A decomposition analysis of first-term attrition in the U.S. Military

Abstract

his thesis analyzes causal factors associated with first-term attrition for all four military services. In particular, it seeks to identify demographic and other factors that have influenced changes in attrition over time. The thesis draws on data provided by the Defense Manpower Data Center on entry cohorts for fiscal years 1984, 1989, and 1994. Separate multivariate models are estimated for each service and each year. These models are used to implement a decomposition analysis of the changes in attrition between 1984 and 1989, between 1989 and 1994 and between 1984 and 1994. The decomposition technique analyzes the portion of the changes in attrition over these periods that is attributable to changes in the demographic composition of the entry cohorts and the portion due to changes in the estimated model coefficients. The thesis finds that sex, education, race, AFQT scores, and months spent in Delayed Entry Program consistently affect attrition behavior while the relationship between age at entry and attrition is not clear. The decomposition technique used in the thesis finds that there are generally big differences between the predicted and the actual changes in attrition and that the direction of predicted and actual change is generally in opposite directions. The thesis recommends that the role of other factors, such as service-specific policies be researched to keep attrition from further rising and that the decomposition technique be replicated for other beginning and end points.