Auxfinder - Finding auxiliary variables for imputation models

To deal effectively with missing data, powerful statistical approaches such as MICE and FIML have been developed. To make optimal use of these methods, researchers can include additional variables in the estimation process that are not part of the final analytical model but that improve the prediction of missing values. The selection of these auxiliary variables is often based on theoretical considerations; however, particularly when large datasets are available, a data-driven selection can yield further improvements. We introduce auxfinder, a new Stata command that enables users to quickly identify the most promising auxiliary variables.