Issue: Volume 6, Number 10
Date: October 2006
From: Mark J. Anderson, Stat-Ease, Inc., "Statistics Made Easy" ™ Blog

Dear Experimenter,

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.

For an assortment of appetizers to get this Alert off to a good start, see these new blogs at http://statsmadeeasy.net:
— "Economists shave hairs on whether basketball games are fixed: Any bets on who wins?"
— "Longer-term perspective on global warming (and other catastrophes)," which provides a link to HOW TO LIE WITH STATISTICAL GRAPHICS.
— "Experimental proof that microwaved water kills plants?"
— "Surveys produce precisely inaccurate findings"
— "Stat-Ease credited with reporting the world's largest flying disk" (see photo)
— "Pinning down the possibility of scents enhancing athletic prowess," which details a DOE that tested jasmine versus beer.

Topics in the body text of this DOE FAQ Alert are headlined below (the "Expert" ones, if any, delve into statistical details).

1. Faq: Analyzing historical data collected "kind of randomly"
2. Expert-FAQ: Puzzling backward-elimination results
3. Info Alert: Case-study applications of DOE to
— Assay development (link to poster)
— Materials processing (link to pre-publication overview)
4. Winners of two free copies of the "Coatings Technology Handbook"
5. Reader response: Professor weighs in against the full-normal plot
6. Events alert: Many chances to hear from Stat-Ease this Fall including a one-day presentation in Charlotte, North Carolina
7. Workshop alert: Two-day crash course for sales and marketing

PS. Quote for the month: Why randomizing may be an experimenter's salvation! (See the end of this message for the answer.)

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1. FAQ: Analyzing historical data collected "kind of randomly"

-----Original Question-----
From: Design-Expert® software user in Florida
"I am afraid that my data were not suitable for using the response surface tab. They were collected kind of randomly in Excel format. Is there a way to do the multivariate regression for this type of data?"

Answer:
(From Stat-Ease Consultant Shari Kraber): "Yes, you can. Select the Historical Data option on the
Response Surface tab of Design-Expert (DX). As instructed on the screen, enter the minimum and maximum values for each factor and how many rows of data you collected. Then proceed to create a blank design layout with the factors coded in an ideal manner by DX. Next, copy the input factors and associated responses from your spreadsheet and paste them into the layout in DX. Now you are ready to perform regression analysis using the powerful modeling and statistical tools of Design-Expert software."

P.S. For a detailed tutorial on how this can be done in DX7,* see http://www.statease.com/x70ug/DX7-04H-HistRSM-P1.pdf. Then take a look at the illuminating evaluation detailed at http://www.statease.com/x70ug/DX7-04I-HistRSM-P2.pdf. It shows a worst case of historical data — presented by James Longley in 1967 for "An Appraisal of Least Squares Programs for the Electronic Computer from the Point of View of the User" in the Journal of the American Statistical Association, volume 62, pages 819-841. Read the following FAQ to see another troublesome situation caused by the nature of the data. My advice is that you be extremely careful with models developed on the basis of happenstance results like this. Whenever possible, be proactive (rather than reactive) with your process by designing an experiment using response surface methods (RSM).

"Trying to glean useful information out of happenstance data is akin to resurrecting a gourmet meal from a garbage can."
— Tryg Helseth (Stat-Ease)

*See http://www.statease.com/soft_abs.html for details and link from there to free 45-day fully-functional trials of version 7 of Design-Expert software.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

2. Expert-FAQ: Puzzling backward-elimination results

-----Original Question-----
From: California
"I've got a historical design here that I'm currently working on (historical because it uses some old data, and current because I'm using it to make setpoint predictions and adding points to it). When I run either forward or stepwise elimination, the following terms stay in — A, B, AB, A^2 (all terms significant). When I run backward, it keeps all of the above plus a B^2 term. This term has a p-value of 0.56, so I can't for the life of me figure out why this term is being kept in (it isn't needed for hierarchy). Since two out of the three elimination methods plus my own intuition/experience say no B^2, that's what I'm going with, but I'm wondering why it stays in under backward elimination."

