LENGTH-OF-SERVICE / SURVIVAL PROFILES METHODOLOGY FOR THE ROYAL AUSTRALIAN NAVY

Using transactional data from Royal Australian Navy (RAN) HR systems from 2002 to 2018, this thesis produces length-of-service/survival profiles for the RAN Officers and Sailors with a focus on testing for heterogeneity in survival profiles along their characteristic dimensions of rank, workforce category/workgroup, gender, age group and calendar year of enlistment cohort. These survival profiles will inform the RAN about the probability of a member remaining in the service at any point in their period of service. The dataset contained 21,820 periods of service comprising enlistment and separation events (or enlistment and final population snapshot events) and represented 21,495 RAN individuals. I determined that all analyzed characteristics were important for predicting RAN separation behavior. In particular, females were significantly more likely to separate than males in the early years (up to 10 years of service). Sailors were more likely to separate than Officers. Across the whole RAN, the first year separation rate is 10.3%, which represents the initial matching period for new recruits. Two methodologies for preparing survival profiles were compared and validated. Kaplan-Meier was found to be the best for single characteristic variable models, and the Cox proportional hazards model was found to be the best for multiple variable specifications.http://archive.org/details/lengthofservices1094561350Lieutenant Commander, Navy, AustraliaApproved for public release; distribution is unlimited