Maximum Likelihood and Bayesian Estimation of Skeletal Age-at-Death from the Human Pubic Symphysis

A number of methodological problems have recently plagued studies of adult skeletal age-at-death estimation. Over the last two decades, researchers have extended considerable effort to place age estimation studies on a firmer statistical ground. However, many of the current methods can still be criticized because they make unjustifiable assumptions or use inappropriate statistical models. Much of the controversy surrounding age-at-death estimation has focused specifically on the question of applying age standards from a reference collection of known-age individuals to a target group of unknown age.

The current study, involving a large sample (n=739) of adult male pubic symphysis data, demonstrates a probability-based method in order to obtain the full posterior distribution for age-at-death conditional on observed symphyseal phases using both a maximum likelihood and Bayesian estimator. With the application of the maximum likelihood or Bayesian estimator (where the prior distribution for age is external to the reference sample) it is possible to produce age estimates that are independent of the reference sample age-at-death distribution.