RESEARCH

Simplifying the Six Sigma Toolbox

through Application of Shainin
includes research articles that
focus on the analysis and
resolution of managerial and
academic issues based on
analytical and empirical or
case research
DOE Techniques
Sunil Sharma and Anuradha R Chetiya

Executive The success of a Six Sigma programme in an organization depends to a large extent on the
success of the Six Sigma projects, which in turn depends on how the team handles the prob-
Summary lem and whether the right combination of tools is being applied to address the root cause. The
Six Sigma toolbox consists of a wide range of tools comprising, on the one hand, simple and
commonly used tools like flow charts, Pareto analysis, and cause-and-effect diagram and the
more advanced statistical tools like design of experiments, regression analysis and many more,
on the other hand. While the simple tools are easy to apply, understand, and analyse, engi-
neers perceive the more advanced tools difficult to comprehend. Design of experiments (DOE)
is one such tool. Two well-known approaches of design of experiments are the Classical DOE,
pioneered by Sir Ronald A Fisher and the Taguchi approach, pioneered by Dr Genichii Taguchi.
A third approach to experimental design—the Shainin DOE techniques, offered by Dr Dorian
Shainin–can be considered as a very good alternative to the other approaches. They are much
simpler than the factorial designs, response surface designs, and orthogonal arrays of the
conventional approaches of DOE, but at the same time are recognized as being very powerful
and effective in solving the chronic quality problems that plague most manufacturers. Shainin
DOE basically works at eliminating suspected process variables by mostly using seven differ-
ent tools, viz.,
• Multi-Vari Charts
• Component Search
• Paired Comparison
• Variable Search
• Full Factorials
• B vs. C (Better vs. Current) Analysis
• Scatter Plots or Realistic Tolerance Parallelogram Plots.
Though not very well documented, these tools have proved to be the key drivers in the suc-
cess of many companies, e.g., Motorola.
This article examines two projects of a leading automotive and general lighting lamp manu-
KEY WORDS facturing company, in which a combination of the standard Six Sigma tools and Shainin tools
has been successfully used to address the root cause of the problems. The advantage of using
Shainin tools is that:
Six Sigma Tools
• Very small sample sizes are required to analyse the problem. Often samples as small as 2
Factorial Designs or 3 are enough to make statistically valid conclusions.
Shainin DOE • Statistical software is not required to analyse the data. In fact, Shainin DOE does not even
require knowledge of complex statistical tools.
Suspected Sources of • It involves employees at all levels, including workers and junior staff in problem solving
Variation that was hitherto a domain of senior technical experts.
• Also, the success of the projects had a very positive effect on the morale of the employees
Red X in terms of convincing them that Six Sigma is not all about using complex statistical tools.

VIKALPA • VOLUME 34 • NO 1 • JANUARY - MARCH 2009 13

A
technical feature that distinguishes Six Sigma
from other quality approaches is its ability to
use statistical methods within a structured for-
mat to reduce defects and improve processes. Some of
the criticisms of the Six Sigma methodology perhaps
stems from the fact that it is sometimes too statistical
and beyond comprehension of the people involved in
implementing it in practice. Eckes (2001) is of the opin-
ion that Six Sigma initiatives can fail if the organization
believes that better quality is possible only through the
use of sophisticated statistical tools. The objective of this
paper is to examine as to how to simplify and demystify
the use of Six Sigma tools. While a critical element of Six
Sigma is the rigour of methodo-logy, it is also true that
the methodologies and tools that work for one organi-
zation may not do so for another. Because of diversity
in products and processes, it becomes imperative to
choose the right combination of Six Sigma tools, to
rightly select projects, to deal with specific challenges
of different structures and processes. It examines the
applicability of a simpler but not very frequently used
methodology known as the Shainin methodology in Six
Sigma projects.
TOOLBOX FOR SIX SIGMA
Pyzdek (2003) has classified Six Sigma tools into three
categories, viz.,
• Basic Six Sigma methods
• Intermediate Six Sigma methods
• Advanced Six Sigma methods
Basic Six Sigma Methods
The basic Six Sigma tools are further categorized as prob-
lem solving tools, 7M tools, and knowledge discovery
tools (Table 1).
Intermediate Six Sigma Methods
These methods include a host of enumerative and ana-
lytical statistical tools.
A few enumerative statistical methods are:
• Distributions
• Statistical inference
Some of the analytical methods that can be used in Six
Sigma problem solving are:
• Basic control charts
• Exponentially weighted moving average (EWMA)
charts
Advanced Six Sigma Methods
These include tools like
• Design of experiments (DOE)
• Regression and correlation analysis
• Process capability analysis
While the basic and intermediate methods are relatively
easier to understand and use, the advanced methods are
perceived to be difficult to comprehend and interpret.
Design of Experiments (DOE) is one such tool.
EXPERIMENTAL DESIGN: CLASSICAL AND
TAGUCHI APPROACH
Six Sigma process improvements consist of analysing
relationships between an output variable (Y) explained
wholly or partly by process variables (Xs) that affect the
output. A key step in Six Sigma projects is the identifi-
cation of the root cause of the problem out of the poten-
tial Xs. Experimental design is one of the tried and tested
statistical techniques long used by industrial engineers
to identify the key variables affecting output. Through

