Stat-Ease
If you are having trouble viewing this email view it online.
 
Vol: 13 | No: 2 | Mar/Apr'13
Stat-Ease
The DOE FAQ Alert
     
 

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 click here.

To open yet another avenue of communications with fellow DOE aficionados, sign up for The Stat-Ease Professional Network on Linked In. A recent thread features “Best Stats or DOE Blogs.”

 

 

 
Stats Made Easy Blog
 
 

-

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

1:  Awarded: Stat-Ease Founder Patrick Whitcomb honored for his distinguished lifetime achievements in engineering, science and technology
2:  FAQ: Multivariate analysis versus DOE—compare and contrast
3:  FAQ: What if the middle factor level is not exactly on center?
4:  Reader comments: Statistician weighs in on transformations and model hierarchy—issues discussed in the previous DOE FAQ Alert (V13, No. 1, Jan/Feb 2013)
5:  Info alert: Mixture design leads to “exceptionally high level of quality” for a molded plastic
6:  Webinar alert: How to get started with Design-Expert® software
7:  Events alert: Speaking in Indianapolis at the World Conference on Quality and Improvement and in Schenectady for the Quality & Productivity Research Conference
8:  Workshop alert: See when and where (India and San Francisco!) to learn about DOE
 
 


PS. Quote for the month: A classic from the inventor of DOE.
(Page down to the end of this e-zine to enjoy the actual quote.)


- Back to top -


1: Awarded: Stat-Ease Founder Patrick Whitcomb honored for his distinguished lifetime achievements in engineering, science and technology

At their annual dinner on February 22, the Minnesota Federation of Engineering, Science and Technology Societies (MFESTS) awarded Stat-Ease Founder, Patrick Whitcomb, the Charles W. Britzius Distinguished Engineer Award.  He earned these honors by being technically outstanding in his professional field making significant contributions to society through efforts in education and community affairs.  Congratulations, Pat, on this well-deserved recognition!


- Back to top -


2: FAQ: Multivariate analysis versus DOE—compare and contrast

Original Question:

From a Consultant:
“I have recently come across an issue that I would be interested in getting your views on.  This is the issue of using multivariate analysis instead of or in conjunction with the standard DOE approaches when studying the relationship between input factors and responses.  I have come across two companies in recent times that have mentioned they are using multivariate analysis for this purpose.  Both companies, in chemical type industries, can identify large numbers of possible inputs and outputs.   Neither company is currently using DOE.  As I understand it, they are essentially using multivariate analysis with large data sets to seek correlations between inputs and outputs.  In one case, the company has special purpose multivariate analysis software with some DOE facilities.  They tell me that they are interested in learning more about DOE, but consider that they need to use the multivariate analysis approach first in order to try and obtain some knowledge about possible relationships among the many variables that are involved.

I am well familiar with DOE having worked with it now for many years, using Design-Expert most of that time.  I did learn about multivariate analysis when I studied for a degree in statistics many years ago, but I have never made any use of it and am currently very rusty on the subject, and I wouldn't be in a position to debate the merits of using multivariate analysis instead of or in conjunction with standard DOE approaches.

I know it is possible to study the relationship between many inputs and outputs with DOE software such as Design-Expert.  So, I would be interested in your views on the value, if any, of using multivariate analysis where large data sets currently exist, and there are large numbers of input and outputs, perhaps as a prelude to undertaking DOE studies, as one of these companies is doing.”

Answer:

From Stat-Ease Consultant Pat Whitcomb:
“Multivariate analysis is usually just linear regression of historical (i.e. happenstance) data.  Although there is the allure of getting something for nothing, it doesn't often work out.  For a good explanation of why it doesn't often work see section 14.7 in Box, Hunter and Hunter's Statistics for Experimenters (my copy is from 1978) on the “Hazards of Fitting Regression Equations to Happenstance Data.”  They explain such hazards as inconsistent data, range of inputs limited by control, semi-confounding of effects, nonsense correlations due to lurking variables, serially correlated errors, and dynamic relationships.  I think it is worth taking a look at data if it is readily available, but go into it with your eyes open and forewarned.”

