Can We Predict Archaeological Site Location? Should We?

Summary

This is an abstract from the "*SE Big Data and Bigger Questions: Papers in Honor of David G. Anderson" session, at the 89th annual meeting of the Society for American Archaeology.

Archaeological predictive models, whether formal or informal, are commonly used on compliance-driven projects, but their efficacy is rarely tested. Too often, we assume that models are “good” or “successful” when more sites are discovered in “high-probability” than in “low-probability” zones. In Florida, state guidelines for cultural resource management professionals mandate the use of a predictive model and stipulate specific testing requirements for each probability zone. Dave Anderson pioneered the use of site file data to answer “Big Picture” questions and his work at Fort Polk set a high bar for developing multiscalar predictive models from broad survey data. As an MA student, Dave taught me not to shy away from large, messy datasets, but always to ask pointed questions and be wary of biases and logical fallacies. In other words, Dave taught me “how to think.” In this paper, I draw inspiration from Dave to test the efficacy of the Florida predictive model using data from the Florida Master Site File and other state-wide datasets. In doing so, I question the efficacy and appropriateness of predictive modeling as a regulatory tool and provide an alternative framework for consideration.

This Resource is Part of the Following Collections

• *SE Big Data and Bigger Questions: Papers in Honor of David G. Anderson   •
• Society for American Archaeology 89th Annual Meeting, New Orleans, LA (2024)

Keywords

Spatial Coverage

Record Identifiers

Abstract Id(s): 41537.0