Peter Hoff

Title: Simple Lq optimization and a Bayes-frequentist mashup

Abstract:  Many of Charlie Geyer's contributions are of use for both Bayesian and non-Bayesian statistical methods, while other contributions have helped elucidate and simplify seemingly complex statistical theory. In this talk I attempt to mimic the nature of such contributions by discussing two topics: The first is the use of "Bayes-optimal frequentist confidence intervals" for multilevel data analysis. The second is a simple method of Lq optimization that requires only knowledge of first semester calculus to understand.