(Learn more about regression modeling by attending the two-day computer-intensive workshop Response Surface Methods for Process Optimization.  Click on the title for a description of this class and link from this page to the course outline and schedule.  Then, if you like, enroll online.)


- Back to top -


3: FAQ: What if the middle factor level is not exactly on center?

Original Question:

From a Pharmaceutical Formulator:
“My middle factor level, which I replicated a number of times, is 16.  This falls above the center point of my 14- to 20-factor range.  Can I still use a two-level factorial design?”

Answer:

From Stat-Ease Consultant Brooks Henderson:
“Although this is not ideal, it is OK to run interior points that are not at center.  Our software cannot calculate the test for curvature but it will provide a test on overall lack of fit.  Go ahead and build your factorial design and then for the center points type in the levels you desire.”

P.S. If lack of fit comes out significant, then look over the effect plots to observe where the fitted surface misses the actual points the most.  It may be that the points on the outside are not far off—only the one in the middle.  If this is the case, then it would suggest that response surface methods (RSM) may be needed to adequately approximate the curvature. —Mark

(Learn more about center points by attending the two-day computer-intensive workshop Experiment Design Made Easy.  Click on the title for a description of this class and link from this page to the course outline and schedule.  Then, if you like, enroll online.)


- Back to top -


4: Reader comments: Statistician weighs in on transformations and model hierarchy—issues discussed in the previous DOE FAQ Alert (V13, No. 1, Jan/Feb 2013)

From Nico Laubscher, Statistical Consultant, InduStat Pro, South Africa
Re FAQ #2 (Do I need a rationale for applying a response transformation?):

“When doing an ANOVA, two of the assumptions for the validity of the F-tests are that the data must be normally distributed and that the ‘within’ cells variance must be constant.  The original idea behind transformations was to ‘stabilize the variance,’ i.e. to ensure that the variance of the transformed data is not dependent on the mean.  (See Bartlett, 1936, who applied the square root transformation).  The Box-Cox transformation family (1964) was constructed to achieve this but fortuitously also that the data are “normalized” (i.e. made Gaussian).  This was pointed out by Kendall in his classical work.  So, the rationale behind the Box-Cox transformation is to stabilize the variance and, simultaneously, to satisfy the assumption of normality.

Re FAQ #3 (Why does Design-Expert request that users maintain model hierarchy?):
“Further clarity may be obtained on this issue by recalling that this deals with fitting a model: Compare it to fitting a parabola to a set of data: Y = a + bX + cX2.  Should you decide to fit Y = a + cX2, rather than the hierarchical model, the model is forced to have (in this example) a specific axis of symmetry, even before it sees the data!  This means a poorer fit (whether the effect of “X” is significant or not).  A model with fewer parameters also cannot provide a better fit (i.e. smaller MSE) than one with more parameters. That is not to say that one should over-parameterize.

I personally never hesitate to use the hierarchical model.”


- Back to top -


5: Info alert: Mixture design leads to “exceptionally high level of quality” for a molded plastic

Check out this Cure for Cratering developed by engineers at resin manufacturer Interplastics with the aid of Design-Expert software.  This case study on the application of mixture design for optimal formulation was posted by Composites Technology from their February publication.

(Learn more about mixture design by attending the computer-intensive two-day workshop “Mixture Design for Optimal Formulations.”  Click on the title for a complete description of this class.  Link from this page to the course outline and schedule.  Then, if you like, enroll online.)


- Back to top -


6: Webinar alert: How to get started with Design-Expert® software

On Wednesday, March 20 at 12 PM (noon) CST* Stat-Ease Consultant Shari Kraber will detail How to Get Started with Design-Expert® Software in a free webinar geared toward novices.  Her presentation will demonstrate how to plan, design and analyze a powerful multifactor test. She will also provide a ‘heads-up’ on mistakes made by unsuspecting beginners that lead to DOE failures. The goal of this webinar is to set you up for success using Stat-Ease software for your experiments!

Shari will reprise this webinar at 9 PM USA-CST on Monday, March 25—this time particularly for those on other side of the globe (India, AustralAsia, Far East) from our headquarters in Minneapolis, Minnesota, USA.

