SPES is proud to announce its schedule of invited sessions at this year's Joint Statistical Meetings, in Dallas, and is honored to have so many distinguished speakers grace its program. We are sure that you will find many sessions to be of interest and highly worthwhile. The five invited sessions are spread throughout the week at JSM. On Sunday, Vida Tigrani and Bob Grossman will be sharing their nuggets of wisdom on Data Mining. Monday's panel discussion on Key Challenges to Today's Industrial Statisticians showcases three fine hurdlers: Gerry Hahn, Doug Montgomery, and Jim Landwehr. On Tuesday, Ker-Chau Li, James Filliben, Ming-Jame Horng, and Chun-Houh Chen will make the problem of Dimension Reduction more manageable for us. Wednesday's Chemometrics session will be brought to you by Cliff Speigelman and Barry Wise, who will both speak on variable selection problems. Finally on Thursday, Benjamin Reiser, Necip Doganaksoy, and Louise Ryan will share their secrets to success with Reliability Models with Incompletely Identified Causes of Failure. Be sure to make it to this year's JSM! Coming in future episodes: More detailed description of the invited program and a sneak-preview of the contributed sessions.
Scott Beattie and Dennis Lin, 1998 JSM Program Chair
The SPES Roundtable Discussion topics for 1998 cover a broad spectrum of interest. John Peterson will be leading "Sequential optimization...or...I can't do 16 experiments!" . This roundtable deals with sequential ways to ease clients into employing statistically designed experiments for process optimization when they balk at batch factorial approaches. Next is Ashwini Mathur, whose roundtable is titled; "Use of statistics in safety assessment of new drugs". His table will discuss use of statistics in safety assessment of new drugs including design and analyses issues for safety assessment studies (for example, reproductive toxicology studies, and carcinogenicity studies) and issues related to multiple end-points and novel endpoints. Darryl Downing will lead the discussion on "High Impact Consulting". The discussion will be about the five points of high impact consultiing (which Darryl would rather call collaboration) including project definition, scope, design, partnering, and deployment. Finally, but not least is the dynamic duo of Lynne Hare and Roger Hoerl doing what they do best - putting on the thinking hat and leading the discussion on "Statistical Thinking for Statisticians". They will each lead a separate table (is this parallel processing?) talking about what Statistical Thinking is and why it is key to greater understanding and use of statistical methods by statisticians and non-statisticians alike.
Darryl Downing, 1998 JSM Program Chair Elect
It is a pleasure to announce that SPES is involved in the sponsorship of two world-class continuing education offerings next year at the JSM's in Dallas. Along with the Quality and Productivity Section, SPES is co-sponsoring a course titled, "Statistical Methods for Failure-Time and Accelerated Test Data" This one-day course will be presented by Bill Meeker and Luis Escobar. As Bill and Luis describe the course, it will illustrate the use of a variety of proven techniques brought up to date with modern computer-based methodology. The topics to be covered include handling censored data, nonparametric and maximum likelihood estimation, probability plotting, likelihood-based confidence intervals, and models and methods related to accelerated degradation testing. Most of the examples will come from the area of testing product reliability, but some biological examples will also be included.
Another tool that has been used extensively in the biological sciences that is now receiving more interest in industrial applications is the generalized linear model. To give our members an opportunity to learn more about this exciting area, we are pleased announce that Doug Montgomery and Geoff Vining have agreed to present a course on this topic entitled, "Methods and Applications of Generalized Linear Models." As Doug and Geoff state in the course abstract, generalized linear models are employed extensively for modeling and analyzing data where the response variable is nonnormal. This one-day short course will describe and illustrate the use of generalized linear models, and show how this approach is an attractive alternative to more traditional analysis. Topics include the framework of generalized linear models, model fitting methods, choice of link function and error distribution, tests on model coefficients, and methods for model adequacy checking. Emphasis will be placed on logistic and Poisson regression and models with exponential and gamma errors. Computer implementation will be discussed extensively. Examples from manufacturing, engineering development, quality improvement, and other fields will be presented.
So make plans now to participate in one or both of these marvelous one day offerings in Dallas!
Dave Sartori, SPES CE Chair