A Unifying Perspective on Statistical Modeling

Abstract

This paper presents a streamlined perspective on the various approaches to statistical modeling. This perspective can bethought of as a way to unify the Bayes, empirical-Bayes, Bayes–empirical-Bayes, parametric, nonparametric, and semiparametric models used to describe uncertainty. It is emphasized here that this paper concentrates on modeling, not on the approaches to inference that can also be categorized under the above labels, but that are quite different in their philosophical orientation.