Reasoning with Zooarchaeological Counting Units and Statistics

Abstract

When zooarchaeologists work with aggregate data derived from archaeofaunas to study past human behavior, they are advised to assess whether patterns in their datasets could be products of their own decisions about quantification rather than choices of past humans. Chapter 18 revisits common zooarchaeological counting units in the context of probability theory and statistical tests based on it. It outlines the nature of zooarchaeological variables and the appropriate application of parametric and nonparametric statistical tests to them. It describes potential problems with each of the counting units described in Chap. 10 and notes methods for checking for whether these are a problem within one’s datasets, a topic that will be visited in Chap. 22. It discusses the vexing issue of which measure is best for estimating element and taxonomic abundance. An experimental blind test with large datasets suggests that NISP does not perform so well as other counting units such as MNE as a measure of element and taxonomic abundance, however, the latter is liable to aggregation effects. Chapter 18 briefly reviews a recently proposed alternative to these measures, introduced an alternative measure of element abundance: Number of Distinct Elements (NDE), a landmark-based method for skeletal element quantification that is argued to transcend many of the problems of other measures.