Issue: Volume 3, Number 3
Date: March 2003
From: Mark J. Anderson, Stat-Ease, Inc.
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 previous DOE FAQ Alerts, click on the links below. Feel free to forward this newsletter to your colleagues. They can subscribe by going to http://www.statease.com/doealertreg.html.
Here's some appetizers to get this
Alert off to a good start. I found them both in the March 2003 issue of "Fast
Company" (http://www.fastcompany.com).
They represent some amusing applications of statistical tools to emotional well-being.
At this time of the year, we in the frozen north of the USA suffer from what's
called "cabin fever," a result of being trapped indoors for long periods.
Perhaps keeping statistics on our mood may alert us and our neighbors before
things go out of control and we become deranged.
a. Click http://www.moodstats.com/pages/about.asp
to learn about a program called Moodstats that generates "statistics showing
you exactly how your moods have been over the last week, month, two months,
six months or year." Click the Download link for a free demo.
b. Andrew Oswald, an economics professor at University of Warwick (http://www.warwick.ac.uk) (update--3/07: new link is http://www2.warwick.ac.uk/fac/soc/economics/staff/faculty/oswald/ijeclarkos.pdf)
reports the following effects on happiness as a function of life's events. Via
application of statistical regression on survey data, Oswald quantified impacts
in monetary terms. Here are a few of his findings:
-$60,000 for loss of job
$100,000 for Marriage
$0 for children
This last result was "one of the more surprising results" said Oswald.
He goes on to say that "there's no value judgment implied...people without
children recorded equally high happiness levels as people with children."
That's interesting! The big negative impact for loss of job makes perfect sense.
I'm not so sure about the value for marriage. For example, a local couple here
in Minnesota by the name of Ole and Lena may not be doing so good. Lena won
the lottery the other day and ran home to tell Ole to get packing. He asked
if they would be heading south for some warm weather and whether to pack the
beach stuff. Lena said: "It don't matter - just pack and get out!"
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: A free update is available for V6.09
of Design-Expert® and Design-Ease® software (if not currently a user,
get the free trial version - now good for 45 days!)
2. FAQ: What should be done if probability ("P")
values fall in an intermediate range such as 0.07?
3. FAQ: Does it make sense that a model is statistically
insignificant when some individual terms are significant?
4. Reader feedback: Gossett's work at the Guinness Brewery
- some history on the development of statistics for DOE
5. Heads-up ("Fore!") on a DOE to do in class: Professor
Lye's golfing machine
6. Info alert: "Using Design of Experiments to Make
Processes More Robust to Environmental and Input Variations," published
in the February issue of "Paint and Coatings Industry" (PCI)
7. Events alert: Annual Quality Congress (Kansas City)
and other places you will find Stat-Ease
8. Workshop alert: Upcoming classes in Philadelphia and
Minneapolis
PS. Quote for the month - the role of statistics in the scientific method.
1 - Software Alert: A free update is available for V6.09 of Design-Expert or Design-Ease software (if not currently a user, get the free trial version - now good for 45 days!)
If you (as an individual user) own a permanently licensed copy of version 6 of Design-Ease (DE6) or Design-Expert (DX6), go to http://www.statease.com/soft_ftp.html#dx6updt for downloads that will patch your software with the latest enhancements. We recommend you do this even though the changes may affect only a few users. To see what got added or fixed, click the "Read Me" link for either DE or DX (whichever program you will be updating).
If you own networked software that needs updating, you must call Stat-Ease customer service at 1.612.378.9449. We do not post patches for networked software on the Web. Be prepared to provide your serial number. We will then send you a replacement CD-ROM to re-install on your network. Here's some details from my partner Tryg Helseth (Principal and VP Programming for Stat-Ease) on specific changes related to how V6.09 supports networking.
"Prior to this latest release, the built-in network metering for DE and DX was controlled by a third-party product called "softSENTRY." That software, and support for it, has been discontinued by the vendor so we needed to find a replacement. Starting with version 6.0.9 the built-in metering is now handled by "Protection PLUS." This new program corrects the following problems with the old system:
- Network license lockup
If the program terminated abnormally, softSENTRY did not release the license
for other users. The PLUS system will release the unused licenses.
- Restricted user issues
If the user had limited privileges on his own computer he might get the error,
"cannot save message," when trying to launch the program. The old
softSENTRY program attempted to write information to protected system areas
on the disk which the user didn't have access to. Protection PLUS does not attempt
to write data to the local system areas and thus will work with restricted user
privilege.
- Launch from a data file
On many systems you could not launch the network version by clicking on a data
file. The error message "cannot write meter file" would appear as
the system could not locate the control folder. This has been corrected."
Before updating or buying version 6 of Design-Expert, feel free to download a fully-functional, but 45-day limited, version of the software from http://www.statease.com/dx6trial.html. Users of Design-Ease or earlier versions of Design-Expert (V5 or before) should consider upgrading their software to DX6. See why you should do it at http://www.statease.com/dx6descr.html. Then call Stat-Ease for a quote. After validating your current serial number, we will give you a special upgrade price.
