|
Activity Number:
|
78
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 2, 2009 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #304165 |
|
Title:
|
Constrain Estimate of Intracluster Correlation for Clustered Binary Data
|
|
Author(s):
|
Hrishikesh Chakraborty*+ and Pranab Sen
|
|
Companies:
|
RTI International and The University of North Carolina at Chapel Hill
|
|
Address:
|
300 Parish House Road, Cary, NC, 27513,
|
|
Keywords:
|
intraclass correlation ; re-sampling ; nonparametric ; cluster randomized trials ; binary ; clustered
|
|
Abstract:
|
Several methods have been proposed to estimate intraclass correlation (ICC) for binary data and some of the methods are very specific to the type of design and underlying distributional assumptions. The ANOVA based estimation technique to estimate ICC is the most widely used method in cluster randomized trials. We proposed a new method to estimate ICC based on nonparametric methods and without replacement re-sampling principles. The main advantage of this proposed method is that it can be used for any type of categorical variable without making any additional distributional assumptions. We found that if the binary proportion is large then both methods provide similar ICC estimates for varying number of clusters and sizes. However, for small binary proportions our method provides more accurate ICC estimates than the ANOVA method does for different numbers of clusters and cluster sizes.
|