Space is limited.  Reserve your Webinar seat now at by clicking one of the links below:

  1. March 20 (12 PM Noon CST) sign-up
  2. March 25 (9 PM CST) sign-up

If this is your first Stat-Ease webinar, see these suggestions on how to be prepared.

Stat-Ease webinars vary somewhat in length depending on the presenter and the particular session: Plan for 45 minutes to 1.5 hours, with 1 hour being the target median.  When developing these one-hour educational sessions, our presenters often draw valuable material from Stat-Ease DOE workshops.

Again, attendance may be limited, so sign up soon via the link above.  Direct any questions you may have to our Communications Specialist, Karen Dulski, via karen@statease.com.  However, if this relates to audiovisual issues, please first research help provided online by GotoWebinar.

*(To determine the time in your zone of the world, try using this link.  We are based in Minneapolis, which appears on the city list that you must manipulate to calculate the time correctly.)


- Back to top -


7: Events alert: Speaking in Indianapolis at the World Conference on Quality and Improvement and in Schenectady for the Quality & Productivity Research Conference

On Monday morning May 6 in Indianapolis I will provide a briefing on Quality by Design (QbD) for Pharmaceuticals and Beyond to members of the American Society of Quality (ASQ).  This talk is sponsored by the Institute for Continual Quality Improvement (ICQI) which runs an annual session concurrent to the World Conference on Quality and Improvement (WCQI).  Quality by Design (QbD) is a hot topic in the pharmaceutical industry, heavily promoted by the U.S. Food and Drug Administration (FDA) and similar agencies in Europe.  However, these tools should be used by every industry interested in producing high-quality products.  The general concepts are not new, but the tools to implement them have dramatically improved in the last few years.  This presentation provides a briefing on QbD along with state-of-the art response surface methods (RSMs) for developing a robust design space.  I hope you can attend my talk and/or stop by our booth at this year’s WCQI.

Please stop by the Stat-Ease table if you attend Quality & Productivity Research Conference (QPRC) in Schenectady, NY on June 5-7.  See meeting details here.

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

PS.  Do you need a speaker on DOE for a learning session within your company or technical society at regional, national, or even international levels?  If so, contact me.  It may not cost you anything if Stat-Ease has a consultant close by, or if a web conference will be suitable.  However, for presentations involving travel, we appreciate reimbursement for travel expenses.  In any case, it never hurts to ask Stat-Ease for a speaker on this topic.


- Back to top -


8: Workshop alert: See when and where (India and San Francisco!) to learn about DOE

Seats are filling fast for the following DOE classes.  If possible, enroll at least 4 weeks prior to the date so your place can be assured.  However, do not hesitate to ask whether seats remain on classes that are fast approaching!  Also, take advantage of a $395 discount when you take two complementary workshops that are offered on consecutive days.

All classes listed below will be held at the Stat-Ease training center in Minneapolis unless otherwise noted.

*Take both EDME and RSM to earn $395 off the combined tuition!

** Take both MIX and MIX2 to earn $395 off the combined tuition!

See this web page for complete 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 Elicia at 612-746-2038.  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. Once you achieve a critical mass of about 6 students, it becomes very economical to sponsor a private workshop, which is most convenient and effective for your staff.  For a quote, e-mail workshops@statease.com.


- Back to top -


Please do not send me requests to subscribe or unsubscribe—follow the instructions at the very end of this message.
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.
2021 East Hennepin Avenue, Suite 480
Minneapolis, Minnesota 55413 USA


PS. Quote for the month—a classic from the inventor of DOE:


"
“Every experiment may be said to exist only in order to give the facts a chance of disproving the null hypothesis."
—Ronald Fisher, 1935

Trademarks: Stat-Ease, Design-Ease, Design-Expert and Statistics Made Easy 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, Wayne Adams and Brooks Henderson
—Statistical advisor to Stat-Ease: Dr. Gary Oehlert
Stat-Ease programmers led by Neal Vaughn
—Heidi Hansel Wolfe, Stat-Ease marketing director, Karen Dulski, and all the remaining staff that provide such supreme support!

Twitter-SmileyFor breaking news from Stat-Ease go to this Twitter site.

DOE FAQ Alert ©2013 Stat-Ease, Inc.
Circulation: Over 6300 worldwide
All rights reserved.

 
  Subscribe