ABSTRACT
In Chapter 3, we have reviewed some commonly used methods for general missing data problems. Missing data methods for regression models are reviewed in Little and Rubin (2002), Ibrahim et al. (2005), and Molenberghs and Kenward (2007), among others. These reviews mostly focus on models for cross-sectional data. In practice, missing data are especially common in longitudinal studies. In this chapter, we describe some statistical methods for various missing data problems in mixed effects models, including LME, GLMM, and NLME models. We will discuss missing data in survival models and frailty models in Chapter 7. We consider the following missing data problems in mixed effects models:
• missing covariates, • missing responses, • missing both covariates and responses, • dropouts.
