EISA 2022 Workshop
Authors: Liam McVay and Josh Walmsley
What is at stake in practices of data visualisation in the social sciences, and what lessons can be drawn from their historic and contemporary use? What are the implications of different techniques and technologies of “data viz” for the knowledge claims advanced in a discipline such as International Relations, in which aspects of data visualisation may also form part of our objects of analysis? And how to go about engaging and critiquing data visualisation – in its discursive, practical, and technological dimensions – and to do this as a collective endeavour?
These were some of the core questions addressed on August 31st, 2022, at the workshop Enacting and disrupting Transnational Spaces: Data visualisation as Exploratory Method. The workshop was convened by PhD researchers Josh Walmsley and Liam McVay (both King’s College London) at The Panteion University in Athens, as part of the European International Studies Association’s (EISA) annual Pan-European Conference. Funded by the EISA’s Early Career Development Group (ECD), the workshop brought together PhD researchers and junior scholars from the UK, Turkey, the Netherlands, Switzerland, and Italy to present and discuss diverse approaches to data visualisation. The expertise of the participants spanned international relations, international political economy, international political sociology, science and technology studies, design and digital art, and computer programming, guided by senior discussant Professor Anna Leander.
The focus on data visualisation emerged out of discussions that we hosted during two previous workshops held in April and May 2022. The first unpacked “Visual Network Analysis and Transdisciplinary Approaches to Digital Methods” with Professor Tommaso Venturini and Dr Alex Meunier, whereas the second on “Identifying Hidden Patterns in Your Data” explored Geometric Data Analysis with Professor Frédéric Lebaron, Dr Sarah Perret, and Dr Médéric Martin-Mazé.
While the logic of unmasking the hitherto unseen had proven a particularly effective strategy for drumming up interest in our workshop – after all, what sort of self-respecting social scientist wouldn’t want to identify hidden patterns in their data? – we came to reflect on this revelatory promise and its free circulation between different sites of knowledge production. A trip to a museum exhibition on data visualisation brought this home, with credits for the exhibition including whole host of stakeholders, from academics to creative and design agencies, from state security institutions (GCHQ; National Cyber Security Centre) to private security firms (BAE Systems; Check Point), from NASA to tech companies (Strava), and from “Data Visualisation Consultants” to “Open Source Investigators.” We began to think with Orit Halpern’s account of how, since the late 1800s, the notion of data visualisation “slowly mutated from the description of human psychological processes to the larger terrain of rendering practices by machines, scientific instrumentation, and numeric measures,” underscoring a technologically-mediated capacity to “make new relationships appear and produce new objects and spaces for action and speculation” (2014, p. 21). We noticed then that this slippery promise appeared to co-exist in our attempts to justify our methodological choices and the discourses of the types of actors we, as researchers of security politics, might take as our objects of investigation and critique. This prompted one reflection that inspired the workshop: how we might go about harnessing the promise of data visualisation in our research practice without reproducing, at best, unrealistic claims about its revelatory capacities, or at worst, contributing to the very injustices and violent hierarchies that we purport to critique?
The workshop was organised into two sections. Part I featured a traditional paper-discussant setup, involving a collective engagement with seven short written interventions shared in advance by the participants. Part II interrogated the questions raised during the morning session through collective practices of “doing data viz”, taking three EISA ECD Workshops as a point of departure for our object of analysis to jointly produce and critique a novel set of data visualisations. The following provides a reflective account of what unfolded.
The papers approached data visualisation in different ways while also speaking productively to one another. To synthesise them here is inevitably to oversimplify of a wide array of points of engagement, but we will try to give an overall flavour. Ruben van de Ven and Cyan Bae’s work of diagramming imaginaries of computer vision in the field of security introduced a new tracing software designed to track the construction of diagrams concerning imaginaries of surveillance, facial recognition, and data processing. Similar empirical issues were explored by Pedro Maia, who elaborated on a methodological approach for exploring the socio-technical characteristics of the use of Geographical Information Systems interfaces in national security infrastructure. The question of technology was likewise present in Liam McVay’s work, which simultaneously reflected on the question of data visualisation as method by reference to his use of multiple correspondence analysis for mapping the symbolic structure of a social space of intelligence professionals. Likewise, Dr Alvina Hoffmann explored the role of data visualisation in the method of elite prosopography, considering the contribution of visual network analysis (VNA) to her analysis of the development of a transnational social space of UN Special Rapporteurs. Along a similar line, Vanessa Ugolini considered what VNA and associated tools (i.e., Gephi) can bring to the mapping various interrelated forms of data flows, reiterating the centrality of caution and reflexivity in the integration of data visualisation into our methodological frameworks. Dr Esra Elif Nartok’s auto-ethnographical reflections on her PHD research explored some similar themes, discussing the relations among forms of knowledge production, visualisation, and corporate logics in her analysis of the World Hindu Economic Forum. And the politics and epistemology of “data viz” were front and centre in Dr Kiran Phull’s historical research into the work of W.E.B Du Bois, exploring the emancipatory potential, and world-making power, of data visualisation for re-humanising data as a tool against oppression.
