Boston University
Testing population-inclusive assigned-sex-at-birth estimation methods on a documented skeletal sample
Abstract
dc:description.abstractHistorically, assigned sex at birth (ASAB) estimation techniques have relied on ancestry estimation and population-specific methods. In cases where ancestry is unknown or is intentionally removed from the biological profile, population-inclusive ASAB estimation methods incorporating significant variation are vital. Therefore, this study tests Kelley and Tallman’s (2022) population-inclusive ASAB estimation method—developed from a demographically diverse U.S. sample of computed tomography scans of the skull—on skeletonized individuals from the University of Tennessee, Knoxville’s (UTK), Donated Skeletal Collection. Standard cranial nonmetric and metric variables were collected on 217 assigned females at birth and 222 assigned males at birth. Kelley and Tallman’s population-inclusive nonmetric, metric, and combined models were applied. These classification accuracies were compared to binary logistic regression and discriminant function accuracies developed on the UTK sample. When applied to the UTK sample, Kelley and Tallman’s models resulted in accuracies of 85.6% (nonmetric), 79.3% (metric), and 82.9% (combined nonmetric/metric). These are lower than the cross-validated classification accuracies developed from the UTK sample: 89.5% (nonmetric), 86.7% (metric), and 90.2% (combined). A third set of models were developed using the UTK dataset, but employed the same combination of traits utilized in the Kelley and Tallman (2022) models. When applied to the data, these UTK-KT models resulted in accuracies of 87.9% (nonmetric), 85.6% (metric), and 89.1% (combined). The reduced performance can be attributed to the heterogeneity of Kelley and Tallman’s sample, which differs significantly from the more homogenous (i.e., majority European American) UTK sample. However, the results indicate that while sample-tailored models may perform better, population-inclusive models developed from CT scans are applicable to dry bone and can produce forensically meaningful sex estimations that do not significantly differ from those generated by population-specific methods.
Degree
thesis:*- Name thesis:degree_name
- Master of Science
- Level thesis:degree_level
- masters
- Discipline thesis:degree_discipline
- Anatomy & Neurobiology
- Grantor
- Boston University
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Conquest, Victoria
- Advisors dc:contributor.advisor
-
- Tallman, Sean D.
- Moore, Tara L.
- ORCID dc:identifier.orcid
Subjects
dc:subject × 1Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/2144/53147