2 - FAQ: What should be done if probability ("p") values fall in an intermediate range such as 0.07?
-----Original Question-----
From: Los Angeles
"For the difference between means [in a simple comparative experiment on
one categorical factor] it is said [in Stat-Ease software] that:
- "Values of "Prob > |t|" less than 0.0500 indicate the difference
in the two treatment means is significant.
- "Values of "Prob > |t|" greater than 0.1000 indicate the
difference in the two treatment means is not significant."
What if the p-value is greater than 0.05 and less than 0.01, for example 0.07?"
Answer:
For fear of being too arbitrary on the "p" values, we left some room
for interpretation. In cases like you suggest where the p-value falls between
0.05 and 0.1, I advise that you look at the effects plots and see whether they
make sense from your subject matter knowledge. If not, you may decide not to
act upon the results. I recall seeing at a conference of statisticians a saying
that relates to this dilemma: "Statistics means never having to say you
are certain."
For more background on p-values, see my "DOE FAQ Alert" Volume 1, Number 4 from June 2001, item 1 FAQ "Why choose a probability of 0.05 (p-value) as the criteria for statistical significance?" posted at http://www.statease.com/news/faqalert4.txt.
(Learn more about p-values in a new
web-based course called "PreDOE" that helps people get over "statistics-phobia"
before embarking on further education in design of experiments. The class content
resides at http://www.statease.net. You can assess your abilities by first taking the free self-assessment
questionnaire—which is available as a pre-test within the PreDOE course. Contact Stat-Ease if you want to learn more
about this exciting new educational opportunity or you need help getting going
on it.)
3 - FAQ: Does it make sense that a model is statistically insignificant when some individual terms are significant?
-----Original Question-----
From: New York
"I should start by stating that I am an undereducated novice at statistics
and DOE. I am currently using Design-Expert 6.0 to analyze a two-level full
factorial design with 3 factors. For some of the responses I have a "not
significant" model but one or more "significant" model terms.
Does the failure to have a "significant" model (based on the model
F value) make the identification of significant model terms meaningless?"
Answer (from Shari Kraber, Stat-Ease
Statistical Consultant):
"First of all, I don't think that there is any other status than "undereducated"
in terms of statistics! There is ever more to learn. In response to your question,
if the model is truly not significant, then you should not use the individual
terms. However, it is likely that by removing the non-significant factors from
the model, you can then create a significant model containing only the significant
terms. Furthermore, there are always the fun possibilities that a transformation
or removing an outlier might help with the modeling."
(Learn more about analyzing factorial designs by attending the 3-day computer-intensive workshop "Experiment Design Made Easy." See http://www.statease.com/clasedme.html for a complete description . Link from this page to the course outline and schedule. Then, if you like, enroll online.)
4 - Reader feedback: Gossett's work at the Guinness Brewery - some history on the development of statistics for DOE
----- Original Message -----
From:Geoff Cole, Specialist - Reliability Modelling, Rolls-Royce Defence(Europe)
"I hope you find the reply I've just received from Dr. Bob Abernethy below
(of Win-Smith Weibull fame) of some help with your studies on linking the efficient
brewing of beer with DoE analysis from your last infosheet! He's an expert on
DoE, the history of Mr Gossett's work at the Guinness Brewery, Sir Ronald Fisher
et al!!"
-----Original Message-----
From: Dr. Robert Abernethy, Florida
Re: DOE FAQ Alert, Volume 3, Number 2 - February, 2003
Gossett was a leader in DOE in his time. His Student's t and F tests were specifically
aimed at his DOE on the taste and quality of beer. He was the principal arbitrator
between Fisher and Karl Pearson, the world's two greatest experts on DOE. For
example he argued with Fisher that in many cases Pearson's systematic arrangements
were better than Fisher's randomization. He was also singularly responsible
for getting both of them into the small sample business which they had both
neglected until Gossett prodded them. He wrote papers on DOE under the pseudo
name of Student.
The reason Mark cannot find references is the same reason Gossett was forced
to use the name Student. The beer industry wants the world to believe we still
make beer in wooden barrels, they do not want the image of high technology and
they are highly proprietary of the methods, including DOE, that they use.
The little book, "Student" is a good source for all of the above.
I do a lot of research on the subject of beer......
Bob
(Dr. Abernethy offers his "The
New Weibull Handbook" plus "ALL THINGS RELATED to RELIABILITY, MAINTENANCE,
TESTING, COST REDUCTION, ETC" at http://www.weibullnews.com.
Check it out!)
