethlehem Steel
Mill on Taylor’s Scientific Management theory.
10
Inspiration for this particular grid came from
Moresteam.com. http://moresteam.com/ Their on-line Six
Sigma Black Belt course is interesting and informative.
11
http://www.fenews.com/
12
http://www.processmodel.com/
Case Studies
C
ase studies needed to meet four criteria. Though
names, places, and data were altered to protect
privacy, each story had to be true. It had to be
entertaining. Each example also needed to graphically explain
how evidence-based decisions produced crowd pleasing
financial returns. Finally, the story needed to be a fair,
representative sampling of what we each have repeatedly seen
over the past 20 years of our professional life.
���������
������� �������
Figure 1 All of the high numbers fell
���������
������� ������� �� �
����
����
The total time to collect data for the data matrix was five
days. The analysis and presentation took one hour. Eventually
one full time position was eliminated through attrition for a
bottom line savings of more than $25,000. This, combined
with avoiding the loss of funding, brought the total value of
the project to $525K.
Define: For more than a year debate had raged over what
could be done to reduce AR days. Suspected causes for this
��������� ���
������ ����� �� ������ �������� ������� ��������
�������� ���� ���� ���������� �������� ��������
������������ ���� ���� ���������� �������� ��������
�������� ���� ���� ���������� �������� ��������
������������ ���� ���� ���������� �������� ��������
Figure 3 P-values less than
������� ���� ���� ���������� �������� ��������
0.05 imply a 95 percent level of ��������������������� ���� ���� ���������� �������� ��������
confidence or more in the results. ����������������� ���� ���� ���������� �������� ��������
The two factors, Customer and ��������������������� ���� ���� ���������� �������� ��������
Relationship, and their interactive ��������������������� ���� ���� ���������� �������� ��������
effect, were statistically significant ������������������������� ���� ���� ���������� �������� ��������
��������������������� ���� ���� ���������� �������� ��������
at this confidence level. ���������������� ���� ���� ���������� �������� ��������
�������������������� ���� ���� ���������� �������� ��������
���������������� ���� ���� ���������� �������� ��������
�������������������� ���� ���� ���������� �������� ��������
���������
������
�� ������ ������ ��
������������
������������
�������� ��������
�����
�����
������
������������
�������� ��������
�����
�����
As they reviewed the actual numbers from data that had been
collected and arrayed in a data matrix, they were surprised.
During the past six months, the Emergency Department
had been closed or on diversion (divert) more than 5300
minutes/month. This totaled eleven, 8 hour shifts, or three
and two-thirds 24 hour days, or 12% of available time. Those
closures penalized patients. They cost the hospital hundreds
of thousands of dollars in potential revenue.
DMAIC
��������������������
���������� �������������������
�������������
�������� ���������
������������������
�������� �����������
������������������
Figure 8 The recommended Six
Sigma closed loop feedback system
is contrary to evidence-based
decisions. Closed loops create
entropy.
������������������������� ����������������
������������������
�������������������������
������������������������� ����������������
����������� �������������
crashing,
�������� ������ ����������
����� ��������� ��������
����� ������ ���� ��������� ��� ������ ���������
�� � ����� �������� ������� ������� �������
�� ����� �������� ������� ������� �������
������������ �������������� ��������� ���������������
Don had come up with a new rule called “75% used up”. He
proposed doing a designed experiment to determine whether
or not the new rule was more cost effective than the old rule.
We met with Don, the area manager and the supplier rep to
discuss the project.
• Line speed
• Die-to-calibrator distance
• Calibrator vacuum
• Screw Revolutions Per Minute (RPM)
• Screw oil temperature
• Barrel zone temperatures
• Die zone temperatures
• Melt temperature
• Melt pressure
• Weight
The die in this case had a dual orifice. This means that two
profiles are extruded at the same time. Results for the two
profiles are distinguished in the matrix as Sides 1 and 2.
