When many people (network researchers included) think about networks, the first thing that pops into their head is the classic network node-link diagram. In its simplest form a network graph is just a collection of points on a page representing entities of some sort with lines drawn to indicate the connections among those entities. Network visuals can be small and include only a few actors and relations where structural patterns and positions can be clearly observed. They can also be dizzyingly complex bundles of thousands, tens of thousands, or more entities and connections where general textures of relations and topological features might be visible but the positions of most nodes and edges are obscured by complexity. In either case, such visuals can paint a fascinating picture of a dataset and help a researcher recognize, interpret, and explain patterns in all manner of relational data that would otherwise be difficult to identify or communicate even with the myriad of network metrics available.

How dense is the network? What are the most centrally positioned nodes in the network? How long on average is a path through the network? Are there any cliques or communities in the network and which nodes are included? How does my network compare to other similarly defined networks in terms of local or global properties? These are the kinds of questions exploratory network methods can be used to answer.

In Chapter 1, we defined a network representation as a formal abstraction created for the purposes of visualization or analysis. Such network representations are created using network data. In this chapter, we define network data, its diverse types and data formats, and we give a wide range of examples of how it can be used to represent abstractions of archaeological data and relational theories. We conclude this chapter with best practice guidelines for how to go about collecting, documenting, storing, and sharing your network data.

In this chapter we take a step back from our in-depth methodological overview to describe what we see as some of the possible future trajectories of productive and critical network research in archaeology. Networks are already beginning to help us address a broad range of archaeological questions, as we have seen in this book (see discussion in Chapter 2), and networks are certainly useful tools and analytical constructs for many common archaeological tasks. We see the increased importance of network methods in archaeology as a trend that is likely to continue. From this vantage point, we ask: What are the profitable next steps that might push network thinking and archaeological network research to the next level? How can network methods and theories help us toward new answers for old archaeological questions or even toward questions we have not yet considered? Can archaeologists contribute to the world of network science beyond archaeology? We believe that network science has transformative potential for archaeological research and that archaeologists can be important players in network science in general, if and only if explicitly formulated relational theories drive network research in the field going forward.

Are my data good enough to create an archaeological network? What if I am missing some sites or contexts, or I have poor or variable quality information for some observations? Can I still apply network methods and models with incomplete and/or imperfect data, or should I not even attempt to use network methods? At this point, some of you may be asking yourselves questions along these lines. It is good to carefully consider potential data issues when conducting any archaeological analysis, but there are also some specific concerns revolving around sampling and data quality that deserve special attention when dealing with network data. In this chapter, we outline some of the most common issues you will encounter and further offer a generalized approach to identifying and assessing the potential impacts of sources of variation and uncertainty in network data through simulation and resampling.