Volume 5, Number 11
Mark J. Anderson, Stat-Ease,
Here's another set of frequently asked questions (FAQs) about doing
design of experiments (DOE), plus alerts to timely information and
free software updates. If you missed the previous DOE FAQ Alert, please click on the links at the bottom of this page.
If you have a question that needs answering, click the Search tab
and enter the key words. This finds not only answers from previous
Alerts, but also other documents posted to the Stat-Ease web site.
Feel free to forward this newsletter to your colleagues. They can
subscribe by going to http://www.statease.com/doealertreg.html.
If this newsletter prompts you to ask your own questions about DOE,
please address them via mail to:StatHelp@StatEase.com.
Here's an appetizer to get this Alert off to a
I have not tried making a potato gun, but I've heard about it. Christopher
Kimball, Founder and Editor of "Cook's Illustrated" called
me some years ago about making their tests more statistical, but evidently
his chefs consider cooking an art and would not go along with this.
However, I suspect that the sort of person who would make a potato
gun might welcome help with their experimentation.* At this time of
the year,post-Halloween in the USA, pumpkins
naturally become the thing to fling. One such pumpkin gun purportedly
shoots these golden globes, the heavier the better (?), more than
*(For ideas on how to build potato cannons, paper
match rockets, fire kites, tennis ball mortars and the like, read
"Backyard Ballistics" by engineer William Gurstelle: See
Here's what I cover in the body text of this DOE FAQ Alert (topics
that delve into statistical detail are designated "Expert"):
1. Software alert (2nd): New
version of Design-Expert® software
2. FAQ: Determining the minimal
effect for power calculation
3. FAQ: What would be a good
R-squared for a manufacturing study?
4. Expert-FAQ: Follow-up to response
surface method (RSM) design
5. Book alert: Positive reviews
for "RSM Simplified"
6. Workshop alert: California
workshops on DOE
PS. Quote for the month: Encouraging
creativity (for example, when considering variables for an experiment).
1. Software Alert
(2nd): New version of Design-Expert® software
Stat-Ease announces a major new releaseversion 7 of Design-Expert
software (DX7). For a free, fully-functional 45-day trial, click
this link: http://www.statease.com/dx7trial.html.
Pricing for new licenses and upgrades can be seen at the Stat-Ease
e-commerce site: http://www.statease.com/prodsoft.html.
Those of you who've used previous versions will be impressed with
the many improvements in V7, including:
> Pareto chart of t-values of effects: Quickly see the vital
few effects relative to the trivial many from two-level factorial
> Min-Run Res IV (two-level factorial) designs for 5 to 50 factors:
Screen main effects with maximum efficiency in terms of experimental
> Full-color contour and 3D surface plots: Graduated or banded
colorization adds life to reports and presentations.
> Magnification feature: Incredible tool for expanding areas
of interest on trilinear mixture graphs.
This is only a small sample from the features in this landmark
upgrade from Stat-Ease that you will see highlighted at http://www.statease.com/dx7descr.html.
2. FAQ: Determining the minimal
effect for power calculation
From: Design-Expert user and "fanatically loyal
customer of Stat-Ease"
"I was looking at the "Eight Keys to Successful
DOE"an article that you and Shari Kraber wrote
for Quality Digest (posted at http://www.qualitydigest.com/july99/html/body_doe.html).
I have re-read this article several times over the past months.
The 3rd key is "Replicate to dampen uncontrollable variation
(noise)." To illustrate this point via the case study
on injection molding, the article states that "... control
charts reveal a standard deviation of 0.60. Management wants
to detect an effect of magnitude 0.85." How would management
come up with that value?"
First of all, thanks for being such a fan of our company and
its software! To assess how much power will be needed in an
experimental design, and thus the number of runs, it's vital
that the signal to noise ratio be estimated.* I think the
concept of noise is readily grasped by technical professionals.
