A0869
Title: Constrained inference for longitudinal data with nonignorable missing responses
Authors: Sanjoy Sinha - Carleton University (Canada) [presenting]
Abstract: Missing data are common in longitudinal studies. When data are nonignorably missing, it is necessary to incorporate the missing data mechanism into the likelihood function to ensure valid inference. The unrestricted maximum likelihood method for incomplete longitudinal data has been extensively studied in the literature. However, parameter orderings or constraints may naturally arise in real-life scenarios, where the efficiency of an estimator can be improved by incorporating these parameter constraints into estimation and hypothesis testing. The purpose is to discuss some novel methods for analyzing longitudinal data with nonignorable missing responses under linear inequality constraints. The proposed method is developed within the framework of the generalized linear mixed model. The empirical properties of the estimators are investigated through Monte Carlo simulations. An application is presented using real data from a health survey.