Table 1: Basic Six Sigma Tools

Problem Solving Tools 7M Tools Knowledge Discovery Tools

Process mapping Affinity diagrams Run charts
Flow charts Process decision program charts Descriptive statistics
Matrix diagrams
Check sheets Histograms
Tree diagrams
Pareto analysis Interrelationship diagraphs Exploratory data analysis
Cause-and-effect diagrams Prioritization matrices
Scatter plots Activity network diagrams

14 SIMPLIFYING THE SIX SIGMA TOOLBOX THROUGH APPLICATION OF SHAININ DOE TECHNIQUES
designed experiments, changes are deliberately intro-
duced into the process to better understand which of
the Xs are affecting the output variable. There are two
well-known approaches to experimental design. The first
approach is the classical design of experiments credited
to Sir Ronald Fisher who initially experimented in the
field of agriculture. However, this method is now widely
used in many fields. The second approach is the Taguchi
approach pioneered by Dr Genichi Taguchi of Japan who
adopted the classical approach to reintroduce the con-
cept of orthogonal arrays. The commonly used classical
Design of Experiment (DOE) tools are the family of fac-
torial experiments consisting of full factorial designs and
fractional factorial designs. A full factorial allows us to
test all possible combinations of factors affecting output
in order to identify which ones are more dominant. A
fractional factorial tests just a fraction of the possible
combinations. Though a very popular tool, many engi-
neers and quality practitioners find design of experi-
ments difficult primarily because of the complexity of
having to create the conditions for conducting the ex-
periments in an industrial environment where interrupt-
ing production lines and changing machine settings may
be sometimes difficult and unproductive.
SHAININ DOE APPROACH
An alternative to the Classical and Taguchi experimen-
tal design is the lesser known but much simpler Shainin
DOE approach developed and perfected by Dorian
Shainin (Bhote and Bhote, 2000), consultant and advisor
to over 750 companies in America and Europe. Shainin’s
philosophy has been, “Don’t let the engineers do the
guessing; let the parts do the talking.” Shainin recog-
nized the value of empirical data in solving real-world
problems. He introduced the concept of Red X, the domi-
nant source of variation, among the many sources of
variation of a problem that inevitably accounts for nearly
all the unwanted effect. In fact, Shainin (Shainin, 1995;
1993b), classified all causes of chronic quality problems
into three X’s, viz., the Red X, the Pink X- the second
most important cause(s), and the Pale Pink X – the third
most important cause(s). According to him, these three
Xs together account for over 80 per cent of the variation
that is allowed within the specification limit and when
captured, reduced, and controlled, these can eliminate
this variation. Shainin developed techniques (Shainin
and Shainin, 1990; 1992a; 1992b; 1993a; 1993b; Shainin,
Shainin and Nelson, 1997) to track down the dominant
VIKALPA • VOLUME 34 • NO 1 • JANUARY - MARCH 2009
source through a process of elimination (Shainin, 1993b),
called progressive search. These techniques, also referred
to as the Shainin System for quality improvement, de-
veloped over a period of over 40 years, are simple but at
the same time powerful and easier to interpret and im-
plement in an industrial environment. In a way, these
may be considered as the non-parametric equivalent of
Taguchi’s DOE as they do not make any restrictive as-
sumptions about population parameters.
“Ninety per cent of the tolerances and specifications are
arbitrarily decided and are not correct for the intended
applications” (http://www.ind.tuv.com). This of course
in no small measure contributes to a host of what are
called ‘chronic quality problems’ – problems that keep
recurring. Finding solutions to these problems then be-
comes a major challenge for engineers and quality prac-
titioners. The Shainin techniques are primarily known
to produce breakthrough improvements in eliminating
chronic quality problems. These are highly effective in
pinpointing towards the root cause and validating it.
Applications of the Classical and Taguchi methods in
various fields have been extensively researched. A few
very recent contributions in this area are from Wang and
Ren (2007), Singh, Maheswari and Pandey (2007), See
and Wong (2008), Abdo (2008), Rao, et al (2008), and
Anawa, et al (2008). In contrast, the Shainin system has
not been extensively reviewed, academically, and very
limited studies have been carried out in this area. Bhote
(1990) compared Shainin techniques with Design of Ex-
periments and Taguchi methods, in the context of the
electronics industry and concluded that the Shainin tech-
niques are simpler, less costly, and statistically more
powerful than the other two. Logothetis (1990) also
evaluated the Shainin techniques in relation to the
Taguchi methods and statistical process control meth-
ods. Bhote (1991) and Bhote and Bhote (2000) described
these tools in their books, but there have been many criti-
cisms regarding their claims and the tools described.
Though Nelson (1991), Moore (1993), and more recently,
Zeigel (2001) criticized the Shainin System as unsubstan-
tiated and exaggerated, Steiner, et al (2008), are of the
opinion that some of the ideas behind the Shainin Sys-
tem are genuinely useful. Goodman and Wyld (2001)
offered a case study involving the use of Shainin DOE
in an industrial operation. This study suggested that ‘the
Shainin methodology is very practical and can be easily
executed in many settings, making it one of the most
15
approachable quality techniques available.’ Verma, et al
(2004) used a slightly different approach to compare the
methods. In their study, three cases of Taguchi experi-
ments were picked up from the available literature and
the Shainin method was then re-applied to find out
whether it had an edge over the other DOE techniques.
A comparison between Taguchi and Shainin techniques
in an aerospace environment was offered by Thomas
and Anthony (2005). A few other authors who have stud-
ied these techniques are Traver (1995), Ledolter and
Swersey (1997), Does, Roes and Trip (1999), De Mast, et
al (2000) and Steiner and MacKay (2005).
Motorola is one of the many companies to have used
these tools. Though Motorola’s phenomenal success with
Six Sigma approach is now history, a little known fact is
that the Shainin techniques were one of the key drivers
of their success. Motorola started using them way back
in 1980 with remarkable results. According to Bhote
(1991), who has done a significant amount of research
in these techniques, “Motorola’s plants, suppliers, and
customers all over the globe conducted about hundreds
of Shainin experiments as part of their Six Sigma qual-
ity movement.” Daimler Chrysler, General Motors, and
Delphi Corporation are some of the other companies
reported to have used the Shainin Methodology success-
fully. The American Society for Quality (ASQ) has very
recently instituted the Dorian Shainin Medal in recog-
nition of his contribution towards developing the Shainin
techniques that are now widely recognized as simple,
powerful, and practical tools to help in solving the manu-
facturers’ problems. In 2005, statistician Karl Bennet was
the first recipient of this award for his work in this area.
The Classical DOE, Taguchi DOE, and Shainin DOE are
compared with each other in Table 2.
An examination of the three approaches clearly indicates
that the Shainin tools have an edge over the other two
approaches in terms of cost, time, training, complexity,
scope, and ease of implementation.

