Analytical techniques and software for the study of intragroup metric variation using principal component analysis

Abstract

Intragroup variation in human cranial samples is much less well understood than intergroup variation. The aims of this study were to develop a flexible and assumption-free approach for detailed explorations and comparisons of intragroup metric variation in any number of samples and to create user-friendly software for these purposes. We revisited the classic study design based on a comparison between the samples from Berg, Zalavar and Oslo from the W.W. Howells craniometric data set. Fourteen mid-facial dimensions were chosen for the analysis. "WorldPCA" software was employed for most of the analyses. This software implements a number of analytical functions aimed at exploring the results of principal component analysis. Our results confirm that the male sample from Berg displays a higher degree of variation. The cluster analyses have shown that intragroup variation in the three samples is mainly of a continuous nature. Arguably, the tendency to separate into distinct clusters is more pronounced in the samples from Oslo and Berg than from Zalavar. Some male individuals from Zalavar display distinct craniofacial features similar to those found in a South Siberian sample. Potential applications of these techniques and software are not restricted to cranial measurements but can be used for exploring any type of continuously varying data. No assumptions about the nature of the data should be made, and any number of samples can be compared simultaneously.

Keywords

• intragroup variation
• principal component analysis
• Howells data set