Together, the papers and discussions articulated several core questions around the pitfalls, traps, risks, but also the potential rewards of a critical engagement with visual methods. For example: what are the politics of doing data visualisation? Who is visible, who is visualisable or visualised, and what is the role of the researcher in navigating this? How do socio-material constraints on academic practice influence our orientation towards data visualisation, our use of it, and the claims we make in its name?
Equally, what do alternative understandings of social and political context (i.e., notions of “social space”), of objectivity and the production of “new” knowledge, imply for practices of data curation and visualisation? And how might we use data visualisation to critique the developing techno-solutionist discourse in which it is often embedded, such as the predictive and anticipatory power of algorithmically generated visualisations?
In the second session, participants were involved in a more “hands-on” endeavour. The task was to apply the reflections of the morning’s session in a collective production of a set of novel data visualisations that took as our research object the “social space” of the three Early Career Workshop funded at EISA PEC 2022.
First, in smaller groups, participants started with a fundamental question: what is at stake in producing a visualisation of the social space of these workshops? How can we delineate this research object? What kind of ontological framework would we use – why even start with a spatial understanding at all? What are the implications of choices around methods and technology in data construction? What were the trade-offs that accompanied these different approaches? Why visualise in the first place? After much deliberation, and a well-earned coffee break, the groups identified three distinct angles.
Tracing workshop discourses over time
The first group sought to map the hierarchies of different concepts by examining the way in which participants appropriated terms and concepts in the day’s discussions. They settled on a kind of improvised temporal discourse analysis in which the history of the workshop was represented by recording the use of each term or concept, who was using it, and when it was used. The results of this quantification were then projected onto a “timeline” that yielded insights into disparities, denoted by differences in the occurrence and order of a terms appearance, among the competing conceptions of data visualisation represented at the workshop.
Visualising workshop relations through “slow data”
The second group took inspiration from the work of W.E.B Du Bois and the “slow data” movement, opting to develop analogue diagrams that challenged formulaic, traditional infographics and network visualisations. Rather than representing variations in the population in statistical charts, the group discussed ways convey multiple data categories simultaneously in the same image. These designs represented relations between actors, different symbolic and material hierarchies, and methodological frameworks, while also eliciting a critical engagement with broader processes of knowledge production often implicit in contemporary methods of data mapping.
Map of socio-professional hierarchies using MCA
The final group sought to analyse the social structures and professional hierarchies that influenced the workshop’s construction and the day’s discussions. In line with more traditional forms of data visualisation, this approach began by constructing a data set around workshop participant’s “traits”, their academic credentials, professional histories, and areas of expertise. Subsequently, there was a debate around the authority of these different variables, and how their inherent fallibility could be addressed or offset. Thereafter, by using data processing software “RStudio” and “Gephi”, the revised data was used to generate two visuals – a multiple correspondence analysis and a visual network analysis, each revealing interesting connections between participants, institutions, and practices, and between different participant “data profiles”.
Doing and critiquing data viz as a humble, collective endeavour
The main takeaways? In an age where the alluring promise of data visualisation circulates freely between the academic field, the arts, state bureaucracies, and private enterprise, the workshop provided an opportunity to interrogate, practically, what it means to do data viz in a cross-disciplinary context. Some participants were encouraged to try alternative forms of visualisation, others were required to reflect more closely on what visualisation actually brought to their work – with one dropping it altogether. As well as providing many points of inspiration for developing our research individually, the workshop underscored that if data visualisation is to live up to its heuristic and emancipatory promise for a discipline such as IR, this can only be realised through a humble, reflexive, and collective engagement with both its potential and pitfalls.
About the authors
Liam McVay is a PhD researcher in International Relations at King’s College London’s Department of War Studies. His current research explores themes of intelligence and policing, history and immigration.
Josh Walmsley is a PhD researcher in International Relations at King’s College London’s Department of War Studies. His current research explores the transnational politics of counter-radicalisation, inquiring in particular into struggles over expert authority and the role of technology.