Partitioning chi-square for the analysis of frequency table data: an archaeological application

Archaeologists often wish to compare observed frequency distributions with expectations generated by a model. We describe a technique of partitioning chisquare which yields information about goodness of fit to a model and about homogeneity among populations simultaneously, and which is often superior to other commonly used methods of evaluating frequency data. The technique is demonstrated with data from the Upper Paleolithic rockshelter Le Flageolet I. The results suggest selective raw material use by the Aurignacian occupants.