Archaeology is plagued by a phenomenon us experts call ‘shit data’. We can rarely be certain that the collection of materials available to us is a representative sample. This has enormous implications when creating archaeological networks from these datasets. Can we trust the network results we calculate? How much would the centrality of nodes change if we assume 10,20, or 90% of our data is unreliable?
Matt Peeples prepared a fantastic tutorial that introduces statistical and network science techniques for performing such sensitivity analyses. You can access it on the resources tab of this blog or on Matt’s website.