MODELING FIRST-TERM ENLISTMENT COMPLETION FOR THE SELECTED MARINE CORPS RESERVE

Abstract

Military retention studies seek to better explain the factors affecting a service member’s decision to complete contractual service agreements. For the purposes of meeting the national security needs of the nation, these studies inform decision makers about recruiting and sustaining a ready and capable force. This thesis explores the significant factors that predict first-term enlistment completion for non-prior service (NPS) enlisted Marines in the Selected Marine Corps Reserve (SMCR). The data available from the Total Force Data Warehouse (TFDW) and provided by Manpower and Reserve Affairs (M&RA), Quantico, Virginia, contains information about first-term NPS enlisted Marines as they progress in the SMCR. A cohort of Marines who enlisted in the SMCR in fiscal year 2009 is selected and tracked for a period of eight years to determine a binary outcome of successful completion. Utilizing binary logistic and random forest regression, two distinct models are developed to determine the significant features indicative of successful completion of first-term enlistments. The first model examines information available prior to entering the SMCR. The second model examines data available prior to enlistment and data obtained as the individual completes obligated service. The end state is to select the best model fit for each distinct model to inform decision makers about first-term enlistment completion.