Date of Award
Master of Science in Operations Research
Department of Operational Sciences
Raymond R. Hill, PhD.
Personnel retention is a matter of great interest in the private and public sectors. In the public sector--specifically the US Air Force--maintaining appropriate retention levels of rated officers is paramount as these officers are the backbone of the Air Force’s mission. As part of the effort to ensure proper retention rates for rated officers, retention models are created by the Air Force Personnel division that assist in predicting future retention patterns and accession needs. The techniques for creating these models, known as the "sustainment line," involve utilizing average retention percentages obtained from historical data. In this study, more statistical-based methods involving logistic regression analysis and survival analysis are utilized to obtain similar retention models for rated officers. The survival analysis curve produces similar results to the sustainment line, but the sustainment line currently employed is a one-dimensional view of retention patterns. It simply models the rate at which officers leave. The value of the survival curve created in this study is that it can be updated very quickly, is flexible in its construction, and can incorporate covariates into the model that are significant to retention rates. The Air Force has long known that there are external (e.g., economic) factors that impact retention. Using a survival analysis regression model instead of simply modeling the rate at which officers leave, this study was able to identify six demographic and one economic factor that may be significant to rated officer retention. This ultimately could lead to the creation of models that reflect the retention behavior of certain subtypes of officer and give insight that could be used to tailor retention and accession programs so that they are more resource-effective.
DTIC Accession Number
Franzen, Courtney N., "Survival Analysis of US Air Force Officer Retention Rate" (2017). Theses and Dissertations. 795.