Introduction to Multiple Comparison Procedures
Sunday morning, July, 8Jason C. Hsu, The Ohio State University
This course will use real applications to illustrate concepts and techniques in multiple comparisons. Error rate concepts covered include Familywise Error Rate, False Discovery Rate, and generalized Familywise Error Rate. Multiple test construction techniques covered include union intersection testing, intersection union testing, closed testing, and partition testing. Conditions for shortcutting closed/partition tests to step-up and step-down tests, and validity of re-sampling methods, will be given. Applications covered include equivalence testing, dose-response studies, and analysis of gene expressions from microarray experiments.
Jason Hsu teaches at the Ohio State University. He and his students have been involved in developing a principle now called the Partitioning Principle which has given useful insights into stepwise testing and confidence set construction. He has worked on implementing multiple comparison methods in statistical packages, and pharmaceutical statistics problems such as bioequivalence and dose-response. His current research includes statistical design and analysis of microarray experiments, control of the generalized Familywise error rate, and model assumption required for validity of re-sampling methods such as permutation tests.