Adaptive Multiple Testing Procedures Controlling the Familywise Error Rate

• Activity Number: 649
• Type: Invited
• Date/Time: Thursday, August 5, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Sanat K. Sarkar*+
• Companies: Temple University
• Keywords: Multiple Testing ; Stepwise Multiple Testing Methods ; FWER Controlling Methods ; Adaptive Multiple Testing Methods

Abstract

The familywise error rate (FWER) is frequently chosen as an overall measure of Type I error to control in multiple testing. Several FWER controlling procedures are available in the literature, such as the Bonferroni, Holm, Hochberg, and Hommel procedures. These procedures can be made less conservative, and hence more powerful, by extracting information about the number of true null hypotheses from the available data and incorporating that into the procedures. Data-adaptive versions of these conventional FWER controlling procedures will be presented in this talk, along with proofs that these adaptive procedures provide ultimate control of the FWER under certain type of dependence structure among the underlying p-values. Results of a numerical study comparing the performances of these different adaptive FWER controlling procedures will be presented.