A Study of the Accuracy and Reliability of Sex Estimation Methods of the Human Pelvis

ABSTRACT\ud A STUDY OF THE ACCURACY AND RELIABILITY OF SEX\ud ESTIMATION METHODS OF THE HUMAN PELVIS\ud by\ud Brenna Kay Blanchard\ud Master of Arts in Anthropology\ud California State University, Chico\ud Spring 2010\ud Sex determination is an important first step in the development of a biological\ud profile in human osteology, as methods to estimate other components of the profile\ud depend on sex. The pelvis is the most sexually dimorphic anatomical region due to obstetrical\ud requirements. This thesis tests sex estimation methods for the pelvis from three\ud sources to determine rates of reliability and accuracy. The sex estimation methods recommended\ud in Buikstra and Ubelaker, the Bruzek method, and the Murphy method were\ud evaluated using remains curated at the California State University ??? Chico Human Identification\ud Laboratory and the William M. Bass Donated Skeletal Collection at the University\ud of Tennessee, Knoxville.\ud Intra- and interobserver error studies were performed to test reliability between\ud two observations by the author and between the author and a second observer.\ud xiii\ud Intra-rater reliability was high for most traits and combinations of traits. The opposite\ud was the case for inter-rater reliability. Concordance was assessed using Cohen???s kappa\ud and paired t-tests.\ud All traits, combinations of traits, and sex determinations for the Buikstra and\ud Ubelaker and Bruzek methods were statistically significant when compared with biological\ud sex. The Murphy method was statistically significant between the sexes. The\ud morphological methods were evaluated using Pearson???s Chi-square and the Murphy\ud method was assessed using independent samples t-tests. The Bruzek method categorized\ud a high percentage of individuals as indeterminate. A revised version of the Bruzek\ud method was proposed, tested, and resulted in a high correct classification rate and low\ud error rate.\ud A movement toward standardization is occurring in the forensic sciences,\ud which includes finding error rates for identification methods. Validation studies are important\ud for the future of the forensic sciences and the process of standardization because\ud they establish the accuracy, reliability and error rates of previously published methods.CSU, Chic