Modeling body height in prehistory using a spatio-temporal Bayesian errors-in variables model

Abstract

Body height is commonly employed as a proxy variable for living standards among human populations. In the following, the human standard of living in prehistory will be examined using body height as reconstructed through long bone lengths. The aim of this paper is to model the spatial dispersion of body height over the course of time for a large archeological long bone dataset. A major difficulty in the analysis is the fact that some variables in the data are measured with uncertainty, like the date, the sex and the individual age of the available skeletons. As the measurement error processes are known in this study, it is possible to correct this using so-called errors-in-variables models. Motivated by this dataset, a Bayesian additive mixed model with errors-in-variables is proposed, which fits a global spatio-temporal trend using a tensor product spline approach, a local random effect for the archeological sites and corrects for mismeasurement and misclassification of covariates. In application to the data, the model reveals long-term spatial trends in prehistoric living standards.