Table 2: Comparison of Classical, Taguchi, and Shainin Approaches of DOE

Classical DOE Taguchi DOE Shainin DOE

Primary tools Factorial experiments Orthogonal arrays a. Component search
b. Multi-vari analysis
c. Paired comparison
d. Product/Process Ssarch or Variable search
e. Full factorials
f. B vs. C (Better vs. Current) analysis
g. Scatter plots
Advantages Effective when interaction Effective when interaction Very powerful irrespective of the presence
effects are not present effects are not present or absence of interactions.
(20 to 200% improvements). (20 to 200% improvements). Maximum optimization possible.
Limited possibilities for Limited possibilities for
optimization. optimization.
Cost/Time Moderate Moderate Low
Training 3 to 5 days 3 to 10 days 1 to 2 days
Complexity Moderate High Low (based on simple & basic mathematical
operations)
Scope Requires use of statistical Requires use of statistical Software not necessary.
software e.g., SAS, SPSS, etc. software e.g., SAS, SPSS, etc. Can be used at prototype, pre-production
Used mainly in production. Used mainly in pre – and production stages.
production. However, can be
used at the design stage
under certain constraints.
Ease of Moderate Poor High
Implementation Requires knowledge of Almost no knowledge of statistics required.
statistics. Easy to understand at all levels including
Engineers find methods shop floor workers, engineers, and suppliers,
complex to comprehend thus creating an overall positive impact.
and interpret.
Source: Bhote, 1991.