Answer (from Stat-Ease Consultant Wayne Adams): "That is exactly how I like to see our clients use the
significance-based selection methods offered by Design-Expert — backward, forward or stepwise: Try them all and let subject matter knowledge sort them out. What is happening is B^2 is significant on its own, but is NOT significant when A is included in the model due to hierarchy. The model terms A and B^2 are highly correlated (r = -0.965). You can see this via the Evaluation feature of DX. (You must first press the Options button and turn on the correlation matrices.) In your case, if B^2 is in the model first, the A will not be significant on its own, but once A is included due to hierarchy, the B^2 effect is washed out of the
model. Because hierarchy adjustments happen after the selection methods routine B^2 shows as in the model even though it is not significant. Statistically speaking, I cannot tell you if the effect is due to A or B^2 so your subject matter knowledge must be the guide."

(Learn more about empirical model-building by attending the three-day computer-intensive workshop "Response Surface Methods for Process Optimization." For a complete description of this class on RSM, see http://www.statease.com/clas_rsm.html. Link from this page to the course outline and schedule. Then, if you like, enroll online.)

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

3. Info alert: Case-study applications of DOE to:
— Assay development (link to poster)
— Materials processing (link to pre-publication overview)

-----Original Contribution-----
From: James D. Batchelor, Ph.D., Hopkinton, Massachusetts
"I am an application Scientist at Caliper Life Sciences (http://www.caliperls.com). We have implemented Design-Expert into our Kinase Assay development. We presented a poster on it at the Society for Biomolecular Sciences (SBS) conference in September. I know there are some groups using DOE, but it isn't widely known or used by biologists."

See http://www.statease.com/pubs/enzyme_assay_design.pdf.

-----Original Contribution-----
From: Fábio Leão, Rolls-Royce Manufacturing Engineer, UK
"I have finished the experimental phase of my PhD, which I am planning to submit to the Journal of Materials Processing Technology. I have used Design-Expert software throughout my work. I thank you and your team for the invaluable support that I received. I have prepared a summary of the article title "Optimisation of EDM Fast Hole Drilling through an Evaluation of Electrode Geometry." It includes drawings and a picture to illustrate the Electrical discharge machining (EDM) process."

See http://www.statease.com/pubs/EDMdrillopt.pdf.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

4. Winners of two free copies of "Coatings Technology Handbook"

The winners of two free copies of the "Coatings Technology Handbook," 3rd Ed., chosen at random (via a tool in Design-Expert software) from 26 requests, are:
— Bill Jones of CP Kelco, San Diego, and
— John McEwen of Time&Again, Div of Ganz, North York, Canada.
This latest edition of the Handbook, which can be seen at http://tinyurl.com/mfj8t, includes a new chapter (15) on "Design of Experiments for Coatings" authored by Pat Whitcomb and myself. You can see the content in our original manuscript posted at http://www.statease.com/pubs/doe_for_coatings.pdf.

For a complete list of books offered for sale by Stat-Ease, go to its e-commerce site at http://www.statease.com/prodbook.html.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

5. Reader response: Professor weighs in against full-normal plot

-----Original Comments-----
From: Statistics Professor Wei-Yin Loh, University of Wisconsin
"I enjoy reading your DOE FAQ Alerts. I am writing to comment on a question that you asked regarding reader preferences for the half-normal vs full-normal plots (FAQ 1 at http://www.statease.com/news/faqalert6-09.html). A basic problem with the full-normal plot is that it is not unique because it depends on factor codings. Please see my article, "Identification of Active Contrasts in Unreplicated Factorial Experiments" in Computational Statistics and Data Analysis (1992), vol 14, 135-148."

Answer:
I found Professor Loh's article very helpful. He illustrates his point via a two-level design on three categorical factors thought to affect glass substrates used in the manufacture of integrated circuits. In actuality, none of the effects prove to be significant. However, the eight normal plots made by arbitrarily switching levels all differ and one in particular, which stems from the design detailed below, could be mistakenly interpreted to reveal a significant effect. It is amazing to see such compelling patterns from this 2^3 design with no active effects! The half-normal plot of effects remains unchanged throughout. In version 7 of software from Stat-Ease, Design-Ease® as well as Design-Expert, effects are color coded so users can see the negative versus positive, thus removing a disadvantage pointed out by inventor Cuthbert Daniels and other DOE experts.

