Data from surveys, experiments and observational studies typically contain missing values. Haphazard methods for handling missing values may lead to bias, inefficiency, reduced power and misleading measures of uncertainty. Since 1996, This email address is being protected from spam bots, you need Javascript enabled to view it and his colleagues at The Methodology Center have been working to develop improved missing-data methods for the social, behavioral and medical sciences. Much of this work involves multiple imputation, a simulation-based approach in which each missing datum is replaced by m > 1 simulated values.The resulting m versions of the complete data can then be analyzed by standard complete-data methods, and the results combined to produce inferential statements (e.g. interval estimates or p-values) that incorporate missing-data uncertainty.

Free software for multiple imputation developed by Dr. Schafer is available here.

A set of introductory pages that give an overview of the missing-data problem is also available.