Bayesian sex estimation from permanent canine metrics in an Argentinian documented collection.

OBJECTIVE: To validate a Bayesian method for sex estimation from permanent canines (Luna, 2019) on an independent contemporary skeletal sample from Argentina. DESIGN: The study analyzed 152 permanent canines (61 upper, 91 lower) from 98 individuals (55 males, 43 females) included in the Prof. Dr. Rómulo Lambre Reference Collection (La Plata, Argentina). Buccolingual and mesiodistal diameters of the crown and neck region were measured. Bayesian posterior probabilities were computed for single- and multi-variable models, and classification accuracy was assessed against documented biological sex. RESULTS: The buccolingual neck diameter was the best single predictor, yielding accuracy rates of 73.77% (lower canines, p = 0.74) and 73.62% (lower canines, p = 0.74). Multi-variable models improved performance: using only the neck diameter, lower canines achieved 81.96% accuracy (p = 0.78), and upper canines, 80.21% (p = 0.80). The four-variable model produced the highest overall accuracy (lower: 85.24%, p = 0.85; upper: 83.51%, p = 0.83). Sex-specific correct classification rates ranged 80.77-83.33% for females and 83.63-88.57% for males, with posterior probabilities of 0.78-0.88 and 0.80-0.86, respectively. CONCLUSIONS: Neck metrics demonstrate strong performance for sex estimation, proving particularly valuable for commingled or isolated remains. This Bayesian probabilistic framework provides transparent uncertainty estimates essential for forensic standards and can be successfully applied to contemporary non-identified individuals of similar geographical provenance.