The tricky part is assessing what signal the experimenter
hopes to detect. This relates to what's considered important
at the bare minimum by the customers and/or management. I
advise bouncing off some numbersbeginning with one that's
ridiculously low. So in this case, ask if your management
would be pleased if your experiment produced a result of 0.01
effect on the response, in this case percent shrinkage of
a molded part. "No way!" they may say. OK, you continue,
"How about a 0.50 response change?" Management now
responds, "Maybe." You then toss out a potential
effect of 1 and get a very definite affirmative that this
would be of interest. Obviously the value of 0.85 used in
the article is somewhat arbitrary, but to calculate power
some number is needed. As a general rule, when signal-to-noise
reaches a level of 1.5 (in this case it is 0.85/.6close
enough), a typical design with 16 runs will likely reveal
even the marginally important effectsif they are produced
(no guarantee of thisdepends whether the right factors
are chosen and the levels set far enough apart).
*(V7 of Stat-Ease software now allows users to enter a specific
signal-to-noise ratio when they do a design evaluation for
(Learn more about power by attending the three-day computer-intensive
workshop "Experiment Design Made Easy." See http://www.statease.com/clas_edme.html
for a course description. Link from this page to the course
outline and schedule. Then, if you like, enroll online.)
3. FAQ: What would be
a good R-squared for a manufacturing study?
"Would 0.5 be a good R-squared for a manufacturing study?
Based on having seen hundreds of analyses over the past three
decades, I'd agree that 0.5 is not an unreasonable benchmark for
R-squared from a manufacturing experiment. However, the real key
is having a low p-value on the overall model ANOVA. I recall Doug
Montgomery talking about how people always pester him about R-squared
and how much he hates this statistic being over-emphasized. After
a day of lecturing at an industrial client, he walked down the
hall and overheard people still talking about their R-squared
results. When these R-squared addicts saw him, the door was quickly
closed so the discussions about this statistic could continue
without Dr. Montgomery intervening! I did some searching on the
Internet and found an article called "R2 is a Big Fat Idiot."
I liked the title, but the article did not prove very helpful.
To see why R-squared is not a reliable statistic, see an article
by Stat-Ease consultant Pat Whitcomb at http://www.statease.com/news/news9709.pdf.
4. Expert-FAQ: Follow-up to response surface method (RSM) design
"After I run an RSM design (such as a central compositeCCD)
I often need to run additional experiments in a region next to my
previous data to explore it more fully. Could you direct me to the
methodology for doing this properly? I thought perhaps you had answered
this question previously. Thank you."
I do not recall fielding a question quite like this before: It is
a good one! When performing evolutionary operation (EVOP), a conservative
move is to keep sequential experiments adjoined to the one done
previously. For example, after replicating two factors such as time
and temperature in a 2^2 design enough to achieve the power needed
for statistical confidence in the path of steepest ascent, the experimenter
would create a new design with at least the corners of the square
regions (DOE #1 and DOE #2) touching.* I infer from your question
that a relatively broadly-ranged CCD for response surface methods
(RSM) may generate a rising ridge, the upper reaches of which you
then want to explore in more detail. The EVOP philosophy dictates
that you take the attitude of "been there and done that"
on the CCD #1 and re-adjust your focus to the new region. My advice
is to set up a new CCD, perhaps with ranges somewhat narrowed, that
stands up on its own with sufficient data to create a new predictive
*(This is pictured at a site maintained by a Swedish consultant:
(Learn more about CCD's by attending the three-day computer- intensive
workshop "Response Surface Methods for Process Optimization."
for a complete description. Link from this page to the course outline
and schedule. Then, if you like, enroll online.)
5. Book alert: Positive reviews for
The October issue of "Quality Progress" magazine provides,
on page 89, a review of "RSM Simplified", a book on response
surface methods co-authored by me and my colleague Pat Whitcomb.
It concludes that "this book is perfect introductory material
for quality and Six Sigma professionals who need to learn the tools
for identifying critical process parameters and optimizing their
Also, I just caught up with the March/April 2005 on-line version
of the "Journal for Healthcare Quality" in which Suzanne
Belanger says*: "Even though I had virtually no idea what I
was reading about, RSM Simplified was entertaining and enlightening...Unless
the reader is engaged in basic research in pharmaceuticals or medicine,
RSM Simplified won't have much applicability to the kind of data
analysis done in most healthcare organizations. Read it for fun."
Hmmm, I would say this is a real compliment!
*(p16 at http://www.nahq.org/journal/online/pdf/webex0305.pdf)
(For more details on "RSM Simplified: Optimizing Processes
Using Response Surface Methods for Design of Experiments" and
sample chapters, see http://www.statease.com/rsm_simplified.html.
