Can-Do Vs. Will-Do Factors: Predicting the Gold-Standard Marine

Abstract

The Marine Corps has historically used the high school diploma and Armed Services Vocational Aptitude Battery scores to define a high-quality enlisted Marine. This industrial-era approach fails to consider the enlistee holistically, despite evidence that a combination of cognitive and non-cognitive assessments paints a more complete picture of an enlistee. In addition to utilizing outdated recruitment methods, the current manpower system fails to identify where a particular Marine falls on a range of skills, with the extremes being generalist and specialist. Using factor analysis, machine learning, and multivariate logistic regression, this research utilizes existing personnel data to develop proxy variables that support Marine Corps efforts to better predict which enlistees will be gold-standard Marines, as well as predicting whether an enlisted Marine is a generalist or specialist. Given that proxy variables are generated to replace data that is provided by the Tailored Adaptive Personality Assessment System (TAPAS), the Marine Corps should validate the predictive accuracy of these models using TAPAS data once it is available. The bottom line is that this research provides evidence that the current manpower and recruiting systems can be refined to support more accurate decision making that will enable the Marine Corps to achieve future manpower and operating environment requirements.