Multiple imputation (MI) is a statistical technique that can be used to handle the problem of missing data. MI enables the use of all the available data without throwing any away and can avoid the bias and unrealistic estimates of uncertainty associated with other methods for handling missing data. In MI, the missing values in the data are filled in or “imputed” by sampling from distributions observed in the available data. This sampling is done multiple times, resulting in multiple datasets. Each of the multiple datasets is analysed and the results are combined to give overall results which reflect the uncertainty about the values of the missing data. This talk will explore what MI is, when it can be used and how to use it. The content will be accessible to a wide audience and illustrated with clear examples.
Disclosure of Interest None declared