1Forensic Science Centre, The University of Western Australia, 2Forensic Science Centre, The University of Western Australia, 3Dipartimento di Scienze Chimiche e Geologiche, Università di Modena e Reggio Emilia, l.go S. Eufemia 19, 41121 Modena, Italy., 4Centre for Anatomical and Human Sciences, The Hull York Medical School, The University of York, Heslington, York, UK
Friday Morning, 301D
The statistical quantification of error and uncertainty is inherently intertwined with ascertaining the admissibility of forensic evidence in a court of law. This issue was especially advocated in the 2009 National Academy of Science review of forensic practice. In the forensic anthropological discipline, the robustness of any given method and classification statistics should not only be evaluated according to its stated error, but also to the accuracy and precision of the raw data (measurements) from which they are derived.
Using a variety of traditional and novel approaches, we discuss here the results of statistical analyses assessing the validity of the raw data that is being used to formulate Australian forensic standards from 3D osseous landmarks acquired in CT-reconstructed bones. We then outline how 3D multivariate descriptors of size and shape can be used to estimate sex with a high degree of expected accuracy. We demonstrate the effect on classification accuracy when non-population specific vs. population-specific sexing standards are applied to an Australian sample. Assessing the relative magnitude of the different sources of errors is fundamental to providing robust and accurate data and results in the forensic sciences. This is especially true in Australia, where there is a general lack of population-specific morphometric skeletal standards for biological profiling, and forensic scientists therefore have little recourse but to apply standards from non-Australian populations that fail accurate representation of the modern Australian population, and thus introduce unacceptably high levels of error.
Supported by Australian Research Council Grant DP1092538