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

P.S. Quote for the month: A cutting comment on certainty (scroll down to view)

1: FAQ: How Stat-Ease augments optimal designs to make them more robust

Original question from a US Air Force Equipment Tester:
“Please explain the inputs offered by Design-Expert^® software for the Optimal (Custom) design choice on its Response Surface tab [see screen shot]:

• How are “Lack of fit” points selected? Can they be identified in the final design matrix? Why is the default 5?
• How do you determine which points to replicate? Why is the default 5?”

Defaults for augmenting an Optimal (Custom) Design for Response Surface experiment

Answer from Stat-Ease Consultant Wayne Adams:
“Lack-of-fit (LOF) points are picked to maximize the minimum distance to another point. This is done on the “five” points as a whole. The goal is to fill the gaps in the design. Version 9 provides the option in the design layout to select a Build Point Type column [see screen shot] to identify LOF and other points by their role in the build.

Selecting Build Point Type for display in design layout

Points are replicated to support the optimality criterion. If I-optimal is chosen, then the replicates will be the points that offer the most gain/least detriment to the I-optimal criterion.

The default of 5 points for lack of fit and replicate points provides a practical tradeoff in cost of runs versus power. If you look at an F-distribution table, you will see that as the degrees of freedom are increased, the critical value decreases. At about 5 df, the rate of change in the F-critical value begins to flatten out on both the numerator and denominator axes. This is a sweet spot. However, if runs are very expensive, then cutting LOF and replicate points to 3 might be chanced. This is the practical lower limit for meaningful tests in our opinion."

(Learn more about optimal design 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.)

2: Book alert: Third Edition of DOE Simplified published—now available as an e-book

I am happy to announce the publication by Productivity Press of DOE Simplified, 3^rd Edition—coauthored by my colleague Pat Whitcomb. See details at the CRC Press (Taylor & Francis Group) web site. I really like the cover showing an intense experiment designer, the traditional bench chemist and a diverse team of manufacturing process improvers.

Beyond the cool cover the new edition is updated throughout from the prior publication in 2007, including a whole new chapter on factorial split plots for handling hard-to-change factors. It ties in with version 9 of Design-Expert software which readers can download freely for use while working through the practice problems.

Although it’s intended for self-study by industrial experimenters, DOE Simplified serves well in conjunction with computer-intensive courses and workshops.

“I would like to use the 3^rd edition of your DOE Simplified textbook. (I have reviewed a copy and it looks good). I want to match your new text by installing DX9 on our classroom PC’s.  (I have about 18 grad students and 6 undergrads each fall semester). I love to teach with your software—it does make it simple! Best regards to you and your team at Stat Ease.”
— David Gore, PE, Associate Professor, Engineering Technology Dept., Middle TN State University

DOE Simplified, 3^rd Edition can be purchased in paperback or e-book from the publisher or via Amazon and other major booksellers. It is also available for purchase from Stat-Ease. Based on all the positive feedback this book has received, going back to the 1^st edition published in 2000, I feel sure you will find the book fun and informative.

3: Webinar alert: “Split-Plot Pros and Cons”

In this webinar titled “Split-Plot Pros and Cons” (presented three times for your scheduling convenience), Stat-Ease Consultant Pat Whitcomb illustrates how to take best advantage of factorial designs geared for hard-to-change process settings. While running through a number of case studies with Design-Expert software, he will provide statistical detail and practical advice on the pluses and minuses created by the split-plot factor layout.

1. Monday, August 17 at 8 pm USA-CDT* for eastern Asia and Oceania (others welcome!)
2. Wednesday, August 19 at 11 am USA-CDT* for the Americas and Caribbean (others welcome!)
3. Thursday, August 20 at 6:30 am USA-CDT* for Europe, Africa, Middle East and western Asia (others welcome!)

If this is your first Stat-Ease webinar, please review these suggestions on how to be prepared. If questions remain, direct them to our Client Specialist, Rachel Pollack, via [email protected].

*(Note that the USA went on daylight savings time, “DT”, on March 8 this year—2015. 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.)

4: Reader response: An alternative explanation as to why unimportant factor effects fall on a straight line in the half-normal probability plot

Comment from Nico Laubscher, Statistical Consultant, InduStat Pro, South Africa:

RE: DOE FAQ Alert 15 (3) June 2015 FAQ 3

“Thanks for the (always) interesting Newsletter. I have read Brooks’ explanation to the question about the straight line for effects on the normal or half-normal probability plot. Perhaps the following few words may be added to it to help the engineering student who asked the question.

Each effect is a linear combination of the raw data obtained from the design. Linear combinations of sampled data, for all practical purposes (and even for smallish sample sizes) are approximately normally distributed under some mild conditions as specified by the central limit theorem. If the estimated effects from an experiment are standardized and all have effect value of zero they should thus follow a N(0, 1) distribution approximately.

The vertical scale of the normal probability plot is designed such that all points from a N(0, 1) distribution will lie exactly on a straight line. Effects that are insignificant (close to zero) should thus lie close to that line and the significant effects will deviate meaningfully from it. The half-normal plot does exactly the same but for the absolute values of the effects, which will show the significant ones on one side of the graph, rather than on both sides.”

5: Events alert: Conferences in USA and Europe

At the Society for Industrial Microbiology & Biotechnology August 2-6 Annual Meeting in Philadelphia, PA, I will present a “Strategy of Experimentation for Fermentation Process and Product Development” and exhibit our software. Register here.

Consultant Pat Whitcomb will provide guidance on “Confirmation—The Final DOE Step” for the Annual Conference of the European Network for Business and Industrial Statistics in Prague, Czech Republic held September 6-10. Click this link for all the details.

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

P.S. 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.

6: Workshop alert: See when and where to learn about DOE

All classes listed below will be held at the Stat-Ease training center in Minneapolis unless otherwise noted. If possible, enroll at least 4 weeks prior to the date so your place can be assured. Also, take advantage of a discount when you register for multiple workshops (or when you register yourself and your colleagues).

• Experiment Design Made Easy (EDME)
  
  • September 29-30*
  • November 9-10, San Jose, California**
    
    
• Response Surface Methods for Process Optimization (RSM)
  • October 1-2*

*Take both EDME and RSM to earn $400 off the combined tuition!

• Mixture Design for Optimal Formulations (MIX)
• August 11-12**
• November 11-12, San Jose, California**
  
  
• Advanced Formulations: Combining Mixture & Process Variables (MIX2)
• August 13-14**

** Take both MIX and MIX2, or EDME and MIX, to earn $400 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, scroll down to the workshop of your choice and click on it, or call Rachel Pollack 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 [email protected].

I hope you learned something from this issue. Address your general questions and comments to me at: [email protected].

Please do not send me requests to subscribe or unsubscribe—follow the instructions at the end of this message. Sincerely,

Mark

Mark J. Anderson, PE, CQE
Principal, Stat-Ease, Inc.
2021 East Hennepin Avenue, Suite 480
Minneapolis, Minnesota 55413 USA

P.S. Quote for the month—A cutting comment on certainty:

 
"Doubt is not a pleasant condition, but certainty is absurd.”
  —Voltaire

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, Brooks Henderson and Martin Bezener
—Statistical advisor to Stat-Ease: Dr. Gary Oehlert
—Stat-Ease programmers led by Neal Vaughn
—Heidi Hansel Wolfe, Stat-Ease sales and marketing director, and all the remaining staff that provide such supreme support!

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

DOE FAQ Alert ©2015 Stat-Ease, Inc.
Circulation: Over 8900 worldwide
All rights reserved.