Back to search

Boston University

Assessing the accuracy of population-inclusive cranial assigned-sex-at-birth estimation models using computed tomography

Abstract

dc:description.abstract

Skeletal estimates of assigned sex at birth (ASAB) have historically relied on population-specific methods. Because applying these methods requires a prior estimate of population affinity, their utility may be limited when population affinity is unknown or intentionally excluded from the biological profile. Population-inclusive methods have been proposed as an alternative approach. The current study evaluated the generalizability of the population-inclusive cranial ASAB estimation models developed by Kelley and Tallman (2022) when applied to an independent sample. The current sample was then used to develop and compare new population-inclusive and population-specific models. The study sample included 500 adults (250 assigned female at birth and 250 assigned male at birth) representing African American, Asian American, European American, Latin American, Native American, and Unknown population affinities from the New Mexico Decedent Image Database (NMDID). Five cranial nonmetric traits were scored, and 18 cranial and mandibular measurements were collected from three-dimensional computed tomography (CT) reconstructions. First, Kelley and Tallman’s (2022) published population-inclusive nonmetric, metric, and combined models were applied to the current independent sample. New population-inclusive models were then developed using binary logistic regression and discriminant function analyses, and population-specific models were cross-applied beyond their reference populations to assess their generalizability. Intraobserver reliability was also evaluated. When applied to the current sample, the Kelley and Tallman (2022) nonmetric, metric, and combined models achieved overall classification accuracies of 88.0%, 90.8%, and 92.2%, respectively, supporting their external validity. The newly developed population-inclusive nonmetric, metric, and combined models produced overall classification accuracies of 95.4%, 91.8-92.2%, and 97.0-97.2%, with no significant differences from the corresponding population-specific models (p > 0.05). In contrast, eight of 30 cross-applications of the population-specific models resulted in significant reductions in classification performance, demonstrating that successful transfer across population groups cannot be assumed. Intraobserver reliability was substantial to almost perfect for the nonmetric traits (weighted κ = 0.696–0.927) and good to excellent for the metric measurements (ICC = 0.823–0.999). These results provide additional validation for population-inclusive cranial ASAB estimation and demonstrate that population-inclusive models perform comparably to population-specific approaches without requiring a prior estimate of population affinity. By eliminating population-affinity-dependent model selection while retaining high classification accuracy, population-inclusive methods offer a practical and ethically grounded alternative in forensic anthropological casework, particularly when population affinity is unknown or not estimated.

Degree

thesis:*
Name thesis:degree_name
Master of Science
Level thesis:degree_level
masters
Discipline thesis:degree_discipline
Forensic Anthropology
Grantor
Boston University
Year dc:date.issued
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Villalba, Haleigh M.
Advisor dc:contributor.advisor
  • Tallman, Sean D.
ORCID dc:identifier.orcid

Subjects

dc:subject × 6

Rights

Language dc:language.iso
en_US

Identifiers

dc:identifier.*
Handle dc:identifier.uri
https://hdl.handle.net/2144/53492

Chain of custody

source
Harvested from
Boston University
Base URL
open.bu.edu/oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Villalba, Haleigh M.. Assessing the accuracy of population-inclusive cranial assigned-sex-at-birth estimation models using computed tomography. masters thesis, Boston University, 2026. https://hdl.handle.net/2144/53492