Files in this item



application/pdfLEE-DISSERTATION-2019.pdf (5MB)
(no description provided)PDF


Title:Multivariate asymmetry and allometry in human limb bones as a basis for osteometric sorting
Author(s):Lee, Amanda B.
Director of Research:Konigsberg, Lyle W.
Doctoral Committee Chair(s):Konigsberg, Lyle W.
Doctoral Committee Member(s):Shackelford, Laura L.; Hughes, Cris E; Ubelaker, Douglas H
Department / Program:Anthropology
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):osteometric sorting
limb bones
multivariate statistics
forensic anthropology
commingled assemblages
biological anthropology
Abstract:The size and shape of a limb bone relative to another bone holds a great amount of information needed to individualize a commingled assemblage. This study maximizes discriminating information found in long bone measurements with multivariate analyses of allometry and asymmetry. Previous research on osteometric sorting are univariate in nature and often make fallacious assumptions of perfect symmetry in long bones. These methods have led to issues of specious statistical assumptions and high error rates in application. This study examines measures and applications of human limb bone asymmetry and allometry in order to improve current methods for individuating commingled human remains. Patterns of asymmetry and allometry are characterized using large data sets (n > 2,000 people) of bone measurements from diverse populations. Based on this information, these patterns are used to develop novel methods in R for the sorting of mixed skeletal elements using multivariate metrics. There are three sets of results. First, the distributional assumptions of previous sorting algorithms are questioned and re-assessed. Through statistical analysis, the true distributional form of human limb asymmetry is described. These findings represent the first characterization of asymmetry distributions. Second, an automated method for osteometric pair-matching using bilateral limb asymmetry is produced and evaluated. Multiple trials are conducted comparing the diagnostic abilities of this method against previous pair-matching algorithms. The results from the new method show an improvement in performance over the other current tests for the humerus and radius, a match in performance for the femur, and underperformance for the tibia. Third, an automated method for osteometric sorting that can match several different bones at once is created and tested. Since there are currently no automated techniques for the matching of more than two bones at a time, this new method is not compared to any other algorithm and the statistics for diagnostic accuracy are presented alone. The results of this study demonstrate that the matching of more than two bones at a time reduces the amount of potential bone combinations that need to be assessed by-hand from hundreds of thousands to more manageable numbers. In total, the results presented here suggest that the incorporation of multivariate statistics in osteometric sorting can improve the diagnostic ability of automated methods and demonstrate the feasibility of the application of research in real world-contexts.
Issue Date:2019-11-22
Rights Information:Copyright 2019 Amanda Lee
Date Available in IDEALS:2020-03-02
Date Deposited:2019-12

This item appears in the following Collection(s)

Item Statistics