Activity Number:
|
87
|
Type:
|
Contributed
|
Date/Time:
|
Sunday, July 31, 2016 : 4:00 PM to 5:50 PM
|
Sponsor:
|
Quality and Productivity Section
|
Abstract #319194
|
|
Title:
|
Optimal Designs for Logistic Mixed Models Using Penalized Quasi-Likelihood Method
|
Author(s):
|
Wanchunzi Yu* and John Stufken and Zhongshen Wang
|
Companies:
|
Arizona State University and Arizona State University and Arizona State University
|
Keywords:
|
Locally optimal design ;
Complete class approach ;
Binary longitudinal Studies ;
General equivalence Theorem
|
Abstract:
|
Optimal design questions for generalized linear mixed models (GLMM) are common but challenging due to the complexity of information matrices. For binary longitudinal responses, we establish a relationship between approximations of information matrices for logistic mixed- effect models using penalized quasi-likelihood and information matrices for corresponding fixed-effect logistic regression models. Using this relationship, we identify locally optimal designs based on widely used optimality criteria. The proposed method for identifying optimal designs is demonstrated through the example of self-reported disability in older women. The robustness of design efficiency to mis-specification of the covariance of random effects is also studied.
|
Authors who are presenting talks have a * after their name.