Abstract

The analysis of residuals can capture departures from a parametrized model. In this thesis we look at how the generalized linear model has become one of the most important developments in statistics in the last thirty years, and on the adequacy of regression model diagnostics that are meaningful and significant in a generalized linear model context. Some asymptotic properties are discussed and numerical examples are provided to illustrate the techniques for binomial, Poisson, and gamma distributed random variables.