Investigating Navy officer retention using data farming

Abstract

The allocation of nearly 30% of the Navy’s budget to personnel costs, and the importance of manning fleet requirements to maintain operational readiness create a critical need for the Navy to effectively manage the size of the force. The Navy’s personnel planners use the Officer Strategic Analysis Model (OSAM) to project officer end-strength based on policies, plans, and historical loss rates. The application of data farming to this model allows for investigation of different scenarios that can provide insight into both the behavior of the model and the behavior of the officer corps under various conditions. This study uses Design of Experiments (DOE) techniques to develop and implement an experimental design that determines the degree of stochastic variation in OSAM and explores the effect of a three-year period of poor retention of Unrestricted Line (URL) officers in paygrades O3 through O6. Analysis of results across multiple replications of a single design point indicate that OSAM produces very little stochastic variation. Regression modeling of the results allows planners to accurately and precisely predict the effect of this poor retention scenario on specific groups. This predictive capability provides the opportunity for proactive approaches to solving potential retention problems.