Predicting Naval Aviator Attrition Using Economic Data.

Understanding and accurately predicting attrition is vital to correctly managing the retention of naval aviators. This thesis investigates the ability of models incorporating a number of economic measures to predict naval aviator attrition rates. Using data from 1978 to 1990, this study examined a wide range of potential economic explanatory variables and their effects on naval aviator attrition rates. The naval aviator data set was grouped into six populations, separated by aviation community helicopter, jet and propeller and by years of service 5-8 and 9-12. Three separate linear regression models for each of the aviator groups were developed, and their predictive ability evaluated. The study found that no single model was best at predicting attrition rates for all groups simple models using one or two variables performed better than complex, multivariate models the most useful predictor variable was the national unemployment rate attrition rates with the highest levels and variability were in the jet and propeller pilot groups with five to eight years of service, and the most significant models, able to outperform a naive prediction, were found for these groups.