And if so, how? Where to start a network research project?
Questions like these guided our week at the Barcelona Past Networks Summer School 2024, where we taught a hand-selected group of 25 graduate, doctoral, and postdoctoral delegates how to tackle a formalised network research project to study the human past — at the backdrop of the beautiful scenery of Barcelona’s eclectic architecture, the gravitas of the repurposed old church of Sant Agustí (our first venue) and the simple austerity of the University of Barcelona’s Physics department class rooms. After planning this for almost 1.5 years, countless meetings, and combining our joint experience of organising events such as the DH Lab Winter School 2022, or the series of HNR Workshops of the HNR Community, it was a feast to finally see this school unfolding!
The summer school 2024 entailed a full program on the how to and why (and the how not to!) of past network research featuring inter- and transdisciplinary perspectives from archeology (Tom Brughmans) and history (Martin Grandjean, Marten Düring, and myself), as well as complexity studies, namely from physics (Albert Diaz Guilera) and urban science (Matteo Mazzamuro). Lectures and tutorials ranged from introductions on the how and why of past network research (Tom) and on formalized network analysis methodologies on micro-, meso- and macro-level (Albert), to bootstrapping the delegates from zero to creating their very first networks in R (Matteo), to experimenting with the more accessible ready-made software of Gephi (Martin), how to design your network research project and modeling networks from an epistemological point of view and critical perspectives on network analysis (Marten and myself), as well as best practices for data sharing and publications in repositories (Tom and Matteo), and assisted study time for advanced network tutorials in R. In another session we offered a reality check: diving deep into initial aspirations of four of our own research projects, and the ways how we adapted to and ultimately overcame the challenges we encountered. The opportunity of receiving individual mentoring by the attending lecturers, as well as peer-to-peer support, for example in our poster session à lieu the poster design workshop, were suited to further advance (very) specific research questions and problems, fostering the individual understanding of our summer school’s delegates.
Special mentions deserve the local organisers (shoutout to Lucce Prignano!) for the seamless integration of much-needed recuperation after mental exertion with coffee, lunch options on site, and even a welcome and farewell dinner and a city tour on Roman road networks in Barcelona.
For those not having attended the summer school: How to start network analysis?
In my opinion, it all starts with an epistemological framework – the way how we generate knowledge. There, I found the framework of modeling most helpful (as outlined in Petz 2022, p. 32): modeling is to approximate the research object through the (inter-)subjective selection of important aspects following the contention by Stachowiak 1973. A model as an approximation is necessarily a reduction of complexity — and as an epistemic tool, a model is based on ultimately subjective selection decisions, which need to become intersubjective (i.e. mutually comprehensible by anyone following your research) by making the research process transparent through documentation. This, too, follows the idea of Open Science and to systematically allow for reproducibility of research as one of the cornerstones of doing good science. (And, of course, documentation in the form of code annotations might even help you to be able to understand your code in a couple years time! Double win.)
Models can be found all around the research process:
- the research design is a model: the subjective selection decisions guided by a (set of) research question(s) thus influencing the selection of sources and methods in order to grasp the object of study best
- the data used is based on a model, i.a., operationalization of information from the source basis; its formatting; which standards, …
- historical periodisation for example is a model, visualisations are models, too, and many more!
- And: networks are models. Networks are constructed through the selection decisions to express the complexity of the research object into a definite set of entities and their relationship(s) – in other words: deciding what are my nodes, what are my edges, in which format do I need those (edgelist vs matrix)?
This needs to be followed by asking yourself: “Is it a fit?” – Does my methodological approach fit my sources, and my research question(s), do my sources fit my research questions, and vice versa? Do my questions/methods/data share the same assumptions, premises and requisites? (This requires method literacy, algorithm literacy, data literacy, …) As this is highly dependent on the domain-specific contexts of the object of research, there are no ready-made answers to these questions. There are many ways for it to go wrong, but ‘the’ right way doesn’t exist. Knowing when to do network analysis is just as important as knowing when not to! (And this of course prevents you from the easy-to-fall-in trap, when having a hammer (the method), everything starts to look like a nail.)
When networks are inevitable
Network analysis is all about relations between entities that are deemed meaningful. Sometimes, however, a network perspective might not even be needed, e.g. when a ‘mere’ frequency statistic would suffice for expressing the frequency of relations. But if the relations between entities should (or: could) explain something more than that, network analysis might become just the right approach – or rather, a step on the way of studying a topic. It is highly likely that network analysis will not provide all the answers (even though ‘networks are everywhere,’ as the famous saying goes). But as an exploratory and investigative approach it is highly valuable. And as part of a bouquet of approaches and methods, network analysis is highly likely to bring you closer to answering your research question(s). Complementation of methodologies is key.
Next steps for your network research endeavors
Mingle! Attend conferences and workshops on network research of the past, learn about other’s solutions for tricky problems, and get into contact with experienced colleagues! Maybe even join the Historical Network Research Community? Or meet up end of May at the next HNR Conference 2025 in Rio? Or learn at the next Barcelona Past Networks Summer School 2025 set to take place again end of June 2025? The respective calls will be published soon.
Experiment! Cooperate! Don’t be mistaken: there is quite a steep learning curve to master the methodologies of network research. There is a whole new world out there: network analysis approaches (and vague or metaphorical uses of networks) situated in very domain-specific contexts within a vast array of the humanities and social sciences (literary studies, historical studies, archeology, anthropology, sociology – you name it) and the formalised study of network structures and development of new (mathematical) methods otherwise known as network science (drawing from mathematics, physics, complexity studies, computer science, computational humanities, computational social science, …). In order to get the best of all worlds, you likely will want to cooperate. But in order to do so, you first have to start speaking a common language. So start learning, start doing. This is your opportunity.
Helpful resources
- Tutorials for Network Science in Archeology in R
- One of the many tutorials for network research in history at the Programming Historian, such as e.g., Network Analysis with Python
Bibliography
- Petz, Cindarella (2022). On Combining Network Research and Computational Methods on Historical Research Questions and its Implications for the Digital Humanities. Dissertation at the Technical University of Munich. http://mediatum.ub.tum.de/?id=1624881.
- Stachowiak, Herbert (1973). Allgemeine Modelltheorie. Wien, New York: Springer Verlag. https://archive.org/details/Stachowiak1973AllgemeineModelltheorie.
Featured image: Barcelona Past Summer School header from https://www.pastnetworks.net/.
OpenEdition schlägt Ihnen vor, diesen Beitrag wie folgt zu zitieren:
Cindarella Petz (12. Juli 2024). “Should I do network analysis?” – Reflections on the Barcelona Past Networks Summer School 2024. DH Lab. Abgerufen am 5. November 2024 von https://doi.org/10.58079/120ba