This book describes a comprehensive Bayesian approach for drawing inferences from incomplete data in longitudinal studies. To illustrate these methods, the authors employ several data sets that cover a range of study designs, variable types, and missing data issues. The book first reviews modern approaches to formulate and interpret regression models for longitudinal data. Then it discusses key ideas in Bayesian inference, including specifying prior distributions, computing posterior distributions, and assessing model fit. The book carefully describes the assumptions needed to make inferences about full-data distributions from incompletely observed data. For settings with ignorable dropout, it emphasizes the importance of covariance models for inference about the mean, while for nonignorable dropout, it studies a variety of models in detail. It concludes with three case studies that highlight important features of the Bayesian approach for handling nonignorable missingness.
At the end of most chapters as well as in many applications to the health sciences, this resource offers a unified Bayesian approach to handle missing data in longitudinal studies. It contains a large number of examples and case studies, including trials and studies on schizophrenia, aging, HIV/AIDS, and smoking cessation. It makes recommendations for the use of standard regression, mixture, selection, and varying coefficient models in different settings. Also, it implements many of the analyses using WinBUGS, offering the code on a supplementary web page. The book is intended for statisticians, data analysts and other scientists involved in the collection and analysis of longitudinal data.