Generalized Linear Models 40 years on and AGM 2012

A chance to meet up and have lunch before the AGM and meeting. This will be a pay as you go lunch but it would be useful for the restaurant to have an idea of numbers and so if you wish to take advantage of this please send an email to Suzanne Clark

The ideas that went together to create generalized linear models had been around for some time, but the power of the 1972 paper by Nelder and Wedderburn was to pull all of these together into a coherent modeling framework. The paper itself is well worth a revisit, with a strong emphasis on applications and teaching. The basic model has now been extended in many different directions and I will consider some topics that have been of particular interest to me, such as overdispersion, random effects, and mixtures. These extensions will be illustrated with a range of real applications.

Objective Bayesian model selection in GLMs

In medical research and elsewhere, variable and model selection in regression is still a common problem.  Bayesian model selection based on test statistics (Johnson, 2005, 2008, Hu and Johnson, 2009) is an approach,  which eliminates the need to specify proper prior distributions on regression parameters. The method is applicable to generalized linear models and the Cox model using the deviance statistic.  In this talk I will review and  extend Johnson's methodology and point out connections to shrinkage estimates for prediction. The methodology is then applied in selected biomedical applications.

This is joint work with Daniel Sabanés Bové