Regression with Archaeological Count Data

Abstract

Archaeological data often come in the form of counts. Understanding why counts of artifacts, subsistence remains, or features vary across time and space is central to archaeological inquiry. A central statistical method to model such variation is through regression, yet despite sophisticated advances in computational approaches to archaeology, practitioners do not have a standard approach for building, validating, or interpreting the results of count regression. Drawing on advances in ecology, we outline a framework for evaluating regressions with archaeological count data that includes suggestions for model fitting, diagnostics, and interpreting results. We hope these suggestions provide a foundation for advancing regression with archaeological count data to further our understanding of the past.