16 SIMPLIFYING THE SIX SIGMA TOOLBOX THROUGH APPLICATION OF SHAININ DOE TECHNIQUES
Shainin DOE Tools: An Overview of Select Tools
Shainin DOE basically works at eliminating suspected
process variables (Xs) mostly by using seven different
tools (Bhote, 1991):
• Multi-Vari Analysis
• Component Search
• Paired Comparison
• Variable Search
• Full Factorials
• B vs. C (Better vs. Current) Analysis
• Scatter Plots or Realistic Tolerance Parallelogram
Plots
Subsequently, Bhote and Bhote (2000) added three more
tools to the list, viz., Product/Process Search, Concen-
tration chart, and Response Surface Methodology. How-
ever, our aim is to study the application of a few basic
but effective tools along with their application in Six
Sigma projects.
A brief discussion and understanding is necessary be-
fore these tools are applied:
Multi-vari analysis: Multi-vari analysis is used to re-
duce a large number of suspected sources of variation
to a smaller family of variables containing the dominant
source of variation. A multi-vari analysis is based on
the assumption that any process variation can occur as
a result of three factors, viz.,
• Positional or Part-to-Part Variation
• Temporal or Time-to-Time Variation
• Cyclical or Stream-to-Stream Variation
Each type of variation is individually measured using a
run of approximately 3 to 5 units produced consecutively
at any given time. After a time lapse, another run of 3 to
5 units are produced. This process is repeated until 80
per cent of the out-of-control variation in the process is
captured. A plot of these results indicates which one of
the three variations is maximum. Simple tools like the
average and the range are then used to identify which
type of variation out of these three is the highest. In case
the positional or part-to-part variation is the highest, the
source of variation is usually attributed to the machine
or the process design. Further tools can be used to iden-
tify the exact cause. If time to time variation is high, the
team needs to try and identify the event that has caused
the shift in average in a variable over a period of time.
VIKALPA • VOLUME 34 • NO 1 • JANUARY - MARCH 2009
And, in the last case of high cyclical variation, the team
needs to focus on identifying which stream causes the
highest variation to get down to the root cause. DeMast,
Roes and Does (2001) have studied this technique in
detail.
Component search: Component search technique is a
simple but powerful tool to lower down the number of
suspected sources of variation to the family of Red X or
the Red X itself. This technique can be used primarily in
assembly operations, i.e., when a product can be disas-
sembled and re-assembled, as it allows finding the com-
ponent of an assembly that is most likely to cause the
problem. This technique is used to find the suspected
source of variation (SSV) that either originates from the
assembly process or the components in the assembly.
Thus, the various stages of the component search method
(Bhote, 1991) are:
• Ballpark stage: This stage determines whether Red X
or Pink X is actually present among the components
or factors being investigated.
• Elimination: The second step involves elimination of
all unimportant main effects and their associated in-
teraction effects.
• Capping run: This stage consists of re-verification of
the important and unimportant effects.
• Factorial analysis: Once the important and unimpor-
tant factors have been double-checked, factorial
analysis is used to determine the magnitude of the
important SSVs and their interactions.
Once the problem components are identified, a paired
comparison can be used to further identify the para-
meters in the component causing the problem.
Paired comparison: Paired comparison is similar to com-
ponent search except that it is used when components
cannot be reassembled or disassembled. This technique
uses a small sample consisting of good parts and bad
parts to further narrow down the range of potential
sources of variation. The good and bad parts are selected
on the basis of a suitable parameter related to the prob-
lem. Selecting the good parts implies selecting the Best
of Best (BOB) parts and selecting the bad parts implies
selecting the Worst of Worst (WOW) parts with respect
to the desired response. The terms BOB and WOW are
attributed to Shainin et al (1997). Steiner et al (2008) are
of the opinion that this comparison of extreme values of
the response variable is unique to the Shainin System
17
and its roots lie in the assumption that the values of the
two groups must be substantially different and hence
