|
Activity Number:
|
481
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #305563 |
|
Title:
|
Goodness-of-Fit Tests for Proportional Odds Model with GEE for Ordinal Categorical Responses
|
|
Author(s):
|
Junxiang Luo*+ and Rakesh Shukla and Qi Zhang
|
|
Companies:
|
University of Cincinnati and University of Cincinnati and University of Cincinnati
|
|
Address:
|
Department of Environmental Health, Cincinnati, OH, 45267,
|
|
Keywords:
|
goodness of fit ; cumulative logit ; GEE ; proportional odds ; ordinal
|
|
Abstract:
|
In 1994, generalized estimating equations approach was extended by Lipsitz, Kim, and Zhao to fit the proportional odds model for analyzing repeated (or clustered) ordinal categorical data. However, few methods exist to assess the goodness-of-fit (GOF) of the fitted models. A U statistic is proposed in this study to do the GOF test, and its performance is evaluated with respect to type I error rates and powers for detecting various model departures by simulation studies and an example illustration. We also compare the proposed U statistic with a kappa-like classification statistic. Results show the proposed U statistic displays much lower type I error rates and appropriate powers compared with the kappa-like classification statistic.
|
