Military Service and Mortality: A Reappraisal Based on Frailty Models

Douglas A. Wolf, Syracuse University
Janet Wilmoth, Syracuse University

Several investigators have examined the influence of military service on men’s life course trajectories, including health outcomes. Studies of late-life outcomes using samples of survivors, such as the HRS, must contend with a potentially serious initial-conditions problem: if military service changes mortality risks during the pre-HRS years, then the chances of surviving from the completion of military service until the HRS baseline will differ between veterans and non-veterans. We address this problem in two ways: we use estimates of pre-HRS survivorship based upon decennial Census data to adjust mortality models estimated using HRS data; and, we develop cohort mortality models that explicitly represent unmeasured heterogeneity, using the analytic machinery of fixed-frailty mortality dynamics. The HRS data fail to produce evidence of differential mortality by veteran status; in contrast, 1960 and 1990 Census data produce clear evidence that black veterans have lower death rates than black non-veterans.