SCRC 2005 / FIM XII
   Hosted by Auburn University

Organizing Committees
Announcement
Important Dates
Featured Speakers
New
Workshops/Special Events
New
R. C. Bose Memorial Keynote
Symposia Topics
Tentative Schedule
Abstract Submission
Student Competition
Registration (on-line)
Sponsors (updated list)
Past Conferences
Hotel (GOING FAST)
Contact Us
Main Page

 
Bachhofer-Gupta-Sobel Memorial Symposium on
Ranking and Selection and Multiple Comparison Procedures Methodologies

 
Peter Westfall is Paul Whitfield Horn Professor of Statistics, James Niver Professor of Information Systems and Quantitative Sciences, and Director of the Center for Advanced Analytics and Business Intelligence at Texas Tech University.  He has published 70+ articles on statistical theory and practice in journals including Annals of Statistics, Journal of the American Statistical Association, The American Statistician, Biometrika, Biometrics, Statistics in Medicine, Genetics, Genetic Epidemiology, Human Heredity, Pharmaceutical Statistics and Journal of Biopharmaceutical Statistics; he is also published in several business journals.  He is lead author of the books Resampling‑Based Multiple Testing: Examples and Methods for P-value adjustment (Wiley, 1993), and Multiple Comparisons and Multiple Tests Using the SAS® System (SAS Book's by Users, 1999). Joint with S. Stanley Young of Glaxo Inc., he received the 1991 "Most Outstanding Applications Paper Award" from the American Statistical Association for developing new multiple comparisons methodology with special application to Pharmaceutical statistics. Funded by a consortium of Pharmaceutical Companies, he developed the SAS/STAT® procedure PROC MULTTEST, which is resampling-based software used for multiple outcomes and multiple comparisons with applications to clinical trials, preclinical animal carcinogenicity studies, gene expression studies, and genetic association studies. He has given numerous invited papers and keynote addresses at major national and international conferences, has ongoing training and consulting arrangements with various pharmaceutical companies, periodically consults with the United States Food and Drug Administration, serves on data safety monitoring boards, and develops protocols for Phase III clinical trials conducted by researchers at Texas Tech University.  His Center for Advanced Analytics and Business Intelligence has been cited by Next-gen data center forum, Database Trends and Applications, and Grid Today for its contributions to grid computing technology. He served as co-chair (with Ajit Tamhane) for the Third International Conference on Multiple Comparisons Procedures, held in Bethesda, MD, August 2002, and co-edited volumes appearing in Journal of Biopharmaceutical Statistics and Journal of Statistical Planning and Inference resulting from this conference. He is editor-elect for The American Statistician (2006-2008), has served as associate editor for The American Statistician and Journal of the American Statistical Association; and he is a Fellow of the American Statistical Association.            

Optimal data snooping: Is Bonferroni admissible for large k?

Modern methods of multiple comparisons, including FDR-controlling, Bayesian, and decision theoretic, are lax relative to the Bonferroni method in their assignment of significances; they are relatively more lax as k, the number of tests, increases. I point out that this laxness is based on an assumption concerning the size of the loss due to Type I errors relative to loss due to Type II errors. I challenge the generality of this assumption, and present an alternative loss function for which the Bonferroni method is appropriate. Using examples, I demonstrate that the proposed loss function is reasonable and applicable to data snooping and data mining, and that it can provide better decisions.

 

 


 

12th Annual Conference of the Forum for Interdisciplinary Mathematics (FIM XII)