An exploratory study of pre-admission predictors of hardiness and retention for United States Military Academy cadets using regression modeling

Abstract

This study uses regression techniques on United States Military Academy (USMA) cadet/ candidate data in order to develop a hardiness-prediction model and explore retention during and after graduation from USMA. We created several data sets using 42 variables from three cohorts (N= 3,716) and analyzed them using regression techniques. Preliminary results showed high school type and the interaction between gender and parents education level as significant. Specifically, private religious high schools and male cadets with less-educated fathers are positive predictors of hardiness (R2 = 0.05). Model quality improved in subsequent regressions by identifying a target population. Among varsity football players (N= 149), less-educated mothers and liberal political views are negative predictors of hardiness while race and parents military service history (African Americans with fathers who served in the military) and prep school attendance are positive predictors of hardiness (R2 = 0.97). Logistic regression results suggest military, physical, and academic performance are positive predictors of USMA retention while hardiness-challenge, participation in varsity athletics, and less-educated fathers are negative predictors. Logistic regression results identified basic branch as the sole positive predictor of U.S. Army officer retention beyond a USMA graduates sixth year of active federal service. Infantry officers, followed by military police, armor and engineers, remain in service longer (medical corps and aviation branch officers excluded).