Austin Chapter of the American Statistical Association
The next meeting of the Austin chapter of the ASA will be held on Monday, October 20 beginning at 5:45 pm. It will be held in room SRH 3.109 at the LBJ school on the UT campus. Free parking is found in front of the building. Maps can be found at http://www.utexas.edu/maps/. This meeting is joint with the Austin section of ASQ.
Refreshments will be served from 5:45-6:30 pm. Announcements from both ASA-Austin and ASQ-Austin will begin at 6:30, followed by the speaker at 6:45pm.
The speaker is Dr. Joseph Voelkel. He will be speaking on "The Comparison of Two Measurement Devices."
This talk addresses the issue of comparing two measurement devices in the common case where (a) there is no standard reference device and (b) each device can make multiple measurements of the part (or unit, such as person) being measured. The most common case occurs when a new device is built and is to be compared to the current best device. Note that this work differs from a standard MSA (measurement systems analysis), in which a comparison to another device is typically not done.
We first present the general problem, and then examine some common, but incorrect, ways of comparing the devices. We show mathematical models that can be used to compare the devices, explain the basic ideas behind them, and then show how a series of hypothesis tests can be used to see how well the devices match up. For example, two devices may be measuring the same general feature but one may be linearly biased with respect to the other. It turns out that statistical packages typically do not possess the ability to perform these analyses correctly, but the general statistical method of Maximum Likelihood does solve the problem well. An example will be presented to illustrate the method.
Joseph G. Voelkel, Ph.D., is an Associate Professor and Graduate Program Chair in the Graduate Statistics Department at The John D. Hromi Center for Quality and Applied Statistics at Rochester Institute of Technology. Prior to this appointment, he was a statistical consultant for Allied-Signal Corporation (now Honeywell), where his clients' concerns included chemicals, plastics, fibers, water-treatment polymers, and agricultural products. He has also served on the faculty at the University of Wisconsin, where he worked in cancer research, and as an adjunct at Canisius College.
Since joining the faculty at RIT, his consulting and training has included clients from the films, optics, electronics, resin, plastics, automotive, laser, and bearing industries. His primary areas of expertise include experimental design, quality control and improvement, mixture experiments, reliability, and analysis of messy data. Dr. Voelkel received his BS in Mathematics from Rensselaer Polytechnic Institute, his MS in Industrial Engineering/Management Sciences from Northwestern University, and his Ph.D. in Statistics from University of Wisconsin, Madison. He is a fellow of the American Society for Quality, and a member of the American Statistical Association and the Institute of Mathematical Statistics. He also serves on the editorial board of Quality Engineering. He was a finalist for RIT's Eisenhart Outstanding Teacher of the Year and is currently writing a book on modern uses of experimental design.
Short course opportunites
ISMI is offering two statistical short courses as part of ISMI Manufacturing Week, October 20-23. Registration information can be found at http://ismi.sematech.org/ismisymposium/registration.htm.
Short Course: "Evaluating The Measurement Process III"
Monday, October 20, 8AM—5PM
Everyone is in favor of good measurements, but what is the relative utility of your measurements and what does it mean in a practical sense? How can you actually use the measurements you have to work with? How can you know exactly how many digits to record for a measurement? How can you do an Honest Gauge R&R Study? What is the effective resolution of your measurements? How can we identify and remove operator effects and instrument effects which degrade our measurements? What is the difference between measurement precision and relative utility and how can we use both?
All who collect or use physical measurements of any kind will find this workshop helpful. It sweeps away the nonsense that so often surrounds measurement error and gives you practical answers to real questions. While the techniques taught are built on a sound theoretical foundation, the focus of this workshop will not be on the theory, but rather on how to use the techniques in business and industry.
The accompanying 300 page text contains over 60 worked-out examples and case histories.
Dr. Donald J. Wheeler is an internationally recognized expert on SPC and data analysis. His knowledge, experience, and ability to communicate have given him a world-wide following. He has conducted over 1,000 seminars in 17 countries on five continents, and he has had students come from 30 countries to attend his seminars in the U.S.
He is the author or co-author of 22 books and hundreds of articles. He has been a monthly columnist for both Quality Digest and Quality magazine. His books have been translated into five languages and are in use in over 40 countries around the world.
He holds M.S. and Ph.D. degrees in Statistics from Southern Methodist University, and for 12 years he taught at the University of Tennessee where he was an Associate Professor of Statistics. He was a student and associate of Dr. W. Edwards Deming for 21 years and is one of a handful of individuals to have been elected as Fellow of both the American Statistical Association and the American Society for Quality. How to Register
You must register to attend this short course. There is a registration fee of $200 for ISMI Members; $400 for all others. The registration fee includes a copy of Dr. Wheeler's textbook, Evaluating the Measurement Process III.
Short Course: "Extended Structure Diagrams: a Powerful Aid in Design & Analysis of Complex Experiments"
Tuesday, October 21, 8AM-5PM
Many experimenters believe that an experimental design can be selected from a book on experimental design or even a computer menu. However, most experimental-design texts, by necessity, show only a narrow slice of designs. Too often only a small fraction of experiments that experimenters would like to run, or statisticians would like to design, are in this slice. As a result, too many experimental designs are simply pigeonholed to fit into a pre-existing list. Or sometimes, experimenters manage to create a good design but then find it difficult to analyze.
But there is a better way. A large class of experimental designs are balanced and complete—this includes most:
This course offers a powerful method—Extended Structure Diagrams, or ESDs—that allows experimenters to first visualize any balanced and complete design, and then find the natural corresponding ANOVA model. With this model, insightful and in-depth analysis may be done.
Thinking correctly about how to construct the appropriate ESD for complex designs is not as simple as one might believe, but this course is designed to help.
Course topics include:
Participants should have experience designing and analyzing experiments and be familiar with ANOVA.
Bring a laptop with JMP® or Minitab® to do some examples yourself—feel free to bring your data as well (complete and balanced!). Send to me ahead of time (firstname.lastname@example.org) if you want share it with the class.
A simpler version of this presentation was the highest-rated talk at a Six-Sigma conference several years ago.
Joseph G. Voelkel is Professor and Graduate Program Chair at the John D. Hromi Center for Quality and Applied Statistics of the Rochester Institute of Technology. His research interests include experimental design, mixture experiments, quality control and improvement, statistical methods in reliability, and non-standard statistical problems.
You must register to attend this short course. There is a registration fee of $150 for ISMI Members; $350 for all others.