Seminars/Colloquia/Invited Talks

Seminars

Zhiyi Chi

Constraints on FDR Control and Ways to Address Them

PLACE:	Clark 314
EVENT:	CIS Seminar Series
DATE:	October 31, 2006
TIME:	1:00 - 2:00 PM

Abstract

False discovery rate (FDR) control is a relatively new and very popular approach to multiple hypothesis testing. However, the mechanism by which it works still needs to be understood. I will show that under regular conditions, e.g., t-tests, since the strength of evidence provided by data to identify true signals from noise is limited, FDR controlling procedures, and more generally, multiple testing procedures, may have severe constraints on their power and ability to control the positive FDR (pFDR). In this case, there is no good trade-off between power and pFDR. I will discuss two methods to improve power and pFDR control:
(1) testing at multiple locations in the domain of p-values
(2) in the presence of multivariate data for each hypothesis, testing using multivariate p-values.
Part of the talk is based on joint work with Zhiqiang Tan at the Johns Hopkins University.

Brief biography

Zhiyi Chi received his PhD in Applied Mathematics at Brown University in 1998. He is currently an Associate Professor of Statistics at the University of Connecticut. His research interests include Applied Probability, Stochastic Processes and Large Deviations.