Date of Award

December 2019

Degree Type

Thesis

Degree Name

Master of Science

Department

Anthropology

First Advisor

Joseph P Gray

Second Advisor

Fred Anapol

Committee Members

Emily Middleton, Robert Jeske

Keywords

Anthropology, Archaeology, Radiogrpahs, Sexual Dimorphism, Tibial Shape, Tibial Size

Abstract

SEXUAL DIMORPHISM AND THE SHAPE OF THE PROXIMAL TIBIA IN A RADIOGRAPHIC SAMPLE

This study investigates the use of radiographs to determine sexual dimorphism in the shape of the tibia. The goal of the research was to identify a small set of markers that would allow researchers to efficiently and accurately determine a person’s sex from a radiograph of the proximal tibia.

The sample consisted of radiographs including 75 females and 46 males ranging in age from 21 to 81. Measurements were taken on 27 points around the area of the knee including the tibia, patella, and femur. The measurements were converted into a compositional data set and the log-ratios between all measurements were analyzed with a classification and regression tree procedure (CART) to predict sex. The sample was divided into training and testing sets prior to analysis. In the training data only three log-ratios were needed to achieve an accuracy of 87.78% for sex determination. The accuracy rate fell to 70.00% in the testing data. This success rate compares favorably with the results achieved by Toon (2014) in her osteological analysis of 200 tibia from the WM Bass Donated Skeletal Collection. Toon used a geometric morphometric approach to define a linear discriminant function that achieved a cross-validated accuracy rate of 57% – 59%. The approach illustrated in my study has the advantage of achieving a higher accuracy while requiring many fewer data points.

Two extensions of the present study are required before its usefulness to forensic anthropologists and archaeologists can be evaluated. First, it must be established whether the log-ratios that identified as useful in predicting sex by the CART procedure are consistent across populations. Second, the log-ratios need to be validated as useful when used to analyze actual, physical skeletal remains.

Download

Share

COinS