See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/286628080

Statistical tools for six sigma

Article  in  Quality Progress · September 2001

CITATIONS READS
16 361

3 authors, including:

Necip Doganaksoy
Siena College
101 PUBLICATIONS   1,194 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Necip Doganaksoy on 28 June 2020.

The user has requested enhancement of the downloaded file.

FRONTIERS OF QUALITY

Statistical Tools for Six Sigma

What to emphasize and de-emphasize in training

by Gerald J. Hahn, Necip Doganaksoy and Christopher Stanard

O ne of the key themes of Six

Sigma is to make decisions
based on data. This idea is
reflected by such popular Six Sigma
sayings as, "We don't know what
function deployment, and softer
organizational skills, such as team
and project leadership, critical to
obtaining favorable business results.
and critical applications.
Most people in the quality field
today have had some exposure to
introductory statistics prior to Six
Sigma training, so the training
we don't know (or don't measure)" Objectives of statistical training should supplement this knowledge
and "In God we trust—all else bring The goal of standard Six Sigma by providing a hands-on review
data!" statistical training is to give Green and a practical perspective on what
To ensure we obtain the right data Belts and Black Belts an apprecia- is most important in the practition-
and transform it into actionable tion of statistical thinking and a er's work. It is also important to
information, we deploy statistical hands-on introduction to the tools present tools often not taught in
tools. These tools and closely relat- needed for successful projects. The most introductory statistics courses,
ed concepts, such as the design of aim is not to convert Six Sigma such as design of experiments.
experiments, are key elements of Six practitioners into statistical experts. Six Sigma generally involves eight
Sigma training and comprise up to Instead, it is to give them the to 15 days of instruction, given in
half of the standard curriculum. The knowledge essential for their suc- one week doses over a three to four
other half consists of various non- cess in obtaining business results. month period. See Figure 1 for a
statistical tools, such as failure Professional statisticians can pro- typical curriculum.1 At GE, all pro-
mode effects analysis and quality vide added support for complex fessional employees receive this or
similar training. For a description of
statistical tools commonly taught in
Six Sigma training see Implementing
Typical Six Sigma Training Curriculum* Six Sigma—Smarter Solutions Using
Statistical Methods by Forest W.
Breyfogle.2
WEEK1 Cause for applause
Define and measure Improve Participants in Six Sigma training
• Six Sigma overview and the Analysis of variance. are taught key concepts and tools
DMAIC road map. 1 Design of experiments. essential to their success, such as the
• Process mapping. • Factorial experiments. design of experiments, gauge
• Quality function deployment. • Fractional factorials. repeatability and reproducibility
• Balanced block design. (R&R), transfer functions (statistical
• Failure mode and effects
• Response surface design. models describing the relationship
analysis.
• Organizational effectiveness between variables) and statistical
concepts, such as team process control. Also, Six Sigma
training emphasizes the power of
development. Control graphical tools over formal statistical
• Basic stats using Minitab. Control plans.
analyses, confirming the saying that
• Process capability. Mistake proofing. a picture is worth a thousand
Measurement systems analysis. Special applications: discrete words—and even more in the wallet.
parts, continuous processes, The underlying approach to
Analyze administration, design. teaching Six Sigma is a key factor to
Statistical thinking. Final exercise. its success. Some of the proven best
Hypothesis testing and Notes: practices in this approach are sum-
confidence intervals (F-tests Project reviews are done each marized in the sidebar on p. 81.
and t-tests). day in weeks one through four. Many of these practices were part of
Correlation analysis. Hands-on exercises are assigned the basic Six Sigma philosophy
most days. espoused by Mikel Harry and his
Multivariate and regression
There are three weeks of applied associates. Others have evolved
analysis.
time between sessions. over time and are based upon
*Slightly modified from original. lessons learned in successfully
DMAIC - Define, measure, analyze, improve, control. implementing Six Sigma.

78 S E P T E M B E R2 0 0 1 W W W . A S Q . O R G
Top 10 recommendations rather than to describe the current
As the scope of Six Sigma has
expanded beyond the original
define, measure, analyze, improve,
5.
process (an enumerative study).7 8
Presentation of the design of
experiments as a step-by-step
learning process, rather than a
Useful Sources
control model (DMAIC), so have one-shot undertaking.9 This idea Articles:
training needs. Tools and concepts is illustrated by the well-known • "Quality Quandaries," a col-
that were not part of the standard helicopter design example.10 This
Six Sigma package, but have proven umn in Quality Engineering
example involves experimenting (a quarterly journal published
their worth in successful projects, with factors such as the length,
need to be added to the standard thickness and weight of the paper by ASQ, www.asq.org).
curriculum. Here is a list of our rec- used to make paper helicopters. • "Statistics Roundtable," a
ommended top 10 additions: 6. Recognition of the fact that in bimonthly column in Quality
1. The shortcomings of much his- many designed experiments all
torical data and the paramount Progress.
variables are not created equal.
importance of getting the right On the Internet:
Some are harder to change than
data.3 others, leading to the frequent • NIST/SEMATECH, Engineering
2. The justification for assuming use of split plot designs.11 Statistics Handbook, www.nist.
data to be normally distributed 7. The use of simulation for discov- gov/itl/div898/handbook/
and the importance of normality ering how large a sample is index2.htm
in different situations.4-5 needed and similar questions
3. How to handle non-normal data.6 • StatSoft, Electronic Statistics
about planning investigations.
4. Why statistical intervals often William Meeker and Luis Escobar Textbook, www.statsoftinc.com/
understate the actual uncertainty. apply this approach to various textbook/stathome.html
This happens because Six Sigma applications in reliability, includ-
projects typically aim to improve ing planning a product life test.12
a dynamic process (what Deming 8. Analysis of categorical data
referred to as an analytic study), and the desirability of having

