SEMINARS
|
|
Spring 2006
STATISTICS
COLLOQUIUM
Wednesday, March 1, 2006
3:30-4:00—Refreshments
4:00-5:00—Talk
Yost Hall, Room 101
Robert Keener
Professor, Statistics Department, University of Michigan
Local Information and the Design of Sequential Hypothesis Tests
In sequential hypothesis testing, information from data is studies as it accumulates and used to decide when to terminate an experiment. For instance, in a clinical trial a study should be stopped as soon as there is enough information to infer which treatment is superior, so that as few subjects as possible receive the inferior treatment. Proper design of a sequential experiment involves balancing inferential risks against costs for a data collection. Various results detailing how this can be done optimally or near optimally for sequential testing will be reviewed. Optimal stopping rules are related to a local measure of statistical information. In some cases, local information can be approximated by L-numbers discovered by Lorden, and simple rules based on these approximations are asymptotically optimal to better order than the cost for a single observation.
|
