Date of Award
Doctor of Philosophy
Richard L. Jantz
Lyle Konigsberg, April Morgan, Andrew Kramer
Population variation in the morphological aging process of the pubic symphysis has generated much debate. The question of whether age parameters derived from an American population will reliably estimate age-at-death for East European skeletal populations is important since the ability to accurately estimate an individual’s age-at-death hinges on what standard is used. Consequently, successful age estimation, individual identification, and demographic profiling rests on the ability to correctly define the skeletal parameters of age-at-death.
The purpose of this study is to assess the aging process of American and East European populations and to determine what age parameters should be applied to estimate the ages-at-death for unidentified victims. A reference sample of identified individuals with known ages-at-death from Kosovo, Bosnia, and Croatia (n=861) is used to determine the age structure of victims and serves as the informed prior in the Bayesian analysis. Skeletal data comes from a subset of the Balkan reference sample and a comparative American sample. Balkan male (n=212) and female (n=84) pubic symphyses were scored in both the manners of Todd (1920 and 1924) and Suchey-Brooks (Katz and Suchey, 1986; Suchey et al., 1986). Further, repeat observations were completed by four observers for the entire Balkan sample. American male (n=1,560) and female (n=518) pubic symphyses were scored in the manner of Todd (1920 and 1921). The American data were converted from the Todd ten-phase system to the Suchey-Brooks six-phase system, as recommended by Katz and Suchey (1986) for comparative purposes.
First, the affect of inter-observer variation on the reliability of estimating age-at-death parameters from the pubic symphyses is investigated. Pearson’s r correlation coefficients among observers for the Todd system are all significant (p < 0.01) and strongly correlate among females. The correlation coefficients among males are more varied and not as strong as the female sample, ranging from r = 0.309 – 0.738. Correlation coefficients among observers for the Suchey-Brooks method are more consistent and higher than those for the Todd method, ranging from r = 0.866 – 0.939 among females and from r = 0.710 – 0.844 among males. Wilcoxon signed ranks tests demonstrate significant differences among observers for both methods.
Second, population variation in the aging processes of American and East European populations is tested using proportional odds probit regression. An analysis of deviance is calculated using an improvement chi-square to test the adequacy of the model. No significant association between the aging process of pubic symphyses and population is found (df = 1, chi-square likelihood ratio = 3.209, p = 0.10). However, when males and females are treated separately, there is a significant association among females and the population (df = 1, chi-square likelihood ratio = 15.071, p = 0.001).
Finally, a Bayesian statistical analysis is utilized to establish accurate age parameters. The ages-of-transition for each phase are calculated using an unrestrictive cumulative probit model for the Balkan males and the log-age cumulative probit model for the Balkan females. The probability density functions (PDF) for the posterior distributions of age at each symphyseal phase are calculated. The age estimates are based on the calculated age distribution from the Gompertz-Makeham hazard analysis and the ages-of-transition. To estimate the age-at-death for an individual, the highest posterior density regions for each Symphyseal phase is calculated and four different regions are provided (95%, 90%, 75%, and 50%).
Kimmerle, Erin H., "Biological and Statistical Variation in Age Estimation from Pubic Symphyseal Morphology with Regard to Individual Identification and Demographic Profiling. " PhD diss., University of Tennessee, 2004.