Endnotes
1
Cartmill, Matt. “Oppressed by Evolution”. Discovery
Magazine, March, 1998, pages 78-83 as reported by Richard
Dawkins on page 20 in his book Unweaving the Rainbow.
2
http://gi.grolier.com/presidents/aae/side/knownot.html
3
Cheryl Payseno, an RN, former hospital administrator and
certified Six Sigma black belt wrote this case study for us.
Cheryl led the charge for the use of Designed Experiments
in health care in 1995 with Daniel Sloan. Results from those
early innovations were published by the American Society for
Quality’s Quality Press.
4
Pfister, Albert J., Zaki, M. Salah, et al. “Coronary Artery
Bypass without Cardiopulmonary Bypass.” Ann of Thorac
Surg 1992; 54:1085-92.
5
Pfister, Albert J., Zaki, M. Salah, et al. “Coronary Artery
Bypass without Cardiopulmonary Bypass.”Ann of Thorac Surg
1992; 54:1085-92.
6
Senge, Peter M. The Fifth Discipline, The Art and Practice of
The Learning Organization. New York. Doubleday Currency.
1990.
Using Profit
Signals
P
rofit signals show you the money. Profit signal vectors
literally and figuratively show you what works best
in any business, financial, health care, manufacturing
or service process. This chapter explains how vector analysis
applied to a data matrix showcases the information contained
in raw data. Once the tools have done their job, the graphic
presentation of evidence paves the way to breakthroughs in
quality, productivity and profitability.
Corrugated Copters
Next, cut or tear the top section to form the “blades.” Finally,
follow the folds at the bottom to form the helicopter’s
������
���
Figure 1 This inexpensive product
is an analogy that works well for ������
teaching data matrix and vector
analysis principles to people in all
industries.
����
���� ����
���� ����
����� ����� ������
One company cuts down the tree, the Supply. Another ships
it as Input to the pulp mill. The pulp mill Process creates
the paper. The packaged Output is sold to its wholesale
Customer. Corrugated Copters is the retail customer who
buys it from the wholesale customer. You and your products
are parts of a system.
The team showed her all their data: 9 seconds, 8.9 seconds
and 10 seconds.
�������������
��������
�������� ������ ����������� ������������ �����
������ ������ ������ ������ ������
� � � ��� ����
��� � � � ���� � ��� � ����
�� � � ��� ���
������������������ � � � � �
��������������� �������������������������� ���� � ���� � ����
they are going to solve the 1,000 year old waste and
rework problems related to the 14th Century’s double entry
bookkeeping system. Our Black Belt CPA Peruzzi told me
the tip off for her was the word “double”. Get it? ‘Double
entry? Rework entry?’ Well, Peruzzi is convinced the entire
double entry ‘bookkeeping system’ is nothing more than a
massive hidden factory loop. With a properly designed data
matrix, the second entry is needless rework.
“Our Black Belt CPAs are arraying entries into a data matrix.
Some have already doubled their personal productivity. It is
a marvel what Six Sigma education and training can do, even
for a Merchant of Venice.”
“Oh it is. It is!” said Avona. “Did you notice that when you
multiply a minus times a minus, the sign becomes a plus? And
look how confusing that –0.4 in the Noise vector column is. I
always have a hard time remembering that a negative number
like –0.1, minus a positive number like 0.3 turns out to be a
bigger negative number!”
“The last time I saw this stuff was when I had to learn to use a
slide rule in Mrs. Beamer’s algebra period.” complained Tom.
“The data will show this is true if and only if the p-value
is small enough. By international standards, a p-value less
than 0.05 gives ‘clear and convincing’ evidence against the
null hypothesis (Table 2). A p-value less than 0.15 gives a
‘preponderance of evidence’ against the null hypothesis. Our
p-value is 0.428, not even close to the lowest standard. This
means there is no evidence at all that the average flight time
is significantly different from 9 seconds. These differences are
Table 2 Standards of evidence probably due to Chance. It is a Law of the Universe.”
table.
