Issue: Volume 3, Number 10
Date: October 2003
From: Mark J. Anderson, Stat-Ease, Inc. (

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 previous DOE FAQ Alerts, please click on the links at the bottom of this page. Feel free to forward this newsletter to your colleagues. They can subscribe by going to If this newsletter prompts you ask to your own questions about DOE, please address them to

Here's some appetizers on scientific graphs to get this Alert off to a good start:  

A. For the better, see (article requires subscription to view). Note the advice that you should never use bar graphs.
B. For the worse (humorous satire on what NOT to do), see For example, they advise you to try putting the dependent variable (the one you measured) onto the x-axis because your readers will "appreciate a bit of a variation from the norm."

Here's what I cover in the body text of this DOE FAQ Alert (topics that delve into statistical detail are designated "Expert"):  

1. FAQ: Viewing two main effects on one interaction graph  
2. Expert-FAQ: Why no PRESS or Predicted R-squared statistics?  
3. Reader response: A question regarding the September Alert, topic #2, on outliers  
4. Info alert: A case study on the application of DOE to defect reduction in a powder coating (links are provided to the article)  
5. Events alert: Link to a schedule of appearances by Stat-Ease
6. Workshop alert: "Experiment Design Made Easy" is coming to Anaheim

PS. Quote for the month: A skeptical comment about graphs.


1. FAQ: Viewing two main effects on one interaction graph

-----Original Question-----
From: Toronto, Canada

"If a response has two significant factors, is it appropriate to overlay the two curves (put the curves in one graph as if there is an interaction)?  I don't see a problem doing this; the curves for the two factors will parallel each other.  I just want to check your thoughts on it."

Answer (from Stat-Ease Consultant Shari Kraber):

"You can do this via the View, Interaction feature in Design-Ease® or Design-Expert® software.  The parallel lines are simply your indication that the interaction is not significant." (That was easy! Mark)

(Learn more about interaction effects by attending the three-day computer-intensive workshop "Experiment Design Made Easy."  See for a complete description. Link from this page to the course outline and schedule.  Then, if you like, enroll online.)


2. Expert FAQ: Why no PRESS or Predicted R-squared statistics?

-----Original Question-----
From: San Jose

"A colleague and I took your class a few weeks ago in San Jose and then did a DOE.  Upon analyzing the results via the ANOVA, we were surprised to see it showing "N/A" for PRESS and Pred R-Squared. Any ideas on why we are not getting a value?  We used a log transform and selected A,B,C,D, AB, AD and BD as significant terms."


Very interesting!  You only performed one center point, so when Design-Expert fitted the term for curvature (done by default), it had a leverage of 1.  Therefore, if it had been removed per the PRESS/R^2pred) process,* the curvature term would no longer be estimable.  Design-Expert was smart enough to recognize this and not even try calculating these two related statistics.

To fix this I went to the View, Effects list and changed Curvature from being modeled ("M") to an error ("e") term.  I also found that only main effects were needed if the transformation is changed from Log to the 4th root (0.25 power), which Box-Cox indicated.  It said to do Log because the minimum residual point is so sharp that the red zone (95% confidence on optimum) fell cleanly between log (represented by the 0 on the x-axis) and the square root (0.5 on the power scale).  Design-Expert cannot discriminate any tighter than this for purposes of making a general recommendation.

*The Help system in Design-Expert (DX) software explains PRESS as follows:

>A measure of how a particular model fits each design point.  The coefficients for the model are calculated without the first design point.  This model is used to predict the first point and then the new residual is calculated for this point.  This is done for each data point and then the squared residuals are summed.<

In other words, if you performed N runs in your DOE, each response is removed before re-fitting the model equation to the remaining N-1 runs.

DX Help provides these details on Predicted R-squared:

>This is a measure of how good the model predicts a response value.  It is computed as 1-(PRESS/ SStotal)<

[PS. Those of you readers who consider themselves experts on analyzing DOE data may enjoy looking at this. The student/user gave me permission to share it.  Set up a standard 2^6-2 (one- quarter fraction of a two-level factorial), with 1 center-point, for a total of 17 runs.  Sort by standard order and enter the following responses via copy/paste:


Then analyze to your heart's delight!]

(If you want to hone your skills on factorial design, bring a Stat-Ease consultant in for a private presentation of the "Real-Life DOE: Tricks of the Trade" (in-house only) workshop.  For a description, see  Link from this page to the course outline.  Call 1.612.378.9449 and ask for a quote to bring this workshop to your site.)


3. Reader response: A question regarding the September Alert, topic #2, on outliers

-----Original Question-----
From: Singapore

"Mark, I would like to ask your advice regarding Topic 2, "Rule-of-thumb for assessing outliers."  If the outlier is caused by typographical error, then one should simply re-type the correct value.  But, if it is due to mechanical breakdowns, then it is clear to me that the particular results associated to the mechanical breakdowns should be ignored.  What would you recommend to do if there is no clear reason for the outlier?  Is it a correct decision to ignore the values for the outliers if a reason cannot be found to explain them?  Or should the values be retained and assume that it is part of the variation although that variation is exceptionally different from the rest of the results?"


I agree that an outlier due to a typo should simply be corrected. Other statistical outliers should be investigated for special physical causes, such as mechanical breakdown.  If this is established, ignore the discrepant value in the analysis. Otherwise, try analyzing with and without the questionable value. It may make no difference, but if it does change things, then you must think this over and be careful about making any definitive judgments without further experimentation.


4. Info alert: A case study on the application of DOE to defect reduction in a powder coating (links are provided to the article)

The April 2003 issue of "Powder Coating" magazine (Vol 14, #3, pp 12-15) features the article "DOE Software Paints Picture of Powder Coating Defects," which details how Morton Powder Coatings in Reading, Pennsylvania used design of experiments to deal with a situation that defied conventional problem-solving techniques.  An abstract is at with an option to purchase the complete article.  A draft version is available for free at


5. Events alert: Link to a schedule of appearances by Stat-Ease

(Second notice) This month we set up shop at:
- Fall Technical Conference, El Paso, TX, October 16-17 (see - site no longer available - for details)  
- Medical Design and Manufacturing (MD&M) Show, Minneapolis, MN, October 29-30, 2003, at Booth #1816 (for details, see )

Click for a complete list of appearances by Stat-Ease professionals.  We hope to see you sometime in the near future!


6. Workshop alert: Experiment Design Made Easy coming to Anaheim

"Experiment Design Made Easy" (EDME), our most popular workshop, has nearly sold out for its next presentation on October 21-23 in Minneapolis, so you may want to set your sights on the next one on November 18-20 class in Anaheim.  Those of you in the Bay Area may want to hold out for the EDME in San Jose on January 20-22, 2004.

We are also seeing a lot of interest in the more advanced "Response Surface Methods for Process Optimization" workshop coming up on October 28-30 in Minneapolis.  If you need RSM training, don't delay—call today.

See 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 J. Anderson, PE, CQE
Principal, Stat-Ease, Inc. (
Minneapolis, Minnesota USA

PS. Quote for the month—A skeptical comment about graphs

"Figures won't lie, but liars will figure."
—Charles H. Grosvenor

Trademarks: Design-Ease, Design-Expert and Stat-Ease are registered trademarks of Stat-Eae, 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 for resumes)
—Statistical advisor to Stat-Ease: Dr. Gary Oehlert (
—Stat-Ease programmers, especially Tryg Helseth (
—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 (see above)

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

Statistics Made Easy™

DOE FAQ Alert ©2003 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
p: 612.378.9449, f: 612.378.2152