Bias and Archaeological Classification

Abstract

Archaeologists increasingly have become aware of the effects of bias and have made strides to identify and correct for error introduced in such areas as sampling and recovery techniques. Much less attention has been paid to the significance of bias introduced during artifact analysis. The potential for analyst-induced error is discussed in terms of: (1) the explicitness of class definitions, (2) differences in perception among analysts, and (3) changes in a single analyst's perception over time. Using a regression-based approach, sources of possible analytic error are detected in an archaeological data set recovered from Steens Mountain, Oregon.