For 2500 years the right triangle has shown us the route to
profitability. Ancient Greek mariners used the sextant to
navigate the Mediterranean Sea’s lucrative markets.
Overcoming Obstacles
The best news for executives and workers alike is that cheap,
reliable, and very user-friendly software makes vector analysis
as easy to learn as sending an E-mail.
Table 3 Vector analysis for comparing two helicopter designs. The raw data are flight times minus
the objective of 9 seconds. The profit signal consists of the average variation for each design. The
average variation for white helicopters is –0.2 seconds of flight time. The average variation for
pink helicopters is 0.2 seconds of flight time. The Profit Signal Vector has one degree of freedom
because a single number, 0.2, determines it. When the numbers in this column are squared, the
minus sign disappears. The squared lengths of all the vectors are connected by their part in the New
Management Equation (NME).
Avona played with all sorts of modeling toys. Her office was
filled with them. She told people they were symbolic. She
would go on and on to anyone who would listen about some
artist named Alexander Calder.
“We haven’t talked about that last vector in the back of the
tetrahedron. This is the vector of hypothetical predicted
values. I didn’t include it in my spreadsheet templates because
it isn’t important in the type of experiments we’ve been doing.
It’s tremendously important in response surface experiments.
That’s where we are optimizing over several continuous
variables.
“Wow, look at that p-value in the table,” said Tom tearing his
gaze away from Mary’s profit signals radio tower, “There really
is a difference between the two designs.”
“Gee whiz Mary,” said Tom after Avona had left. “Everyone
can see pink helicopters are best. Why is Avona such a stick-
in-the-mud? And why does she keep saying ‘we’ when she
really means us?”
Table 5 Vector analysis for comparing three helicopter designs. The raw data are flight
times minus the objective of 9 seconds. The profit signal consists of the average variation
for each design. It has two degrees of freedom because it is determined by two numbers,
-0.125 and - 0.200 in this case. The third number, 0.325, is minus the sum of these two.
“Of course,” added Dick. “It’s obvious that flight time should
depend on blade length, not on color.” Tom and Mary
said nothing, but they each wondered why Dick had not
mentioned this “obvious” thing earlier.
“Oh no,” whispered Mary to Tom. “It’s bad enough when she
talks about evidence. Now it’s cubes.”
“I notice this table is just the same as the others, except it’s
wider.”
“By looking at the profit signal vector for blade length (Z), we
can see that using the short blade subtracts 0.20 second from
the overall average flight time. Also, we can see that using
the long blade adds 0.20 seconds to the overall average flight
time. Overall, this means that using the long blade instead
Tom asked, “I know that X, Y and Z are code names for the
three factors. But what do XY, XZ and YZ mean?”
Avona said, “They are code names for the interactive effects
among the factors. An interactive effect exists when the
effect of one factor depends on the level (choice or setting)
of another factor. In this case there were no significant
interactions. Usually there are.”
Mary asked, “Is that why it was OK to just add together the
effects of paper clip and blade length?”
“Exactly!”
Everyone was taken aback to see Avona use a bar chart. “For
heaven’s sakes Avona,” cried Mary. “Have you become a bar
chart bamboozler?”
“If I read this right,” observed Dick, “It looks like we could be
over-engineering our product. Very few of the other factors,
including the expensive paper, makes a difference.”
“Roctev needs to meet this team and hear about these results
soon,” said Avona.
Chapter Homework
Closing Arguments
Endnotes
1
Sloan, M. Daniel. Using Designed Experiments to Shrink
Health Care Costs. Milwaukee. ASQ Quality Press, 1997.
2
Dilson, Jesse. The Abacus, The World’s First Computing
System: Where It Comes From, How It Works, and How to
Perform Mathematical Feats Great and Small. New York, St.
Martin’s Press,1968.
4
Sagan, Carl. Science as a Candle in the Dark, The Demon
Haunted World. New York, Ballantine Books, 1996. Page 328.
5
Sagan, Carl. Science as a Candle in the Dark, The Demon
Haunted World. New York, Ballantine Books, 1996. Page 32.
