Missing Data

Paul D. Allison, Ph.D., is Professor of Sociology at the University of Pennsylvania where he teaches graduate courses in methods and statistics. He is also the founder and president of Statistical Horizons LLC which offers short courses on a wide variety of statistical topics.After completing his doctorate in sociology at the University of Wisconsin, he did postdoctoral study in statistics at the University of Chicago and the University of Pennsylvania. He has published eight books and more than 60 articles on topics that include linear regression, log-linear analysis, logistic regression, structural equation models, inequality measures, missing data, and survival analysis.Much of his early research focused on career patterns of academic scientists. At present, his principal research is on methods for analyzing longitudinal data, especially those for determining the causes and consequences of events, and on methods for handling missing data.A former Guggenheim Fellow, Allison received the 2001 Lazarsfeld Award for distinguished contributions to sociological methodology. In 2010 he was named a Fellow of the American Statistical Association. He is also a two-time winner of the American Statistical Association’s award for “Excellence in Continuing Education.”

"…an excellent resource for researchers who are conducting multivariate statistical studies."

• Series Editor′s Introduction1. Introduction2. AssumptionsMissing Completely at RandomMissing at RandomIgnorableNonignorable3. Conventional MethodsListwise DeletionPairwise DeletionDummy Variable AdjustmentImputationSummary4. Maximum LikelihoodReview of Maximum LikelihoodML With Missing DataContingency Table DataLinear Models With Normally Distributed DataThe EM AlgorithmEM ExampleDirect MLDirect ML ExampleConclusion5. Multiple Imputation: BascisSingle Random ImputationMultiple Random ImputationAllowing for Random Variation in the Parameter EstimatesMultiple Imputation Under the Multivariate Normal ModelData Augmentation for the Multivariate Normal ModelConvergence in Data AugmentationSequential Verses Parallel Chains of Data AugmentationUsing the Normal Model for Nonnormal or Categorical DataExploratory AnalysisMI Example 16. Multiple Imputation: ComplicationsInteractions and Nonlinearities in MICompatibility of the Imputation Model and the Analysis ModelRole of the Dependent Variable in ImputationUsing Additional Variables in the Imputation ProcessOther Parametric Approaches to Multiple ImputationNonparametric and Partially Parametric MethodsSequential Generalized Regression ModelsLinear Hypothesis Tests and Likelihood Ratio TestsMI Example 2MI for Longitudinal and Other Clustered DataMI Example 37. Nonignorable Missing DataTwo Classes of ModelsHeckman′s Model for Sample Selection BiasML Estimation With Pattern-Mixture ModelsMultiple Imputation With Pattern-Mixture Models8. Summary and ConclusionNotesReferencesAbout the Author