Here's the most interesting of the eight ways Loh's case can be coded:

A:Operator experience?
-1 Yes
+1 No
B:Sliding process?
-1 No
+1 Yes
C:Flatness (microinches)
-1 <60
+1 60-100

The results in standard order for flatness improvement are:
466
303
573.8
366
244.6
439.8
284.2
257.4

I have the other seven permutations saved in "dx7" format, which I will e-mail upon request.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

6. Events alert: Many chances to hear from Stat-Ease this Fall!

Look over this line-up and see if any of our events scheduled this Fall will be convenient for you to attend and/or affiliated with a conference in your field of expertise:

— Minnesota Quality Conference, Bloomington, October 9-10, exhibit and pre-conference workshop "DOE/RSM Simplified" by me (for details, see http://www.mnasq.org/MNQWorkshopDescriptions.htm)

— National Association of Printing Ink Manufacturers Technical Conference, Ft. Lauderdale, FL, Oct 11, mixture design workshop by Pat Whitcomb (see http://napim.org/TC2006/Tech06Program.htm)

— 50th Fall Technical Conference — Statistics & Quality: 50 Years of Exploration & Discovery, Columbus, OH, October 12-13, exhibit and talk on "A Factorial Design Planning Process" by Shari Kraber and Pat Whitcomb (see http://www.asq.org/cpi/conferences/2006/ and link from there to the program)

— MD&M Minneapolis, October 25-26, booth 954, conference site http://www.devicelink.com/expo/minn06/, click http://tinyurl.com/mu4zh to register for free exhibit hall admission

— Managing Improvement via Modern Design of Experiments, Stat-Ease workshop, Charlotte, NC, November 14, $349.00 (0.8 CEU's), print form at http://www.statease.com/pubs/managing_improvement_DOE.pdf to register, fax when completed to 612.378.2152

Click on http://www.statease.com/events.html for the complete list of appearances by Stat-Ease professionals. We hope to see you sometime in the near future!

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

7. Workshop alert: Two-day crash course for sales and marketing

Paul Selden will present his two-Day Crash Course on DOE for Sales & Marketing ("SMDOE2") on November 8-9 at the Stat-Ease training center in Minneapolis. Please pass along the word to your business colleagues along with this link to all the details: http://www.statease.com/clas_smdoe.html.

Seats remain for two workshops this month, also in Minneapolis:

— Experiment Design Made Easy ("EDME"), October 17-19: See http://www.statease.com/clas_edme.html.

— Mixture Design for Optimal Formulations ("MIX"), October 24-26: Details at http://www.statease.com/clas_mix.html.

If you are uncertain about which workshop would be best, see http://www.statease.com/training.html for the curriculum in hierarchical form. Click any workshop box for course details. Link from there to online enrollment in the next public class.

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 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.

*Believe it or not, it only takes a class of 4 students to make it economical for Stat-Ease to come and teach at your site versus sending them out to one of our public presentations. The economics are detailed in the July 2006 issue of the Stat-Teaser newsletter at http://www.statease.com/news/news0608.pdf.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

I hope you learned something from this issue. Address your general questions and comments to me at: Mark@StatEase.com.

Sincerely,

Mark

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 month Why randomizing may be an experimenter's salvation:

"Designing an experiment is like gambling with the devil: only a random strategy can defeat all his betting systems"

— R.A. Fisher
(found in the appendix to the second edition of the classic book on DOE by George E. P. Box, J. Stuart Hunter, and the late William G. Hunter "featuring quaquaversal quotes from a variety of sources ranging from noted statisticians and scientists to famous philosophers that embellish key concepts and
enliven the learning process." For more details on this new edition from Box, Hunter and Hunter, click this link for the listing at Amazon: http://makeashorterlink.com/?K1F42272B. Heads-up — check out the mixed reviews. Unbelievable!
 
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
—Stat-Ease consultants Pat Whitcomb, Shari Kraber and Wayne Adams (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 and Neal Vaughn (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, #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, #5-12 Dec 05, #6-01 Jan 06, #6-02 Feb 06, #6-03 Mar 06, #6-4 Apr 06, #6-5 May 06, #6-6 Jun 06, #6-7 Jul 06, #6-8 Aug 06, #6-9 Sep 06, #6-10 Oct 06 (see above)

Click here to add your name to the DOE FAQ Alert newsletter list server.

Statistics Made Easy™

DOE FAQ Alert ©2006 Stat-Ease, Inc.
All rights reserved.

 


Software      Training      Consulting      Publications      Order Online      Contact Us       Search

Stat-Ease, Inc.
2021 E. Hennepin Avenue, Ste 480
Minneapolis, MN 55413-2726
e-mail: info@statease.com
p: 612.378.9449, f: 612.378.2152