Regression for Dependent Ordinal Data
The lion's share of the ORD_EE module was written by
when he was a graduate
student at Hopkins. Then Paul Rathouz wrote a few S functions
and alot of very helpful documentation about Patrick's module.
I wrote a somewhat verbose help document in LaTeX (which can be
downloaded below) and made the module available on the web.
Students in the Department of Bisotatistics at Hopkins should refrain
from downloading this module because it already exists on the
network and is easy to access.
Abstract from User's Guide
This document will walk the user through an elementary example which
uses estimating equations for correlated ordinal data. This document
is written at a Masters in Biostatistics level and
will outline details of Heagerty and Zeger's
technology and the software Heagerty
developed to process this technology.
The statistical methods extend the
proportional odds model for regression analyses with ordinal response
data to clustered (ordinal) reponses.
With this technology, we
can easily model nested data, such as in hierarchical models, or
longitudinal data. Logistic regression is a special case of the proportional
odds model and was extended to clustered responses by Zeger and Liang
Download the rest of the User's Guide in LaTeX.
Instructions on how to download Patrick's stuff.