Online Program Home
My Program

Abstract Details

Activity Number: 311
Type: Contributed
Date/Time: Tuesday, August 2, 2016 : 8:30 AM to 10:20 AM
Sponsor: Biometrics Section
Abstract #320194
Title: Multiple Intraclass Correlations for Higher-Level Nested Logistic Regression
Author(s): Kyle Irimata* and Jeffrey Wilson
Companies: Arizona State University and Arizona State University
Keywords: Intraclass Correlation ; Generalized Linear Mixed Models ; Hierarchical Data ; Overdispersion ; Binary Outcomes

In the analysis of hierarchical data, correlation often occurs at each level of the hierarchy. The resulting correlation amongst the responses is commonly accounted for through the use of more complex and parameterized models such as multilevel logistic regression models; however it is not uncommon to find that these more complicated models present the same conclusions as less parameterized models. While there are a variety of methods available for measuring this intraclass correlation, these methods are usually used to address the correlation at the lowest level of the structure and are often limited to the analysis of continuous outcomes. We provide a measure of association at each stage of a three-level nested structure with binary responses, as well as appropriate tests to determine when the dependency between observations needs to be taken into account and when it can be ignored. A threshold at which the intraclass correlation will begin to affect the significance of the covariates is established using simulated hierarchical data. The use of these methods are illustrated through the analysis of patient visit data in Bangladesh.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2016 program

Copyright © American Statistical Association