Bayesian Exponential Random Graph Modeling of an Iron Age Burial Network in Northeastern Taiwan

Author(s): Li-Ying Wang; Ben Marwick

Year: 2021

Summary

This is an abstract from the "Bayesian Archaeology" session, at the 86th annual meeting of the Society for American Archaeology.

Burials provide valuable information to study social structures and discuss social inequality. The relationship between prestige goods among burials may reflect the social relations between individuals, since prestige goods usually relate to social practices of trade, exchange, and gifting. We ask whether European colonial activities in seventeenth-century Taiwan resulted in the emergence of social inequality in an indigenous society. We use social network analysis (SNA) where burials represent actors (nodes in the network) linked by sharing the same prestige goods. Do the observed burial data indicate a more clustered network than a distribution of random networks with similar qualities? Exponential random graph models (ERGMs) are an important family of statistical models for analyzing network data and evaluating models of network structure. However, ERGMs are difficult to compute because their normalizing constant, which depends on model parameters, is intractable. A Bayesian framework allows for parameter inference using MCMC strategies that avoid the need for computationally intensive calculations of the normalizing constants. We use Bayesian SNA to study burials from Kiwulan, an Iron Age site in northeast Taiwan. This study helps to expand the use of burials in understanding the indirect effects of a colonial presence on indigenous groups.

Cite this Record

Bayesian Exponential Random Graph Modeling of an Iron Age Burial Network in Northeastern Taiwan. Li-Ying Wang, Ben Marwick. Presented at The 86th Annual Meeting of the Society for American Archaeology. 2021 ( tDAR id: 467156)

This Resource is Part of the Following Collections

Spatial Coverage

min long: 70.4; min lat: 17.141 ; max long: 146.514; max lat: 53.956 ;

Record Identifiers

Abstract Id(s): 32896