STA 223: Biostatistics, Generalized Linear Models

Subject: STA 223
Title: Biostatistics, Generalized Linear Models
Units: 4.0
School: College of Letters and Science LS
Department: Statistics STA
Effective Term: 2002 Fall

Learning Activities

  • Lecture - 3.0 hours
  • Discussion/Laboratory - 1.0 hours

Description

Likelihood and linear regression; generalized linear model; Binomial regression; case-control studies; dose-response and bioassay; Poisson regression; Gamma regression; quasi-likelihood models; estimating equations; multivariate GLMs.

Prerequisites

STA 131C

Cross Listed

Same course as BST 223.

Expanded Course Description

Summary of Course Content: 
Introduction (1 lecture); Likelihood and Linear Regression (3 lectures): classical linear regression model (review), likelihood principle (review), exponential family, likelihood in exponential families; Generalized Linear Model (3 lectures): components, link functions, iterated weighted least squares, asymptotics, residual analysis and goodness-of-fit; Binomial Regression (3 lectures): logistic regression, variable selection, over-dispersion, application to screening tests; Case-control Studies (4 lectures): design, relative risk, odds ratio, binomial regression approach, confounding, matching; Dose-Response Relations (4 lectures): bioassay, probit and logit, ED50, binomial regression approach, comparison of dose-response curves, additivity and combination of drugs; Poisson Regression (2 lectures): clustering, random effects models; Gamma Regression (2 lectures): variance functions, constant coefficient of variation models; Quasi-Likelihood Models (2 lectures): properties of quasi-likelihood, computational issues; Estimating Equations (2 lectures): properties, examples and applications; Multivariate GLMs (2 lectures): multinomial model; Exams and Reviews (2 lectures). 

Illustrative Reading: 
The course material is extracted from a variety of sources, including journals such as Biometrics, Biometrical Journal, Biometrika, Statistics in Medicine, JASA, Applied Statistics, and the New England Journal of Medicine. Relevant reference books are: Collett, D. Modeling Binary Data. Chapman and Hall, 1991. Dobson, A. An Introduction to Generalized Linear Models. Chapman and Hall, 1990. Fahrmeir, L and G. Tutz. Multivariate Statistical Modelling Based on Generalized Linear Models. Springer, 2001. Finney, D.T. Statistical Methods for Biological Assay. Wiley, 1987. McCullagh, P. and J. Nelder. Generalized Linear Models. Chapman and Hall, 2nd edition, 1989. Morgan, J.T. Dose-response analysis. Chapman and Hall, 1993. 

Potential Course Overlap: 
None