Attrition in the All Volunteer Force: A Prediction Model for Non High School Graduates

摘要

FY74 and FY75 Army enlistment data on non prior service, non high school graduates were used to estimate the parameters of both a grouped linear and grouped logistic attrition probability assessment model based on individual demographic attributes. Both models yielded consistent results finding significant inverse relationships between years of education, age and AFQT group standing and the dependent variable - probability of attrition. The models were then tested using FY76 data. Both model forms did equally well in terms of prediction and both did a better overall prediction of job as compared to using either education level or AFQT group as the sole selection criterion. The implications of the analysis are discussed in terms of the relative expected value of alternative recruiting cost/attrition risk tradeoffs between different labor market segments. (Author)