Multiple Imputation (MI) is a well established technique for handling missing data, although its conceptual and computational aspects are often confused. It was originally developed in the sample survey setting by Rubin in the 1970’s. Since then, its range of application areas has steadily
increased. This talk will focus on research on MI that has been done in the last few years at the London School of Hygiene and Tropical Medicine, in particular in developing the approach for multilevel data and for sensitivity analysis. The latter is a role envisaged by Rubin at the inception of MI, but not greatly developed in practice.
Presented by:
Mike Kenward (London School of Hygiene and Tropical Medicine)
Date & time:
October 25, 2010 3:00 pm
External seminars home