A comparison of techniques for handling missing data in scenarios with different missing data mechanisms

The purpose of this study was to illustrate the influence of missing data mechanisms on results of a multiple regression analysis and to demonstrate the influence of the use of traditional techniques (including listwise deletion, pairwise deletion and mean substitution) versus use of multiple imputation (MI) for handling missing data. A methodological approach involving a real generated dataset was adopted. Results from descriptive analyses and multiple regression models indicated that traditional missing data handling methods and MI yield similar regression coefficients and standard error estimates. Although the means and correlations are almost always biased regardless of missing mechanism and missing data techniques, the bias was less severe when using MI.