ABSTRACT
Multiple imputation (MI) is a popular and accessible method of model-based imputation. Standard references for MI include Rubin (1987) and Little and Rubin (2002). Multiple imputation is flexible and therefore has a number of specific implementations. However, the three basic steps to MI are:
1. Impute the missing data (typically) using Bayesian predictive distributions of missing data, conditional on observed data, resulting in multiple (m) completed data sets.
