Interpretation and Testing in Multivariate Statistical Approaches to Physical Anthropology: The Example of Sexual Dimorphism in the Primates

Abstract

During consideration of general problems of testing and interpretation of multivariate statistical methods in physical anthropology, the question of bias due to pooling of sex subgroups has been an issue. Though, as expected, recent tests have confirmed the validity of such pooling, they have also revealed unexpected information about structural sexual dimorphism itself. Thus, canonical variate analysis of primate overall bodily proportions shows that sexual differences are most marked in proportions representing breadths. In conjunction with high-dimensional displays, they further demonstrate that a number of different structural sexual dimorphisms exist rather widely in the entire Order. This finding has implications for our understanding of the causal mechanisms of structural sexual dimorphism in living species, for our attempts to assess sexual differences in fossils, and, by extension, for our understanding of change in non-structural sexual dimorphisms in humans.