Buehler-Martin Lecturer Series - March 30, 2005
University of Minnesota
School of Statistics
College of Liberal Arts

Established by Mr. and Mrs. Thomas Martin
in Memory of
Robert J. Buehler, Professor of Statistics (1963-1988)

Longitudinal and Clustered Data and Non/Semiparametric Regression

Raymond J. Carroll
Department of Statistics
Texas A & M University

Wednesday, March 30, 2005
3:30 PM, 210 Tate Lab of Physics
Minneapolis, East Bank Campus
Social at 3:00 PM, 300 Ford Hall

Abstract

I will review the problem of nonparametric and semiparametric regression for longitudinal and clustered data. More generally, I will consider problems in which a nonparametric function is evaluated repeatedly for each subject. The semiparametric efficient score is easily calculated, although implementation is not. I will describe a constructive solution to this problem, contrast backfitting and profiling, and discuss several examples. In particular, I will consider the Gaussian case, and compare spline and kernel-based approaches to the problem.