ABSTRACT
This chapter examines strategies for testing the strength and significance of statistical relationships in large multivariate datasets. In addition, it discusses the logical problem of multicollinearity in archaeological data analysis, or in other words, how to determine causality in which multiple variables may interact with one another in complex ways. Techniques such as multiple regression and partial correlation are used to identify relationships between several different variables and to mathematically hold constant the effects of particular variables. This chapter also considers the assumptions of these models, and it discusses the use of generalized linear modeling (GLM) as an alternative.
