ABSTRACT
The objective of the present discussion is to provide a framework for the relatively systematic analysis of a very wide range of possible F. This chapter focuses on the likelihood function and quantities derived from it. It presents the primary mode for summarizing evidence about a one-dimensional parameter of interest to be a nested series of upper and lower confidence limits based as directly as possible on the likelihood. The chapter deals with general F, it is important to have some special families in mind, both to illustrate general concepts and because of their intrinsic importance. It considers three such families, returning later to more thorough discussion. These are the normal theory linear models, the exponential models and the transformation models.
