Panel regression formulas for estimating stature and body mass from immature human skeletons: a statistical approach without reference to specific age estimates

Abstract

Anthropologists require methods for accurately estimating stature and body mass from the human skeleton. Age-structured, generalized Least Squares (LS) regression formulas have been developed to predict stature from femoral length and to predict body mass in immature human remains using the width of the distal metaphysis, midshaft femoral geometry (J), and femoral head diameter. This paper tests the hypothesis that panel regression is an appropriate statistical method for regression modeling of longitudinal growth data, with longitudinal and cross-sectional effects on variance. Reference data were derived from the Denver Growth Study; panel regression was used to create one formula for estimating stature (for individuals 0.5–11.5 years old); two formulas for estimating body mass from the femur in infants and children (0.5–12.5 years old); and one formula for estimating body mass from the femoral head in older subadults (7–17.5 years old). The formulas were applied to an independent target sample of cadavers from Franklin County, Ohio and a large sample of immature individuals from diverse global populations. Results indicate panel regression formulas accurately estimate stature and body mass in immature skeletons, without reference to an independent estimate for age at death. Thus, using panel regression formulas to estimate stature and body mass in forensic and archaeological specimens may reduce second stage errors associated with inaccurate age estimates.