Multiple Regression of Body Weight on Stature and Waist for Estimation in Forensic Anthropology

Questions have been present in the forensic literature for many years about the most proper and accurate method for calculation of human body weight. Many have tried and failed to measure aspects of the human skeleton and find some correlation with body weight at the time of death. This has never been possible and selected skeletal measurements regressed against known weight will bear this out. Simple measurements of the femur were regressed with weight and with stature to show that skeletal measurements in general have a higher correlation with stature than with weight. The bulk of this thesis will deal with comparison of five methods of weight regression. Body weight from a skeletal data set, the Terry collection, and a living data set, was regressed with five sets of parameters to see which method would prove to be most accurate. The multiple regression of weight on stature and waist proved to be superior and more accurate than regression with stature alone, waist alone, age, or a more obscure parameter called the index of obesity. Multiple regression proved to have the highest R² overall for the five regression methods.