Date of Award


Degree Type


Degree Name

Master of Arts



Major Professor

Susan R. Frankenberg

Committee Members



Physical anthropologists and bioarchaeologists often seek to generate biological profiles of individuals represented by skeletal remains. One particularly informative component of the biological profile is skeletal age-at-death. Age-at-death estimation is vital to numerous contexts in both paleodemography and forensic anthropology. Throughout the history of the discipline, numerous authors have published methods for adult age-at-death estimation. These methods have proved invaluable, but they are not free from error. As a result, workers have continually worked to improve the methodological toolkit for estimating age-at-death.

In June of 1999, researchers gathered in Rostock, Germany for the sole purpose of evaluating and testing age-at-death estimation methods. The hallmark of this symposium was a theoretical framework known as the Rostock Manifesto published in volume edited by Hoppa and Vaupel (2002a) entitled Paleodemography: age distributions/rom skeletal samples. Included in this work was a new age-at-death estimation method called transition analysis published by Boldsen and colleagues. Transition analysis utilizes traits of the pubic symphysis, auricular surface, and cranial sutures to produce likelihood age-at-death estimates. In their publication, Boldsen et al. (2002) report a remarkable correlation between estimated age and real age in addition to asserting that this method adequately ages individuals in the 5O+ years category.

This purpose ofthis research was to perform a validation study of the transition analysis method by utilizing 225 skeletons from the William M. Bass Donated Skeletal Collection curated by the Forensic Anthropology Center at the University of Tennessee. Data were collected in the manner of Bolds en et al. (2002) and used to generate age-at- death estimates. These results were then statistically compared to known ages from the Bass Collection. Results from the study were not as favorable as those published by Boldsen and colleagues. Correlation coefficients were low and analyses of data using the forward continuation ratio, ordinal cumulative pro bit, and unrestrictive cumulative probit models suggest such problems arise from a combination of the method's statistical framework and its lack of applicability.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Included in

Anthropology Commons