6
Jakab, Peter L. Visions of a Flying Machine, The Wright
Brothers and the Process of Invention. Washington, Smithsonian
Institution Press. 1990. Page 140.
Predicting Profits
M
aking accurate predictions is an important, difficult
task. By now, you may not be surprised to learn
vector analysis is the international standard for
making predictions as well as for making comparisons. This
is good news for Corrugated Copters and your company too.
The vector analysis methods for solving prediction problems
are known as regression modeling and analysis.
Fingerprint Evidence
Three Wishes
Prediction Practice
“Wait a minute,” said Mary. “This isn’t the same as the other
things you showed us. We aren’t trying to find the best way
of doing something. We’re trying to determine a relationship.
Can you give me a preview of how we’re going to do this?”
“OK, but what’s this ‘coded drop height’ about?” asked Mary.
“The values -1, 0 and 1 are codes for low, medium and high
drop heights. At the minus setting I was sitting in my chair.
At zero I was standing. At 1 I got up on my desktop.
“This means the forecast goes up 1.25 Y units for every coded
X unit. We get the profit signal vector by multiplying this
slope times the coded X data vector.
Figure 5 Picture of the vector “Exactly,” Avona exclaimed. “Well done. Here is the drawing
analysis for fitting Y as a linear for this vector analysis (Figure 5).”
function of X.
8.5
9.5
11.0
Raw
Data
e
Lin
n
sio
es
gr
Re
Coded
Now Mary had a question. “OK, but how do I use all this to
make a prediction?”
“9.67 is the average flight time in our practice data set, and
1.25 is the slope. If the coded drop height is 0.5, then we get:
= 9.67 + 0.67
= 10.34
“Right again,” said Avona. “But let’s save that for when you
get your real data. The statistical software will automatically
show you the prediction error.”
Is that right?”
Table 5 Vector analysis for fitting flight time (Y) as a linear function of coded drop height (X). The
raw data are Mary’s actual flight times minus the objective of 9 seconds.
Closing Arguments
Endnotes
1
Gould, Jay Stephen. The Mismeasure of Man. New York:
W.W. Norton & Company, 1996. page 272.
2
Galton’s complete works are available from a variety of
library and Internet sources. This quote comes from http:
//www.mugu.com/galton/essays/1890-1899/galton-1890-
nareview-kinship-and-correlation.html The origin of this web
page is the comprehensive http://www.mugu.com/galton/
start.html
3
http://www.mugu.com/galton/
© M. Daniel Sloan and Russell A. Boyles, All Rights Reserved, 2003
192 Predicting Profits
4
http://www.fme.fujitsu.com/products/biometric/pdf/Find_
FPS.pdf
5
http://www.fme.fujitsu.com/products/biometric/pdf/Find_
FPS.pdf
6
http://www.wvu.edu/~bknc/BiometricResearchAgenda.pdf
7
Bernstein, Peter L. Against the Gods, The Remarkable Story of
Risk. New York, John Wiley & Sons. 1966. Page 95.
8
Bernstein, Peter L. Against the Gods, The Remarkable Story of
Risk. New York, John Wiley & Sons. 1966. Page 182.
Sustaining
Results
S
tewardship entails honorable conduct in the
management of other people’s property. But, it is
broader than that. It also includes respect for the
people’s moral responsibilities. These responsibilities require a
constant demonstration of trustworthiness in economic and
personal conduct.”1 observed William G. Scott and David K.
Hart in Organizational Values in America.
“What?”
“You’re kidding.”
“No, I’m not. This is old news. Six Sigma is not just a passing
fad. It is an extremely disciplined way of competing for
market share. Let me show you how to calculate a Cpk value
from your data.”
and
and
“But still not good enough for Six Sigma,” said Avona. She
drew a third picture. “Here’s a process with Cpk = 1.33. The
average is 9.6 and the standard deviation is 0.3 (Figure
3). The average is 8 standard deviations above LSL and 4
standard deviations below USL. Therefore,
and
“OK, now for your quiz.” Mary and Dick groaned. “What
would Cpk be if we moved the average to 9, right in the center
of the specification range?”