identifiable. Usually 8 good and 8 bad parts are selected.
Once the sample is selected, each SSV is measured on
the 16 parts and the results are recorded in the form of a
table. The analysis to pinpoint a suspected source of
variation as being significant or not significant is based
on a minimum “endcount” (Bhote and Bhote, 2000),
which is described later. This technique can be used only
when the suspected source of variation is measurable
on both the good and bad parts. This technique is ex-
plained in more detail in the project illustration.
Product/Process search or Variable search: The first
three methods are generally used to reduce the SSVs to
a smaller number. Once this is achieved and the number
of SSVs has been reduced to a range of 5 to 15, the vari-
able search technique can be used to further zero-in on
one or two variables that could possibly be the domi-
nant sources of variation. It is very useful to separate
the important variables from the unimportant ones, con-
trol the former, and open up tolerances on the latter.
This technique can be used to identify factors of varia-
tion related to the process parameters and is especially
useful in case the sources of variation are related to in-
put material dimensions which are subject to change
during processing. Ledolter and Swersey (1997) have
given a critical evaluation of variables search as an ex-
perimental design method for uncovering important
factors. Anthony (1999) has given a brief description of
this method, showing with the help of an experiment
that it is a powerful tool for identifying key process vari-
ables and therefore recommending its utilization by the
engineering fraternity in manufacturing organizations
as a problem-solving tool. Anthony and Cheng (2003)
have illustrated the potential of Shainin’s variable search
method as a powerful methodology for problem solv-
ing with the help of a simple catapult example.
Full factorials: This is same as the full factorial of Clas-
sical DOE except that here Shainin recommends its use
once the number of SSVs has been bought down to four
or less than four by using one or more of the above tech-
niques. Thus, the maximum size of the full factorial ex-
periment would be 24.Also the objectives of this tool
remain the same as those of the variable search tech-
nique, i.e., pinpoint the Red X or the Pink X, separate
and quantify the important and unimportant factors, and
open up tolerances on the unimportant ones.
18 SIMPLIFYING THE SIX SIGMA TOOLBOX THROUGH APPLICATION OF SHAININ DOE TECHNIQUES
Better vs. Current (B vs C) analysis: This is a final vali-
dation tool useful in validating the root cause identified
using the other methods. Here B stands for the better
process and C stands for the current process. Some ap-
plications of this method are:
• to validate the root cause
• to validate product/process changes made
• to validate improvement actions.
An advantage of this method is that it can be applied
using very small sample sizes starting from 3 from the B
process and 3 from the C process, to make a final con-
clusion with a certain degree of statistical confidence
whether the B process is better than the C process. This
technique is explained in detail in the project illustra-
tions.
Scatter plots or Realistic tolerance parallelogram plots:
The purpose of scatter plots or realistic tolerance paral-
lelogram plots is to determine an optimum and realistic
level of tolerance value. It is a graphical representation
of thirty observations of the independent variable against
corresponding values of the dependent variable. If there
is correlation between the two, then the independent
variable is the dominant source of variation and an ap-
propriate target value along with a tolerance can be de-
termined graphically. In case of absence of correlation,
the independent variable is not important and the toler-
ance can be opened up to an economical level.
Although the tools are systematic and follow a logical
sequence, these can be applied individually or in groups
depending on the nature of the problem. A compara-
tive analysis of these tools is presented in Table 3.
The Shainin tools have the following major advantages.
These tools are:
• simple to comprehend by both engineers and line
workers
• logical and based on common sense
• practical in terms of implementation in prototype
building, production as well as easy to work with
suppliers
• universally applicable in a wide range of industries
• statistically valid and powerful
• able to produce excellent results with quality gains
of up to 100-1,000 per cent.
Table 3: The Shainin DOE tools

