J.A. Ferreira,
Department of Clinical Epidemiology and Biostatistics
Academic Medical Centre, University of Amsterdam

On the Benjamini-Hochberg method of multiple testing
 (joint work with A.H. Zwinderman)


The Benjamini-Hochberg procedure has become a popular method for
testing several hypotheses simultaneously, especially in contexts
where the number of hypotheses is very large such as those of
"microarray analysis" and "gene discovery". Under certain conditions,
the method is known to control the false discovery rate - the expected
proportion of incorrect rejections among rejected hypotheses - and to
be less conservative than Bonferroni-type procedures. Several variants
of the method have recently been proposed in the literature with the
aim of increasing its power (of rejecting false hypotheses) while
maintaining the control of the false discovery rate. In this talk we
shall look at some of the properties of the Benjamini-Hochberg method
and of another variant of it, namely those related to the convergence
of the false discovery rate and of the proportion of incorrect
rejections (as the sample size goes to infinity).