Estimation indirecte de l'âge en paléodémographie : approche bayésienne (original) (raw)

In paleodemography, the age at death can only be estimated by means of biological indicators and reference data which provide estimates of the conditional distribution of age knowing an indicator. Several methods have been proposed for the estimation of age-structure from these data. Unfortunately, they do not work well in most cases, due to the basically unstable structure of the problem as well as the small size of available samples. We show how to improve the estimation by introducing a simple Bayesian method to incorporate useful prior information on the unknown parameters. Mots cles : Estimation, methodes bayesiennes, paleodemographie, lois de Dirichlet, loi de mortalite.