There is a growing literature on methods for statistical analysis of datasets with missing values, with much of this research focused on the development of several modeling frameworks to handle missing data. These include the classes of selection, pattern-mixture and sharedparameter models under a likelihood perspective, and the inverse probability weighted and doubly robust estimators under the generalized estimating equations framework. The interested reader is referred to Chapters 4 and 8 of this volume. The most important lesson that has been learned from all this work is that in missing datasettings, inferences can be strongly influenced by modeling assumptions. Therefore, there has been much interest in sensitivity analysis and the investigation of how results change when key components of the model are altered.