BioStat 571 Syllabus - Advanced Generalized Linear Models II: Correlated Data

Lectures: M W F, 1:30 - 2:20 in HSB T-498

Class Web Page: http://faculty.washington.edu/heagerty/Courses/b571/
This will be used for posting homework assignments, discussion of homework, data sets and announcements. Please visit the web site regularly, especially on MWF.

Useful References:
. title author
*+DHLZ: Analysis of Longitudinal Data Diggle, Heagerty, Liang, and Zeger
*+VM : Linear Mixed Models for Longitudinal Data Verbeke and Molenberghs
+MN: Generalized Linear Models McCullagh and Nelder
+PB: Mixed-Effects Models in S and S-PLUS Pinheiro and Bates
FT: Multivariate Statistical Modelling Based on Generalized Linear Models Fahrmeir and Tutz
SCM: Variance Components Searle, Casella and McCulloch
+CH: Analysis of Repeated Measures Crowder and Hand
Li: Models for Repeated Measurements Lindsey
He: Quasilikelihood and its Application Heyde
Lo: Random Coefficient Models Longford
Go: Multilevel Statistical Models Goldstein
MS: Generalized, Linear, and Mixed Models McCulloch and Searle

* = (required text; copies at SCC)
+ = (to be put on 2 hr reserve at health sciences library)

Computing:
We will emphasize implementation in R/Splus, but SAS or STATA will be used for several specialized analyses.

Grading:

Course Objectives:
The prerequisites for the course are a good knowledge of linear model theory for ANOVA and regression models for independent observations such as obtained in BIOST/STAT 570 and elsewhere. This includes knowledge of least squares, model selection, regression diagnostics, maximum likelihood theory, and an introduction to generalized linear models.

This course will extend linear model methods to methods for analysis of data with non-iid errors, including generalized linear models for data with nonconstant variance, quasilikelihood for overdispersed data, and several methods for correlated data such as random effects and mixed models, analysis of repeated measures, longitudinal data analysis, and generalized estimating equations. The primary objective of the course is to provide some tools for analysis of data with non-iid errors, and highlight the common features of these tools to give a general approach to the analysis of such data.