Mortality Crossovers: Reality or Bad Data?

Abstract It has been argued in the literature that the observed mortality crossover among older black Americans relative to the white population is a result of 'differential early mortality which selects the least robust persons from the disadvantaged population at relatively earlier ages so that, at advanced ages, the disadvantaged population has proportionately more robust persons' (Kenneth G. Manton). The authors examine the plausibility of the observed black mortality crossover and the heterogeneity argument supporting its existence. In addition to citing evidence from the literature, they use life tables from various countries known to have good mortality data to explore the relation between mortality in childhood and at younger adult ages and mortality in old age for cohorts and periods. Analysis suggests that the association between childhood and old-age mortality for cohorts is positive, implying that observed mortality crossovers are produced by deficient data rather than population heterogeneity.