Objectives: In any study involving individual level data, the problems associated with incomplete observations are an obstacle to analysis. For this reason methods have been developed to complete these datasets. Multiple imputation is considered the most robust method of handling missing data, however it is also the most complex and computationally intensive. Whether multiple imputation is needed depends on the mechanism of the missing data. For example, if data is missing completely at random simpler methods can be used. For this reason, we conduct an analysis to inform the appropriate imputation method by identifying the mechanism of missing data. Methods: To determine the mechanism of missing data we fit a probit model to a dataset from a study comparing the use of Endovascular Repair (EVAR) versus the use of Open Surgical Repair (OSR) in repairing Abdominal Aortic Aneurisms. From this we determined the appropriate method to complete the dataset. We then ran a sensitivity analysis on the different methods to determine the potential consequence of utilizing the inappropriate method. Results: The results of the probit model indicated that the dataset had data which was missing at random and thus the missingness is predictable by observables in the dataset. This implied that the most appropriate method is imputation by stochastic regression or multiple imputation( the stronger of the two methods). The sensitivity analysis, however, showed no statistically significant difference between the two methods in terms of QALYs - total QALY difference between EVAR and OSR: -0.09952(-0.13202,-0.0670) for SR and -0.0866(-0.12344,-0.04977) with significant deviations from other methods. Conclusions: This study demonstrates the importance of appropriate imputation and how determining the mechanism of missing data informs the appropriate imputation method. A probit model using missing data dummies can effectively identify the mechanism of missing data and inform the appropriate method for imputation.