Stojanowski C M
Department of Anthropology, University of New Mexico, Albuquerque 87131, USA.
Am J Phys Anthropol. 1999 Jun;109(2):245-52. doi: 10.1002/(SICI)1096-8644(199906)109:2<245::AID-AJPA8>3.0.CO;2-T.
The use of metacarpal dimensions for determining skeletal sex is investigated. Previous studies on sexual dimorphism in the metacarpals (Scheuer and Elkington [1993]; J. Forensic Sci. 38:769-778; Falsetti [1995]; J. Forensic Sci. 40:774-776; Smith [1996]; J. Forensic Sci. 41:469-477) used multiple variables, which limits the application potential of the methodology. Using six measurements for each metacarpal, I generated 35 linear discriminant functions based on expected taphonomic or pathological preservation scenarios. The number of variables per function ranged from 2-5. Normal, jackknifed, and cross-validation classification matrices indicated a sex prediction accuracy in the 79-85% range. MC4 produced the most consistent functions. Overall accuracy in the validation samples ranged from 75-90%. ANOVA and MANOVA analyses indicated that population effects are insignificant, which may allow for the application of the functions without knowledge of the ancestral background of the individual. This, combined with the variety of preservation scenarios considered, provides accurate sex estimators for incomplete individuals. However, the population specificity of the insignificant population group effects remains untested.
