A Method for Detecting Bias in Human Archaeological Cemetery Samples

Abstract

This paper aims to provide a methodological approach to identify potential bias in cemetery sample age-at-death distributions and provide an alternative way to report fertility despite underenumeration. The method involves comparing total fertility rate (TFR) estimates from two empirically derived models developed on a United Nations mortality and fertility dataset. The models utilize different age cohorts in their calculations (one relies on the proportion of pre-adults aged < 15 years, whereas the other excludes all those aged < 15 years). The tested hypothesis is that similar TFR estimates using both models indicate a relatively unbiased sample, although the converse would suggest cemetery sample bias in one broad age cohort. Results comparing the respective TFR estimates from D0–14/D and D15–49/D15+ models confirm that fertility estimates are comparable for unbiased samples. From this, a method for the coordinated application of the D0–14/D and D15–49/D15+ models were found to be valid in determining if a cemetery sample was biased. Following the determination of potential underrepresentation, an approach is outlined for dealing with biased and unbiased cemetery samples in terms of reporting on demographic variables such as TFRs.