Master of Arts (MA)


Geography and Anthropology

Document Type



Analyses of human skeletal remains by physical anthropologists necessitate precise methods for the determination of age at death. Physical anthropologists recognize that the most accurate estimations of age at death are produced from evaluation of multiple indicators of age. To this end, three skeletal sites, the auricular surface, pubic symphysis, and right fourth sternal rib end, are often evaluated for age in adult skeletal remains. However, as of yet, no standardized method of combining the respective estimates of age provided at these sites has been reported. The above mentioned skeletal sites of 49 females and 77 males from the Hamann-Todd collection were evaluated for age at death. Regression analyses were used to assess the relationship between the estimated age and actual age. Furthermore, regression analyses were used to combine the estimates of age in an effort to provide an accurate estimate of age at death. Results of the regression analyses show that a relationship exists between the respective estimates of age provided at each site and actual age. The values of R for regressions on the male and female samples reflect an 85% and 90% correlation between the actual ages of individuals and the predicted ages resulting from combination of the respective estimates through linear regressions. However, the large predicted age ranges associated with predicted ages make the results of the regressions questionable for forensic use. Although mean absolute differences between the predicted and actual ages for the male and female samples are low, 4.35 and 5.62 years, respectively, the large maximum absolute differences demonstrate the vast amount of variability possible in indicators of age in the human skeleton. An exception to the problem of large predicted age ranges produced through regression analysis occurred for the sample characterized as young adult males. This group showed both narrower predicted age ranges of roughly 11 years and a maximum absolute difference between actual and predicted age of 8.43 years. However, none of the other sectioned samples showed similar results, making the method useful for only a small sample of human skeletal remains submitted for analysis. Regression analyses performed after the removal of outliers on samples divided into young and mature adults did provide more accurate estimates of age at death and smaller more useful predicted age ranges. However, use of these models requires that outliers be recognizable in a laboratory context without comparative data, which may not be practical. Ultimately, the possibility of using additional skeletal indicators of age at death could serve to narrow the predicted age ranges provided by regression analyses. This research question should be addressed in future studies.



Document Availability at the Time of Submission

Release the entire work immediately for access worldwide.

Committee Chair

Mary Manhein