5. Heads-up ("Fore!") on a DOE to do in class: Professor Lye's golfing machine
From: Dr. Leonard M. Lye, PEng,
FCSCE, Professor, Faculty of Engineering and Applied Science, Memorial University
of Newfoundland, Canada
"Your idea of teaching hands-on toys like the hockey experiment in your
book ["DOE Simplified" pages 73-81*] really works very well. Following
up on that I have used two more toys in my DOE class. One is the well-known
catapult and another is a golfing machine I have invented. I find that the golf
machine works really well because the students can play a golf tournament with
it, naturally based on the predictive equations they get via their experiments.
The golfing toy can consider up to 5 or 6 factors (e.g. brand of ball, length of club, weight of club, angle of swing, type of greens (carpet), direction (floor may not be flat), etc. Students can design FFD, RSM (CCD +BBD) and other designs using the toy. I have made two of them and they can be used for blocking experiments also. Of course the students do all the design and analysis using Design-Expert software.
The team that requires the least number of strokes to get to within 0.5 inches of a line is the winner and I used 3 lines in different directions. Also, we played a 3-hole tournament with it. The team that won had a well-designed experiment and even had a hole-in-one!
Attached please find a write-up about the golfing toy. I wrote it as an interview with myself just for fun. I have included more pictures as well. Also attached is the DX6 file for an experiment I did on my home carpet. I used 3 reps and a face-centered CCD. It gave excellent predictions. I can take on Tiger Woods anytime - on my home carpet of course!"
Comments:
See Professor Lye's write-up, photos and data at http://www.statease.com/golftoy.html.
There you will find details on how to buy one of these "toys" for
teaching DOE.
Mark
*For details and links to online purchasing of the "DOE Simplified" book see http://www.statease.com/doe_simp.html.
6 - Info alert: "Using Design of Experiments to Make Processes More Robust to Environmental and Input Variations," published in the February issue of Paint and Coatings Industry" (PCI)
Click http://makeashorterlink.com/?M1DB23DA3 to see a publication entitled "Using Design of Experiments to Make Processes More Robust to Environmental and Input Variations" by me and Shari Kraber. The article, which appeared in the February 2003 issue of "Paint and Coatings Industry" (PCI) Magazine, explains why standard factorial designs (one array) offer a cost-effective alternative to parameter designs (two array) made popular by Taguchi.
(Learn more about parameter design and its alternatives by attending the 3-day computer-intensive workshop "Robust Design, DOE Tools for Reducing Variation." For a complete description see http://www.statease.com/clasrdrv.html. Link from this page to the course outline and schedule. Then, if you like, enroll online.)
7 - Events alert: Annual Quality Congress (Kansas City) and other places you will find Stat-Ease
Mark your calendars and make plans to come to American Society of Quality's (ASQ) 57th Annual Quality Congress (AQC) in Kansas City for talk T310, "How to Use Graphs to Diagnose and Deal with Bad Experimental Data", presented by me on Tuesday May 20 from 2:00 PM - 3:15 PM. The material, co-authored by fellow Stat-Ease consultant and Principal Pat Whitcomb, is rated "Basic." It deals with thorny issues that confront every experimenter - how to deal with individual results that do not appear to fit with the rest of the data. I will show graphical tools that diagnose whats really wrong with questionable data: damaging outliers and/or a need for response transformation. Stop by our booth (#115) to talk with me and other representatives of Stat-Ease. We'd like to see you! Go to http://aqc.asq.org/ for details on how to register for the exhibit and conference.
Also, we've been asked to pass along
an alert for the Conference on New Directions in Experimental Design to be held
in Chicago, Illinois on May 14-17, 2003. The focus of the conference will be
on design of experiments in the pharmaceutical and related industries. See http://www.math.uic.edu/~kjryan/dae2003.html (Update 3/07: Link no longer available)
for more details.
Click http://www.statease.com/events.html
for a listing of where Stat-Ease consultants will be giving talks and doing
DOE demos. We hope to see you sometime in the near future!
8 - Workshop alert: Upcoming classes in Philadelphia and Minneapolis
Experiment Design Made Easy (EDME) will be presented April 1-3, in Philadelphia, PA. This class and many others will also be available in the Stat-Ease home training center in Minneapolis. See http://www.statease.com/clas_pub.html 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 questions and comments to me at:
Mark J. Anderson, PE, CQE
Principal, Stat-Ease, Inc. (http://www.statease.com)
Minneapolis, Minnesota USA
PS. Quote for the month - the role
of statistics in the scientific method:
"Statistics is the guts of scientific method. As such,
statistics is useful for solving just about any type of problem. To be effective,
statistics must be understood and put to use, not just by a few experts, but
as widely as possible by everybody, everywhere."
- Professor Soren Bisgaard, University of Massachusetts (see http://intra.som.umass.edu/fomgt/Faculty/Bisgaard.htm no longer a faculty member),
in remarks made while accepting the 2002 American Society of Quality (ASQ) Statistics
Division's Hunter Award
Trademarks: Design-Ease, Design-Expert and Stat-Ease 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 (see http://www.statease.com/consult.html
for resumes)
- 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 staff.
Interested 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 (see above)