Transcript what they don` teach you – final Calumet

What they don’t teach you at
Lean and Six Sigma School
Bill Hooper
Elkay Manufacturing
President, William Hooper Consulting Inc.
ASQ CSSMBB, CSSBB, CRE, CQE, CMQ
So what do they teach you in LSS school
Computer generated analysis and design
A language only a few can understand
Statistics! Statistics! Statistics!
4-6 months to get results
30 days for the rebound
How many understand the following?
 Fractional Factorial Design
 Factors: 8
 Runs: 67
 Blocks: 1
Resolution: IV
Replicates: 2
Fraction: 1/8
Center pts (total): 3
 Design Generators: F = ABC, G = ABD, H = BCDE
Dr. George Box’s analysis
Typical LSS courses
teach to
What is typically
needed
What if we taught statistical based
process improvement to the 95%
that need it and don’t get it?
The real Customer
Un-geeking the teaching process
Hypothesis theory
Gauge R and R
Probability theory
Capability Analysis
Design of Experiments
Hypothesis Theory – Un-Geeking
What is an Alpha error?
Rejection of the null hypothesis when the null hypothesis is true.
What is a Beta error?
Failure to reject the null hypothesis when the null hypothesis is false.
What the jury found
guilty
What really
happened
He did
it
He did not
do it
Not guilty
Alpha vs. Beta Errors
Power
Beta Error
Alpha Error
Reduction in Variation
Probability Theory – Multiplication principle
P(A and B) = P(A) X P(B/A)
Ugly!
….but what if this could be explained by
card magic?
Probability of a pair of kings
P(K and K) = P(first King) x P(second King given the first King) =
.5 % or 5 out of a thousand
… and on a control chart
So what to do with the process?
Nothing!
Probability of drawing three kings
P(K and K) = P(first King) x P(second King given the first King)
x P(third King given the first two)
= .02%
or 2 out of a 10,000
… and on a control chart
….. And what should happen?
What if that was process flowed?
Person
chosen at
random
Random
deck
Made four piles
Split the
cards in the
middle
Handed card
out to audience
Right or left
hand
(50/50).
Audience turns
over cards one
at a time
Shuffled the
cards four
ways.
What if that was probability plotted by
group exercise?
45
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
40
Likely candidates
35
30
25
20
15
10
5
0
shuffle the Random deck Right or left person chosen cut the cards handed out
Audience created four
cards 4 ways
hand?
at random in the middle
cards to
turns over
piles
audience cards one at a
Members
time
Process
Probability of drawing four kings
P(K-K-K-K) = P(first King) x P(second King given the first King) x P(third King
given the first two) x P(fourth given the first three)
= .0000036 or 3.6 PPM
… and on a control chart
What is easier to learn
P(K-K-K-K) = P(first King) x P(second King given the first King) x P(third King
given the first two) x P(fourth given the first three)
= .0000036 or 3.6 PPM
or
Design of Experiments
What if we could all learn the basics of a 2 level – 3 factor DOE?
What if interaction effect was understood by everyone?
Reverse hand experiment
(or the solution may not be obvious and it may hurt)
A very simple but effective DOE
Bowl wrinkle
Bowl Break
The basics of a 2 level – 3 factor DOE
Breaks
Bowl
Wrinkle
Deck
Wrinkles
Possible sweet spot
Additional runs to
define sweet spot
…and it does work
So one more time – the reverse hand
experiment
Probability theory/Capability Analysis
Hypothesis/Probability Theory
Probability Theory
Hypothesis/Design of Experiments
How to make it work – how to make it stick
Finally
Probability of you beating my 4
kings :
.0000005 or
.5 PPM
Bill Hooper
Principle Engineer, Elkay Manufacturing
President, William Hooper Consulting Inc.
Visit his website at www.williamhooperconsulting.org
To learn more about innovative techniques to teach statistics
and process improvement methodology.