According to official United States life tables, black males and females have higher death rates than their white counterparts at each age through about 75, after which they have lower rates. The implausibility of the precipitous decline in the ratio of black to white mortality rates in the older ages motivates this investigation into the accuracy of population and death data in the United States. Differences between whites and blacks in the type and/or extent of age misstatement could produce a series of age-specific death rates such that a crossover in mortality would be merely an artifact of poor data. In order to construct correct life tables, it would be most useful to have at our disposal a set of techniques with which we could examine the consistency between census and death registration data as well as that within each of the individual sources of data. By so doing we could adjust for apparent inconsistencies and extract unbiased estimates of the various life table parameters. Such an array of methods has only recently been developed. Implementing these methods, the analysis will result inlife tables reflecting the mortality experience of native-born Americans categorized by race (black/white) and sex. These life tables will be purged of the biasing effects of age misstatement in deaths and population, underenumeration in the censuses, and incomplete registration of deaths.
