Improved screening for Navy enlistment

Abstract

This thesis analyzes the effect of selected demographic characteristics on first-term enlisted attrition from the U.S. Navy. The characteristics include age, marital status, dependency status, gender, race, Armed Forces Qualification Test (AFQT) score, and education credential. The analysis draws from a Defense Manpower Data Center file containing Navy enlisted cohorts of recruits from fiscal years 1999 through 2003. Probit regression models are constructed using these data to identify differences in the attrition likelihood of recruits who possess the selected characteristics. Results show that the current Educational Tier system is flawed with respect to education credential assignment and attrition predictability. The data also reveal that different factors correlate with attrition during the first 90 days (or less) of service and attrition occurring later. Finally, dependency status of single Sailors is found to be the single strongest predictor of attrition once education and aptitude (AFQT score) are controlled. The Navy uses AFQT score and Educational Tier to determine enlistment eligibility. This thesis presents matrices for screening applicants based on education credential, AFQT score, age, marital status, and dependency status, with the intent of more accurately predicting first-term attrition.