An Introduction to Missing Data: Multiple Imputation

Cindy Chen, Ph.D., Assistant Professor of Biostatistics

Frank Harrell, Ph.D., Professor of Biostatistics and Chair of the Department; Director, Statistics and Methodology Core

Missing data are very common in various experimental settings, including clinical trials, surveys, and environmental studies. There are basically three classes of missing data as discussed by Rubin (1976). These are missing completely at random (MCAR), missing at random (MAR), and nonignorably missing (NINR). In this workshop, the speakers will focus on MAR scenario and introduce the principles and algorithms of multiple imputation method, including chained equations, predictive mean matching and regression imputation. Case studies for linear regression, logistic regression, and survival model will be analyzed using the multiple imputation method.

Intended Audience: The target audience is the quantitative researcher who faces the missing data problem and is interested in learning the statistical techniques.

An Introduction to Missing Data: Multiple Imputation

Please call (615) 322-8240 to request braille materials or a transcription.

July 6, 2012
Statistics and Methodology Core Training Mini-Workshop
Cindy Chen and Frank Harrell