Network Analysis

Among the more fundamental problems in archaeological and scientific inquiry is one of study identification; namely, is a study confirmatory or exploratory (Tukey 1980; Jaeger and Halliday 1998)? The answer will depend on whether the study will test a hypothesis (confirmatory), or whether it will focus upon generating hypotheses (exploratory). Both approaches are important for scientific inquiry and discovery; however, different approaches and workflows should be used for studies that aim to generate hypotheses and those that aim to test hypotheses. Confirmatory analyses are determined in advance (often preregistered) of the study and harness the diagnostic value of statistical inference (Wagenmakers et al. 2012), while exploratory analyses are reflexive to your observation(s) of the data and provide an opportunity for the discovery of unexpected outcomes (Goeman and Solari 2011; Muliak 1985). It is important to identify and explicitly state which approach your study will pursue, since each step in the research cycle will build upon that decision (MacKay and Oldford 2000).

Developmental workflow for exploratory analyses used in my research.

While I have used networks in the past to illustrate co-presence of artifact types between sites (and populations) based largely upon similarity indices, I am also using networks to explore epistemological issues that articulate with a number of my projects. These models are employed in exploratory analyses as a method of generating hypotheses that are used and tested in subsequent confirmatory analyses.

Weighted co-presence network for Caddo ceramic types in East Texas.

Adopting methods from scientometrics, bibliometrics and machine learning has helped me to mine the literature associated with geometric morphometrics as well as predictive modeling in archaeology. These methods aid in crafting a deeper understanding of how the various arguments in each research domain are structured, helping to identify resources that may be useful in this research program. The bipartite citation networks are filtered by a degree of two, and a modularity algorithm is used to identify communities of practice within the network by identifying those nodes that have more in common with each other than with the rest of the network.

Directed bipartite citation network for predictive modeling in archaeology.

Further still, network analysis is being used to take a closer look at the variable legal challenges that employ laws associated with cultural resources and heritage management. This has proven to be a useful tool in identifying trends associated with variable legal responses through time.

Directed bipartite citation network for legal challenges associated with the Native American Graves Protection and Repatriation Act.

The contribution of network analysis to the study of archaeology is substantial, although the full scope of applications remains unclear. Within the context of those research domains mentioned above, it offers a substantive contribution that challenges a number of our ideas, while reinforcing others.

“That is what learning is. You suddenly understand something you’ve understood all your life, but in a new way.” –Doris Lessing (Nobel Laureate)