Activity Number:
|
287
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 14, 2002 : 8:30 AM to 10:20 AM
|
Sponsor:
|
General Methodology
|
Abstract - #301193 |
Title:
|
Two Stage False Discovery Controlling Procedure for Multiple Testing
|
Author(s):
|
Yoav Benjamini*+ and Abba Krieger and Daniel Yekutieli
|
Affiliation(s):
|
Tel-Aviv University and University of Pennsylvania and Tel-Aviv University
|
Address:
|
Ramat Aviv, Tel Aviv, International, 69978, Israel
|
Keywords:
|
FDR ; Multiple comparisons
|
Abstract:
|
The linear step-up procedure, which controls the false discovery rate (FDR) in multiple testing, is inherently conservative. It controls the FDR at a level too low by a factor of the number of true hypotheses to the number tested. Various efforts have been made to increase the power of the procedure by estimating the factor from the marginal p-values. We present a very simple and intuitive two-stage procedure of this nature. We prove that this procedure controls the FDR when the test statistics are independent. We show by simulations that it does so when the test statistics are positively correlated. We discuss the asymptotic optimality of this procedure as a multiple testing procedure.
|