Tool Objective Where Applicable When Applicable Sample Size

Multi-vari chart To reduce a large number
of unrelated variables to a
family of smaller related
causes
Components search To determine if the Red X is
in the assembly process or
the components in the
assembly
Paired comparisons Provide clues to determine
the red X by using a paired
comparison of good and
bad parts
Variable search Pinpoints the Red X,
Pink X etc
Finds the magnitude of the
important main and
interaction effects
Open up tolerances on
unimportant variables
Full factorials Same as variable search
B vs. C analysis To validate a new/better
process over a current one
Scatter plots or Determines optimum values
Realistic tolerance for Red X, Pink X variables
Parallelogram plots and their maximum
allowable tolerances
Source: Bhote, 1991.
To get a snapshot of how At engineering pilot run, Min 9- 15
a product/ process is production pilot run or
currently running in production
In assembly operations At prototype, 2
engineering pilot run,
production pilot run
or in production
Where there are matched Same as component 1 to 20
sets of differently search
performing products that
cannot be disassembled
When the number of In R &D, development 1 to 20
variables is between engineering and in
5 to 20 production
When the number of Same as variables 1 to 16
variables are between search
2 and 4
Usually follows one or In prototype, pilot run Min 3B’s and
more of the above tools or production 3C’s
Follows the above tools In pilot run 30

Some other benefits reported by users of Shainin DOE
techniques are accrued by virtue of their direct role in:
• solving chronic quality problems
• reducing process variations
• eliminating defects
• reducing/eliminating scrap/rework/testing
• reducing cost of poor quality
• increasing field reliability of products
• improving financial parameters, like return on invest-
ment (ROI) and market share, etc.
Among the Shainin tools, factorial designs and scatter
plots are commonly known and used. However, the
main differentiators of the Shainin tools are component
search method, paired comparison method, product/
process search, and B vs. C analysis. In order to high-
light the application of these tools, we choose two
projects from the selected predominantly automotive
lamp manufacturing company known to apply Shainin
tools extensively in their Six Sigma projects.
LAMP MANUFACTURING COMPANY
Company Profile
The company is a leading manufacturer of automotive
and general lighting lamps with a group turnover ex-
ceeding Rs 250 crore. With exports to nearly 60 different
countries, it is the largest exporter of lamps in the coun-
try. The company gets nearly 33 per cent of its income
from exports and the rest from the domestic segment. It
has five fully integrated state-of-the-art manufacturing
facilities at Noida in NCR and Dehradun and Haridwar
in Uttaranchal. The five units together deliver over 150
million lamps annually, all benchmarked to international
regulations for quality, performance, and safety. One of
the five facilities, is a 100 per cent export-oriented unit
(EOU), catering to exports mostly to USA and European
countries like Italy and Germany. The company also has
all major quality certifications like ISO 9001:2000, ISO/
TS 16949:2002, ISO 14000:2004, and OHSAS 18001:2007.
It saw a change in ownership in the year 2007 following

VIKALPA • VOLUME 34 • NO 1 • JANUARY - MARCH 2009 19