Activity Number:
|
347
|
Type:
|
Topic Contributed
|
Date/Time:
|
Wednesday, August 14, 2002 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Section on Statistics in Epidemiology*
|
Abstract - #301839 |
Title:
|
Longitudinal Categorical Data and Likelihood Inference
|
Author(s):
|
Patrick Heagerty*+
|
Affiliation(s):
|
University of Washington
|
Address:
|
Box 357232, Seattle, Washington, 98185, USA
|
Keywords:
|
|
Abstract:
|
One common goal of a longitudinal study is to compare two or more groups with respect to their average response at various follow-up times. This talk will discuss flexible likelihood-based regression methods for the analysis of serial categorical data that can be used to estimate response profiles. The approach is called a "marginalized transition model" (Heagerty 2002, Biometrics) since the likelihood is constructed using two assumptions: a marginal mean regression model that characterizes systematic variation in the average response as a function of covariates (including group and time); and a Markov dependence model, or transition model, that describes serial correlation within an individual. One advantage to a likelihood approach is that a profile likelihood can guide selection of appropriate covariate transformations that may facilitate parsimonious models for group differences over time.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |