FIFTEEN. Modeling Correlated Outcomes with Generalized Estimating Equations

  1. Mari Palta

Published Online: 11 AUG 2003

DOI: 10.1002/0471467979.ch15

Quantitative Methods in Population Health: Extensions of Ordinary Regression

Quantitative Methods in Population Health: Extensions of Ordinary Regression

How to Cite

Palta, M. (2003) Modeling Correlated Outcomes with Generalized Estimating Equations, in Quantitative Methods in Population Health: Extensions of Ordinary Regression, John Wiley & Sons, Inc., Hoboken, NJ, USA. doi: 10.1002/0471467979.ch15

Author Information

  1. Madison, Wisconsin, USA

Publication History

  1. Published Online: 11 AUG 2003
  2. Published Print: 15 AUG 2003

Book Series:

  1. Wiley Series in Probability and Statistics

Book Series Editors:

  1. Walter A. Shewhart and
  2. Samuel S. Wilks

ISBN Information

Print ISBN: 9780471455059

Online ISBN: 9780471467977

SEARCH

Keywords:

  • Poisson;
  • binary;
  • GEE;
  • empirical variance;
  • conditional logistic regression;
  • correlation;
  • longitudinal;
  • drop-out;
  • pattern mixture;
  • PROC GENMOD

Summary

We consider correlated or longitudinal binomial and Poisson outcomes. Motivation for generalized estimating equations (GEE) based on score equations in previous chapters. Empirical variance formula with GEE. Comparison of PROC MIXED and PROC GENMOD for normally distributed data. Application of PROC GENMOD with different correlation structures. Analysis of longitudinal hypertension data as example of correlated binary outcome. Analysis of longitudinal hospitalization data as example of Poisson outcome. Hypertension data analyses include example of conditional logistic regression as alternative to GEE. Consideration of how to adjust for drop-out via pattern mixture model.