Complements and Strategies for Application

Abstract

A broad array of families of estimating functions has been presented in this monograph but many others are available and may be of particular use in special circumstances. The focus of the theory has usually been on polynomial functions of the data and these will not always provide ideal families of estimating functions. It must be remembered that the general theory requires estimating functions which are at least square integrable and if the underlying variables of interest do not have finite second moments then transformations are required. A number of useful approaches are indicated below.