K LEADERSHIP K SIX SIGMA

> ISO 9000:2000 > CEMARK
> ISO 17025
> TL9000
> SPC
iso 14000
PROCESS MAPPING
HBSfflBHB^^iSHSSBBBI >•
>
TEAM BUILDING
LEAN TOOLS
TQM * COST of QUALITY

EuroQuest...your passport for success in the global marketplace

September October November December
10-14 Dallas 8-12 Otlando 5-9 Houston 3-7 Indianapolis
3 ISO 9000:2000 Lead Auditor 17-21 Indianapolis 15-19 Nashville 5-9 Atlanta 10-14 Atlanta
=1 24-28 Atlanta 22-26 Raleigh
H 17 Dallas 1 Atlanta 5 Indianapolis 14 McAllen
=• Understanding ISO 9000:2000
24 Lexington 29 Orlando 27 Atlanta
1 18-20 Dallas 2-4 Atlanta 6-8 Indianapolis
mm Implementing ISO 9000:2000
25-27 Lexington 30-1 Orlando 27-29 Atlanta
Eel 13-14 Dallas 15-16 Orlando 12-13 Chicago 3-4 Indianapolis
I Auditor Transition (RAB approved)
29-30 Atlanta 15-16 Nashville
3 24 Lexington 1 McAllen 12 Greenville, SC
£J Planning for TS 16949
CT
d• Transitioning from QS to TS 16949 25-26 Lexington 2-3 McAllen 13-14 Greenville, SC

9 Certified EMS Lead Auditor 10-14 Atlanta 29-2 Indianapolis 3-7 Orlando

I Integrating EMS into your QMS 27-28 Dayton 17-18 Houston 17-18 Atlanta

Circle #180
Call for Additional Dates, Locations and In-House Training
Atlanta, GA • 800-355-3876 • 770-395-0124 • Fax 770-395-0737 • www.euroquest.net

Q U A L I T Y P R O G R E S S SEPTEMBER 2001 79
FRONTIERS OF QUALITY com.

continuous data, where possible.13 Many of the statistical tools from es, such as classification and
9. Informative ways of analyzing the standard training have general regression trees and logistic
and displaying data graphical- applicability for these audiences, regression for categorical response
ly 14,15,16,17 but the examples should be relevant variables.23'24
10. Tools for quantifying individual to the participants' interests. For • Chemical and processing indus-
sources, or components, of varia- example, in presenting Six Sigma tries: Mixture experiments, the
tion.18 These have become impor- for commercial transactions, you marriage of engineering process
tant because people are increasing- might illustrate gauge R&R in terms control and statistical monitoring
ly realizing one of Six Sigma's of the consistency of assessments by and concepts of combinatorial
major goals needs to be the reduc- different insurance appraisers. Also, chemistry.25-26-27
tion of variability, in addition to, or you might use sampling human • Software development: Design of
instead of, improving the mean. populations for marketing evalua- experiments to identify faults and
See the sidebar "Useful Sources" tions and sampling bills to quantify the use of reliability growth
(p. 79) for more information on error rates as examples, while you curves.
these and other tools. downplay the formal aspects of the
design of experiments.20 What to de-emphasize
Focus on targeted audiences In addition, the different areas of Which tools in the standard train-
Initially, Six Sigma was aimed application call for specialized tools ing have been less effective and
principally at managers and practi- that should be included in targeted might be de-emphasized to make
tioners directly involved in the training: room for additions? Hypothesis
manufacture of a product, and it • Design for Six Sigma: Tools for tests, such as F-tests and t-tests, top
was based on the DMAIC model. reliability improvement and prod- our list. These tests deal with statis-
Since then, it has proven its applica- uct life data analysis.21-22 tical significance. However, in
bility to the entire business, from • Commercial and other transac- applications we are generally inter-
design to commercial transactions, tions: Discrete event simulation to ested in practical significance.
and even to enterprises other than model nonmanufacturing process- Unfortunately, these two concepts
manufacturing operations, such as es, and graphical and data mining are far from equivalent.
banks, hospitals and schools. 19 tools that deal with large databas- Hypothesis tests are sample size

'• ISO BOO1 :SOOO

|K CONVERSION TOOLS
NEW POCKET GUIDE NEW ISO B001 :SOOO
The innovators of the ISO 9000 pocket guide (with over 250,000 in CONVERSION KIT
print) are now offering, What Every Employee Needs to Know This conversion kit contains raluabk information needed to quickly
About ISO 9001:2000 and easilv transition your ISO 9001 svstein to ihe new ISO 9001:20(IO
requirements.

