Missing Data Imputation Using the Multivariate t Distribution

When a rectangular multivariate data set contains missing values, missing data imputation using the multivariate t distribution appears potentially useful, especially for robust inferences. An efficient technique, called the monotone data augmentation algorithm, for implementing missing data imputation using the multivariate t distribution with known and unknown weights, with monotone and nonmonotone missing data, and with known and unknown degrees of freedom is presented. Two numerical examples are included to illustrate the methodology, to compare results obtained using the multivariate t distribution with results obtained using the normal distribution, and to compare the rate of convergence of the monotone data augmentation algorithm with the rate of convergence of the (rectangular) data augmentation algorithm.