“Right on,” said Avona. “And FYI, a process with that level
of capability would produce no more than 2 or so defective
outcomes per billion.”
“Just for grins, let’s roll some dice,” suggested Avona. “I will
record them as you roll. Viva Las Vegas!”
“Let work out how many ways there are to get each possible
outcome with two regular dice,” said Dick (Figure 5).
“It does,” said Avona. “In fact, it’s a nice segue into what I
wanted to show you. We can use the throwing of two or more
“Work with me, Dick. It’s just a simulation. OK, let’s get
started. Our initial process involves four dice.”
For each simulation, one of them threw the four dice, one
of them called out the result, and one of them entered the
result into their data matrix statistical program. They traded
jobs once in a while. After completing 1000 simulations, they
decided they had had enough. Avona showed them how to do
a process capability analysis with two mouse clicks. (Figure
7).
“You got it, girl,” said Avona. “Let’s assume now that we
have data-mined our process data base, using vector analysis
of course, prioritized the causes of defects, and successfully
eliminated one of the top causes. Our improved process
involves only three dice.”
“This does look a lot better than the old process,” said Tom.
“But what if the average drifts up over time?”
“We used to,” answered Avona, “but not any more. Not since
Rotcev took over. He immediately insisted that we apply
standards of evidence everywhere, not just in manufacturing.”
“They did,” said Avona. “That was the problem. They could
make the numbers say pretty much whatever our previous
CEO wanted them to say.”
(185TION
scale. (335 NOISE
)
2)
IA
DA (162)
TA
VAR
AV
E RA
GE
(33
50)
IT
R OF AL
P GN
SI (90)
“We can confirm this visually by plotting the profit signal and
noise vectors together on a single graph (Figure 13).
“The solid line is the profit signal vector and the dotted line
is the noise vector. They are plotted in time sequence. The
overall degrees of variability are about the same.
“This control chart tells us the same thing as the F ratio: the
quarter-to-quarter changes are just noise.”
“OK, here’s your quiz. What are two ‘events’ on this chart that
would indicate a real change of some kind?”
Taking Action
After the others left, Avona realized there was another basic
fact about control charts that she needed to teach them. It
wasn’t about how to set up the charts, or how to interpret
them. She felt that was pretty easy.
She knew from experience that control charts were all too
often used as “window dressing”. Maybe “wallpaper” is a
better analogy. In manufacturing at least, she knew that
control charts add real value only when they are used as a
basis for action.
Closing Arguments
Endnotes
1
Scott, William G. and, Hart, David K. Organizational
Values in America. New Brunswick, Transaction Publishers,
1991. Page 139.
2
Committee on Quality of Healthcare in America. Kohn,
Linda T., Corrigan, Janet M., Donaldson, Molla S. Editors.
To Err is Human, Building a Safer Health System. Washington,
D.C. National Academy Press. 2001.
3
http://www.student.math.uwaterloo.ca/~stat231/stat231_
01_02/w02/section3/fi4.4.pdf and http://www.ralentz.com/
old/space/feynman-report.html
4
http://www.uri.edu/artsci/ecn/mead/306a/Tuftegifs/
Tufte3.html
6
AT& T Technologies. Statistical Quality Control Handbook,
copyright 1956 by Western Electric. Copyright renewed by
AT&T Technologies, Inc. 1984.
7
Shewhart, Walter A. Statistical Method from the Viewpoint of
Quality Control. New York, Dover Publications, Inc. 1986.
The Three Rs
E
ducation and training are the first steps in building an
organization founded on evidence-based decisions and
the New Management Equation. Knowledge and skill
necessarily change the nature of authority.1 Trust, decency and
respect replace fear and favor as social adhesives.2
Roctev the CEO, Avona, Tom and Mary listened. They were
smiling.
