A General Procedure for Obtaining Maximum Likelihood Estimates in Generalized Regression Models

Abstract

This paper describes an iterative procedure for obtaining mnaximum likelihood estimates of the parameters of a generalized regression model when direct maximization with respect to all parameters is difficult.A proof of convergence and some interesting applications are provided.' The first draft of this paper was written while both authors were at the University of Bonn.Helpful comments were received from Werner Oettli and from an anonymous referee.2 An investigation of a similar problem in the context of general optimization procedures has been carried out by Warga [4,5].3 This lemma is also given in Sargan [3].579