Bayesian Inference for Categorical Data Analysis: A Survey
by Professor Alan Agresti
Abstract: his seminar surveys Bayesian methods for categorical data analysis, with primary emphasis on contingency table analysis. Early innovations were proposed by I.J. Good for smoothing proportions in contingency tables and by Dennis Lindley for inference about odds ratios. These approaches primarily used conjugate beta and Dirichlet priors. Pat Altham presented Bayesian analogs of small-sample frequentist tests for 2x2 tables using such priors. An alternative approach using normal priors for logits received considerable attention in the 1970s by Tom Leonard and others. Adopted usually in a hierarchical form, the logit-normal approach allows greater flexibility and scope for generalization. The 1970s also saw considerable interest in Stein-influenced Bayesian shrinkage methods and in log-linear modelling. The advent of modern computational methods since the mid-1980s has led to a growing literature on fully Bayesian analyses with models for categorical data, with main emphasis on generalized linear models such as logistic regression for binary and multi-category response variables and hierarchical generalizations.
For More Information: Professor Alan Agresti from University of Florida will be visiting Melbourne University from Monday 27 until Wednesday 29th June. His visit is co-sponsored by the Melbourne Business School and the Department of Mathematics and Statistics. He will also be in Sydney from 30th-31st and presenting in a special session on Exact Methods Of Statistical Inference at the SSAI/NZSA conference in Auckland.