IDA seeks to estimate the effect of attending the Uniformed Services University F. Edward Hebert School of Medicine (USU SoM) on military physicians’ years served on active duty. Observed differences in years served between USU SoM graduates and others likely capture preference-based selection in addition to a causal effect. To isolate the causal effect, IDA implements a two-stage estimation strategy. In the first stage, IDA uses the share of medical students from the same state that attend medical school out-of-state as an instrument for USU attendance. To account for right-censoring in the available data and non-proportional hazards, IDA uses a neural network survival model in the second stage. IDA estimates that USU SoM graduates would serve a mean of 2.25 fewer years (95% confidence interval of [0.94, 3.61]) if they graduated elsewhere.