From there you can link to an on-line page to order this soft-cover
book, which is accompanied by a free CD-ROM of Design Expert V7
software for educational use.)
6. Workshop alert: California workshops
Stat-Ease makes its annual escape from the cold
Minnesota winter with two presentations of Experiment Design Made
Easy in California:
—Anaheim, December 6–8
—San Jose, January 10–12
If you work elsewhere in the country (nowhere near the west coast
of the USA), your management might be more amenable to funding the
workshop hosted on the 8th through the 10th this month of November
at our Minneapolis training center. These are likely to be great
days for being indoors studying statistics!
for schedule and site information on all Stat-Ease workshops open
to the public. To enroll, click the "register online"
link on our web site or call Stat-Ease at 1-612-378-9449. If spots
remain available, bring along several colleagues and take advantage
of quantity discounts in tuition, or consider bringing in an expert
from Stat-Ease to teach a private class at your site. Call us to
get a quote.
I hope you learned something from this issue.
Address your general questions and comments to me at: Mark@StatEase.com.
PLEASE DO NOT SEND ME REQUESTS TO SUBSCRIBE OR
UNSUBSCRIBE - FOLLOW THE INSTRUCTIONS AT THE END OF THIS MESSAGE.
Mark J. Anderson, PE, CQE
Principal, Stat-Ease, Inc. (http://www.statease.com)
2021 East Hennepin Avenue, Suite 480
Minneapolis, Minnesota 55413 USA
PS. Quote for the
monthEncouraging creativity (for example, when considering
variables for an experiment):
"It is easier to tone
down a wild idea than to think up a new one."
Alex Osborne, "Father of the Brainstorm" according
Pat Whitcomb suggests some other methods to enhance creativity:
—Dr. Edward de Bonos Lateral Thinking methods (see http://www.debonoonline.com/Lateral_Thinking.asp)
—Triz theory of solving inventive problems (see http://www.triz-journal.com/archives/2002/03/a/01a.pdf)
Warning: This is a big file, so it takes a while to download.
are registered trademarks of Stat-Ease, Inc.
Acknowledgements to contributors:
Students of Stat-Ease training and users of Stat-Ease software
Fellow Stat-Ease consultants Pat Whitcomb and Shari Kraber
Statistical advisor to Stat-Ease: Dr. Gary Oehlert (http://www.statease.com/garyoehl.html)
Stat-Ease programmers, especially Tryg Helseth (http://www.statease.com/pgmstaff.html)
Heidi Hansel, Stat-Ease marketing director, and all the remaining
in previous FAQ DOE Alert e-mail newsletters?
To view a past issue, choose it below.
#1 Mar 01, #2 Apr 01, #3
May 01, #4 Jun 01, #5
Jul 01 , #6 Aug 01, #7
Sep 01, #8 Oct 01, #9
Nov 01, #10 Dec 01, #2-1
Jan 02, #2-2 Feb 02, #2-3
Mar 02, #2-4 Apr 02, #2-5
May 02, #2-6 Jun 02, #2-7
Jul 02, #2-8 Aug 02, #2-9
Sep 02, #2-10 Oct 02, #2-11
Nov 02, #2-12 Dec 02, #3-1
Jan 03, #3-2 Feb 03, #3-3
Mar 03, #3-4 Apr 03, #3-5
May 03, #3-6 Jun 03, #3-7
Jul 03, #3-8 Aug 03, #3-9
Sep 03 #3-10 Oct 03, #3-11
Nov 03, #3-12 Dec 03, #4-1
Jan 04, #4-2 Feb 04, #4-3
Mar 04, #4-4 Apr 04, #4-5
May 04, #4-6 Jun 04, #4-7
Jul 04, #4-8 Aug 04, #4-9
Sep 04, #4-10 Oct 04, #4-11
Nov 04, #4-12 Dec 04, #5-1
Jan 05, #5-2 Feb 05, #5-3
Mar 05, #5-4 Apr 05, #5-5 May 05, #5-6 Jun 05, #5-7 Jul 05, #5-8 Aug 05, #5-9 Sep 05, #5-10 Oct 05, #5-11 Nov 05 (see above)
Click here to add your name to the FAQ
DOE Alert newsletter list server.
Statistics Made Easy
DOE FAQ Alert ©2005 Stat-Ease, Inc.
All rights reserved.