The MAD legacy: How meaningful is mean age-at-death in skeletal samples

Abstract

This study examines the representativeness of palaeodemographic reconstructions from human skeletal remains. Mean age-at-death (MAD) is the primary statistic used in interpretations of changing patterns of health and well being from palaeodemographic analyses. A series of sampling experiments were conducted on three documented 19th century samples representing the total cemetery population from which skeletal samples could be drawn. Comparisons of the age-at-death distributions of simulated skeletal samples to the parent population were made to assess the relative magnitude of deviation associated with different types of bias (age, sex, temporal). From the examples presented, variability in age-at-death distribution is high in samples of less than 100, suggesting that for samples of less than 100analyzable individuals, it is probable that the mortality profiles constructed are not an accurate reflection of the cemetery. It is proposed that whateverprocess mean age-at-death reflects for past populations (fertility or mortality), is irrelevant if the sample on which the statistic is calculated is not representative of the population. Given that most cemetery samples will be subject, differentially, to biases at a variety of levels, comparative studies based on palaeodemographic data cannot be considered reliablewithout careful control for those biases. It is suggested that representativeness is the primary theoretical obstacle for researches to overcome, and that it is necessary to shift our focus to rigorously exploring those factors that bias our samples. Without some direct quantification of the representativeness of a sample, palaeodemographic estimators such as mean age-at-death are meaningless and any subsequent interpretations regarding the past, dubious at best.