Estimation of adult skeletal age-at-death : statistical assumptions and applications.

Abstract

We examine the statistical assumptions underlying different techniques of estimating the age-at-death of a skeleton from one or more age indicators. The preferred method depends on which property of the distribution of the data in the reference sample is preserved in the skeleton to be aged. In cases where the conditional distribution of age given indicator is preserved, we provide look-up tables giving essentially unbiased age estimates and prediction intervals, using a large reference sample and the auricular surface and pubic symphysis age indicators. Where this assumption is violated, but the conditional distribution of indicator given age is preserved, we find that an alternative model which attempts to capture the biological process of development of an individual has some attractive features, which may make it suitable for further study.