"ef
,,n
9001:2000

Kit includes Quality M

Transition Checklist, i
*Cliwiges to the new Standard ISO 9000:2000 Qualit rocedures.
are highlighted in blue

Orders of 100 or more receive a free CD with a narrated PowerPoint

training presentation. A S250 value! . . . .,
H. J. STEUOEL. a ASSOCIATES, INC.

Telephone 608.271.3121, Facsimile 608.371.4735

6417 Normandy Lane, Suite 205, Madison, Wl 53719
Circle #27 www.hjsteudei.com, [email protected]

80 S E P T E M B E R 2 0 0 1 W W W . A S Q . O R G
 dependent. With a sufficiently large variance (except as a tool for estimat- How To Contact'Quality
sample size, you can disprove most
statistical hypotheses, therefore
ing components of variation) in
favor of graphical displays. Progress'About Advertising,
establishing statistical significance
even though the results are not of
We would also place less empha-
sis on R-squared (the percentage of
Subscriptions and Reprints
practical significance. Conversely, variability accounted for by a fitted
lack of significance might often be regression line) as a measure of For general information, go to
due to inadequate sample size, association because this, unlike the www.asq.org.then click on Quality
rather than lack of true effects, standard deviation of a fitted Progress or the magazine cover; or
therefore resulting in inability to regression, provides limited infor- call 800-248-1946 or 414-272-8575.
establish statistical significance, mation on prediction ability.29
even though the effect may not be
of practical significance. The evolution of Six Sigma Subscriptions, back issues
Statistical interval statements, such Six Sigma training is meeting its and missing issues
as confidence intervals that quantify goals by impacting the quality experi-
the statistical uncertainty, are gener- enced by customers, streamlining the Call ASQ customer service at 800-
ally more informative.28 Similarly, we hidden factory and strengthening the 248-1946, or checkwww.asq.org
would de-emphasize the analysis of bottom line for the corporations under "Quality Progress," then
"subscribe" or "report a missing
issue." Rates appear on page 4 of
this issue of QP.

Proven Best Practices in Six

Reprints and permissions
Training Back issue and bulk photo-
copies—call ASQ customer
• Road map: Integrate statistical tools into an overall road map, such service at 800-248-1946 or
as the traditional define, measure, analyze, improve, control model. University Microfilms International
This allows students to learn how the tools fit together. It contrasts at 800-521-0600, ext. 533 (313-761-
with a typical college statistics course in which the training sequence 4700 in Michigan).
is determined by mathematical complexity, rather than a tool's natur-
al place in addressing real-life problems.
Internal or personal use photo-
• Theory vs. applications: Teach the tools and their applications, and
omit the underlying theory. copy authorization—call Corinne
• Appreciation of assumptions: Don't confuse de-emphasizing theory Johnson at 800-248-1946, ext.
with omitting assumptions. Six Sigma practitioners need a clear 7373, or the Copyright Clearance
understanding of the key assumptions underlying the use of each Center at 978-750-8400 for autho-
tool, the assumptions' importance, how to evaluate the assumptions rization and fee information.
in a given situation and how to proceed if the assumptions are violat-
ed.
• Trainers: These people should be enthusiastic, experienced Black
Advertising and media kits
Belts and Master Black Belts who can speak knowledgeably about
business applications and who also understand the basic principles Display advertising—call 800-248-
of adult learning. 1946 and ask for ASQ ad sales, or
• Hands-on implementation: Instruction on a tool should come with e-mail [email protected].
easy-to-use, and often familiar, statistical software, such as Excel,
Minitab1 and JMR2
Classified advertising—call 800-
• Project tie-in: Six Sigma practitioners should learn the tools by
immediately applying them to their projects. However, we must 248-1946 and ask for ASQ ad sales,
resist being dogmatic and should not require participants to apply a ore-mail [email protected].
specific tool. Instead, we need to emphasize using the method that is
most effective for the problem at hand. For a media kit, call ASQ customer
• Tailored training: Adapt the materials and examples to the special- service at 800-248-1946, or find
ized business needs of the target audience. complete media and contract
information atwww.asq.org under
REFERENCES
"QualityProgress/mediakit"
1. Minitab, Minitab User's Guide 1: Data Graphics and Macros and Minitab User's Guide 2: Data
Analysis and Quality Tools, release 12 (State College, PA: Minitab, 1997).
2. SAS Institute Inc., JMP User's Guide, Version 3.1 (Gary, NO SAS Institute Inc., 1995).

QUALITY PROGRESS SEPTEMBER 2001 81

View publication stats