Chapter 16: Other Approaches to
Analysis of Correlated Data
These examples illustrate the GEE approach
for modeling data containing correlated
dichotomous outcomes. However, use of the
GEE approach is not restricted to dichotomous
outcomes. As an extension of GLM, the GEE
approach can be used to model other types of
outcomes, such as count or continuous out-
comes.We suggest that you review the material cov-
ered here by reading the detailed outline that
follows. Then, do the practice exercises and
test.The GEE approach to correlated data has been
used extensively. Other approaches to the anal-
ysis of correlated data are available. A brief
overview of several of these approaches is pre-
sented in the next chapter.V. SUMMARY
üChapter 15: GEE Examples
This presentation is now complete. The focus
of the presentation was on several examples
used to illustrate the application and interpre-
tation of the GEE modeling approach. The
examples show that the selection of different
correlation structures for a GEE model
applied to the same data can produce differ-
ent estimates for regression parameters and
their standard errors. In addition, we show
that the application of a standard logistic
regression model to data with correlated
responses may lead to incorrect inferences.Presentation: V. Summary 557