Formulation of Regression Models for Stature Estimation Using Selected Anthropometric Variables

Abstract

Background: Stature is a vital anthropometric parameter in forensic science, anthropology, and medical practice. In situations where complete body remains are unavailable, reliable regression models using alternative body measurements are essential for stature estimation. Objective: This study aimed to develop regression models for estimating stature using selected anthropometric variables, elbow height (EH), buttock-knee length (BKL), and buttock-popliteal length (BPL), in a Nigerian population. Methods: A cross-sectional study was conducted on 240 adult participants. Standard anthropometric procedures were employed to measure stature, EH, BKL, and BPL. Data were analyzed using descriptive statistics, independent t-tests, and linear regression. Multicollinearity was assessed using the variance inflation factor (VIF), and model accuracy was evaluated with the standard error of estimate (SEE). Results: The average height was 169.96 ± 8.02 cm, average EH was 20.99 ± 2.56 cm, average BKL was 58.53 ± 3.51 cm, and average BPL was 49.04 ± 3.27 cm. Regression analyses showed that BKL and BPL were significant predictors of height (p < 0.05), while EH was not significantly associated. The regression models demonstrated good predictive accuracy (SEE < 1, VIF < 2). Sex-specific analyses indicated slightly higher predictive accuracy among females (r = 0.61, SEE = 5.159) compared to males (r = 0.56, SEE = 5.846). The overall regression model produced a moderate correlation coefficient (r = 0.61) with an SEE of 6.411. Conclusion: BKL and BPL are reliable predictors of stature in this Nigerian population, whereas EH is less useful. The regression models developed provide a population-specific tool for stature estimation, with practical applications in forensic identification, medical evaluation, and ergonomic design. Future studies with larger sample sizes and additional anthropometric parameters are recommended to enhance predictive accuracy.