Multiple imputation: The Monte Carlo technique in which missing values in the data set are replaced by m> 1 simulated versions, where m is usually small (say 3-10). Each of simulated complete datasets is analyzed by the technique appropriate to the investigation at hand, and results are later combined to generate estimates, confidence intervals etc. The imputations are created by the Bayesian approach which needs specification of the parametric model for the complete data and, if necessary, a model for mechanism by which data become missing.

Hear also required is a prior distribution for unknown model parameters. Bayes' theorem is taken in use to simulate m independent samples from the conditional distribution of the missing values provided the observed values. In most of the cases special computation techniques such as Markov chain Monte Carlo methods will be required.