Activity Number:
|
114
|
Type:
|
Topic Contributed
|
Date/Time:
|
Monday, August 1, 2016 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Committee on Career Development
|
Abstract #320696
|
View Presentation
|
Title:
|
Conditional Logistic Regression in a Cluster-Specific M-to-M Treatment-Control Study with Binary Outcomes
|
Author(s):
|
Zhulin He* and Gabriel Demuth and Zhengyuan Zhu
|
Companies:
|
Iowa State University and Iowa State University and Iowa State University
|
Keywords:
|
Conditional logistic regression ;
Treatment-control design ;
Binary outcomes ;
Ordinary logistic regression ;
Cluster-specific
|
Abstract:
|
Conditional logistic regression (CLR) method is derived by maximizing the conditional likelihood based on the sum of outcomes in each cluster. Using CLR, we can estimate the parameters of interest without estimating unwanted cluster-level nuisance parameters. Our main interest is to study the performance of CLR comparing with ordinary logistic regression (OLR) in a m:m treatment-control design with binary outcomes, where m is a study design factor. We compare CLR and OLR estimators under different designs indexed by the design factor m. We also illustrate the results with simulation studies and an application of National Resources Inventory image study.
|
Authors who are presenting talks have a * after their name.