Analytics

A recent study illuminated the bias toward publishing significantly positive results by researchers in the social sciences, raising substantive questions regarding the treatment and dissemination of null or statistically non-significant data. In archaeology, we also tend towards emphasizing the latest discovery, the big site, or the conclusive analysis. While it is satisfying to be able to present the latest and greatest in one’s field, what then becomes of the rest of the data? Typically, these are relegated to summary tables, supplementary technical reports, or cursory discussions of miscellany that note how the rest of the data support or relate to the “big find”. The preferential bias towards positive significance, however, generates both analytical and interpretive self-selection biases in our archaeological understandings. Null data sets play an absolutely critical role in the inferential methods of spatial, quantitative, and archaeometric analyses. In addition, null or negative results can provide epistemic boundaries on the evaluation of interpretation and theory.

Archaeologists have likely collected, as a conservative estimate, billions of artifacts over the course of the history of fieldwork. We have classified chronologies and typologies of these, based on various formal and physical characteristics or ethno-historically known analogues, to give structure to our interpretations of the people that used them. The simple truth, nonetheless, is that we do not actually know how they were used or their intended purpose. We only make inferences—i.e., educated guesses based on the available evidence as we understand it—regarding their functions in the past and the historical behaviors they reflect. Since those inferences are so fundamental to the interpretations of archaeological materials, and the archaeological project as a whole, the way we understand materiality can significantly bias the stories we construct of the past. Recent work demonstrated seemingly contradictory evidence between attributed purpose or function versus confirmed use, however, which suggested that a basic premise of those inferences did not empirically hold to be true. In each case, the apparent contradiction was resolved by reassessing what use, purpose, and function truly mean and whether certain long-established functional categories of artifacts were in fact classifying by function. The resulting triangulation, presented here, narrows the scope on such implicit biases by addressing both empirical and conceptual aspects of artifacts. In anchoring each aspect of evaluation to an empirical body of data, we back ourselves away from our assumptions and interpretations so as to let the artifacts speak for themselves.