Exploring the Information in p-Values for the Analysis and Planning of Multiple-Test Experiments

Dan Nettleton
Department of Statistics
Iowa State University

Wednesday, October 3, 2007
3:30pm
MoosT 1-450G
Minneapolis Campus

Abstract:

A new methodology is proposed for estimating the proportion of true null hypotheses in a large collection of tests. Each test concerns a single parameter delta whose value is specified by the null hypothesis. We combine a parametric model for the conditional cumulative distribution function of the p-value given delta with a nonparametric spline model for the density of delta under the alternative hypothesis. The proportion of true null hypotheses and the coefficients in the spline model are estimated by penalized least squares subject to constraints that guarantee that the spline is a density. The estimator is computed efficiently using quadratic programming. In addition to providing an estimate of the proportion of true null hypotheses, our methodology produces an estimate of the density of delta when the null is false. We discuss the use of our estimate of the density of delta in sample size calculations for future multiple-test experiments. We compare our estimator to leading competitors through simulation and illustrate our method using microarray data sets generated at Iowa State University.

This talk covers joint work with David Ruppert and Gene Hwang at Cornell University.

A social tea will be held at 3:00 P.M. in A434 Mayo. All are Welcome.
For more details contact 612-624-4655 or see http://www.biostat.umn.edu/seminar_academic.html