“Yes. I do.”
“Well, anybody who has taken Statistics 101 at USC can tell
you don’t even know how to do an Analysis of Variance.”
“What!” cried Tom and Mary who had just mounted and
framed their Black Belt certificates. Avona chuckled quietly
and looked at her shoelaces. Roctev nodded his head.
“Let me get this straight Dick,” said Tom with a stern look on
his face. “Are you saying that Black Belts aren’t needed?”
“No. No.” replied Dick. “You are an expert. You are a teacher.
We need experts and good teachers. But people respect you
and Mary because of what you know and do, not because of
your numbered certificates.
“Literacy?”
“Yes. Let’s call the way we work literacy and be done with it.”
vectoR
analysis
wRiting
Reading
Our Proposal
Each age, as Emerson pointed out, must write its own books.
The books of an older generation will not fit ours. Motorola’s
Six Sigma business initiative was designed at a time when
a dual 5.25-inch floppy disk drive IBM computer with an
amber screen was an executive luxury. Harvard Graphics bar
charts on a dot matrix printer were breakthrough technology.
Yes. We do.
We must negotiate and we, all of us, must get to Yes together.9
Endnotes
1
Wood, Gordon S. The Radicalism of the American
Revolution. New York, Alfred A. Knoph. 1992. page 189.
2
Wood, Gordon S. The Radicalism of the American
Revolution. New York, Alfred A. Knoph. 1992. page 189.
3
http://www.dropbears.com/b/broughsbooks/history/articles/
john_adams_quotations.htm
4
Jeffereson, Thomas. The Life and Selected Writings of Thomas
Jefferson, edited by Adrienne Koch and William Peden. New
York, Random House. 1944. Page 48.
5
Paine, Thomas. Collected Writings. New York. Library of
America. Pages 630-633.
6
Emerson, Ralph Waldo. Circles from The Portable Emerson,
edited by Carl Bode in collaboration with Malcolm Cowley.
New York. Penguin Books. Page 229.
7
Sloan, M. Daniel and Torpey, Jodi B. Success Stories in
Lowering Health Care Costs by Improving Health Care Quality.
Milwaukee, ASQ Quality Press. 1995. Pages 87-97.
8
Emerson, Ralph Waldo. Self-Reliance.
9
Fisher, Roger, and Ury, William, Getting to Yes, Negotiating
Agreement Without Giving In. New York, Penguin Books.
1981.
© M. Daniel Sloan and Russell A. Boyles, All Rights Reserved, 2003
Appendices
Black Belt Core Study Texts and Free On-Line Resource book
Knowledge
Application
Analysis
Synthesis
Evaluation
Bamboozle 56, 57
belt grinding 142
Bernstein, Peter L. 191
Black Belt 19, 37, 76, 79, 90
Box, George E.P. 115, 153
break-even analysis 11, 179
Imagineering 31
JCAHO 128
Jefferson, Thomas 217
Kaizen-blitz 145
Keats, John 50
knowledge 36, 39
Mackay, Charles 56
main effect 39
Marconi 43
math phobia 220
Matreshka 106
Maxwell, James Clerk 91
measurements 9, 19, 153
Michelangelo 43
Minitab 88
Motorola 223
multiplication 15
N
NASA 194
Netter, Frank 33
Newton, Isaac 50
New Management Equation 63, 161, 175
Normal distribution 70
n dimensions 31
OFAT 103
P-value 81
p-value 72,
Paine, Thomas
Paper Bags 14
Pareto chart 127, 140, 171
perpendicular planes 41
PERT 97
Picasso, Pablo 37
predicted values 11, 44, 165
process capability 113
process maps 94
Profit Signals 44
Pythagoras 21
Pythagorean Theorem 13
reasoning 81
records 123
refraction 50
regression modeling 177
Ronald Fisher 9, 12
Rothamsted 32
Russian dolls 106
V
variation 10
vector 10, 60, 152
vector analysis 9, 10, 13, 70, 76, 158, 171, 208