The statistical method of nearest neighbor analysis is presented for the study of distributional patterns of artifacts over occupation floors. It is compared with the previously presented method of dimensional analysis of variance. Nearest neighbor analysis is found to be much more sensitive in its detection of non-random spatial clustering. It has the advantage of not being particularly limited in application by problems of size or shape of the area under study, although it does require coordinates for each artifact and cannot be applied when only counts per grid unit are known. On the other hand, nearest neighbor analysis encounters considerable problems in defining the artifact clusters on an area and in comparing the distributions of several artifact types. These problems severely limit the utility of nearest neighbor analysis at the moment. Dimensional analysis of variance handles them better.
The standard typology of Owasco pottery from New York State is examined. This typology is ostensibly based on Krieger's definition of a type as a consistently recurring combination of attributes. The application of statistical techniques suggested by Spaulding, Sackett, and others as appropriate to the discovery of such types revealed that the standard Owasco typology does not exist in Krieger's terms. This typology, however, has been a reliable spatial and temporal indicator for many years. The real, underlying logic and foundations for this typology were therefore sought through a careful inspection of the actual type definitions. This made it clear that the definition and recognition of these pottery types depend on 2 important but unstated principles, the existence of a hierarchy of importance among attributes and the shifting of defining criteria from type to type. Krieger's definition of a type is thus seen to be inappropriate here and a type is recognized to be defined by proceeding through a hierarchical, "tree-type" series of decisions. Statistical procedures appropriate to discovering types according to this definition are developed, and their application to the Owasco material is seen to come close to duplicating the original typology. Such a tree-type structure based on the 2 principles just outlined closely resembles cultural classifications elicited by ethnologists. It is suggested therefore that the attribute association or "Spaulding" approach and the currently popular techniques of numerical taxonomy will not work for establishing space-time reflective types, not because the techniques are not well developed, but because the basic concept or definition of such a type on which they are based is inappropriate.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.