The Estimation of United States Army Reenlistment Rates
Abstract
The U.S. Army uses cash selective reenlistment bonuses (SRB) to encourage soldiers in selected military occupation specialities (MOS) to reenlist. Estimates of the reenlistment rate as a function of bonus level are needed for each MOS as input to a bonus allocation model. This thesis outlines and uses a new method for predicting the reenlistment rates as a function of bonus level. The approach involves partitioning the soldier population into cells with stable reenlistment rates using demographic variables. The cells are aggregated using clustering techniques to produce groups of cells which exhibit homogeneity of reenlistment behavior. Regression models are developed for each group of cells. MOS reenlistment rates are determined as a linear combination across cells. Cross-validation techniques are used to lend credibility to the predictive model. The study points out the usefulness of identifying categories of soldiers who display unique reenlistment behavior. Integration of this technique with existing econometric reenlistment models is recommended to further improve the predictive model. Keywords: SRB, Reenlistment; Retention, Logistic regression, Hierarchical clustering,(EG)
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1989
- Accession Number
- ADA219811
Entities
People
- Michael J. Streff
Organizations
- Naval Postgraduate School