Should (large) datasets be used to study and anticipate wars? Interview with Louise Beaumais, Iris Lambert, and Frédéric Ramel
In an era where data drives decision-making across countless domains, how do numbers shape our understanding of war and conflict? This question lies at the heart of Quantifying International Conflicts: Data on War or Data for War, a collective1 work that challenges conventional approaches to conflict analysis in the digital age. The book emerges from the DATAWAR research project, which examined how quantitative databases on armed conflicts influence both academic research and professional practice. The project's findings reveal a striking paradox: despite the growing availability of conflict data, practitioners—from diplomats and military officials to journalists and humanitarian workers—often remain sceptical of purely quantitative approaches to understanding war.
In this interview, co-editors Fréderic Ramel, Louise Beaumais, and Iris Lambert explore how methodological choices carry profound political implications, and advocate for a more reflexive approach that acknowledges both the possibilities and limitations of data-driven conflict studies.
In order to introduce this interview, can you briefly describe the aims and objectives of the DATAWAR project of which this book is a result, as well as the methods used?
At a time when the use of numbers in contemporary public policy is becoming increasingly systematic, the project aimed to analyse the influence of quantitative data in the strategic field, both in the production of scientific knowledge and in professional practice (diplomats, soldiers, humanitarians, journalists). The study focused on a specific tool that has been little studied in the academic literature: the production and use of databases on armed conflicts. It was structured around three axes: the creation and use of these databases by social scientists; their use by practitioners in France, Great Britain, and Germany, and the representations they induce for understanding conflicts; and reflection on the conditions and ways of disseminating thinking on quantification to civilian and military decision-makers. Influence is not defined as a causal link between a type of scholarly discourse and a specific type of perception, but through the prism of how the results of quantitative conflict analysis are received, filtered, and interpreted. These three objectives are reflected in the structure of the book, which is divided into three parts. The first, “Academic Practices of Conflict Data Collection and Analysis”, reviews how academics provide alarmist biases as well as a tendency to neglect other explanatory factors such as norms, or even emotions when using this kind of databases. They express a positivist alarmism that occurs with a certain regularity, to the detriment of rarer phenomena (such as nuclear accidents or covid). The second, “Practitioner Perceptions and Use Patterns”, examines the uses of this data by practitioners in a comparative manner. It highlights the different functions that this data serves in describing, diagnosing, or even reinforcing the rhetoric of agents in the production of an argument. The last part, “Dissemination Challenges and Lessons”, identifies ways to teach and promote a more reflective use of this quantitative data. Richard Ned Lebow has done us the honour of reading the full manuscript in order to put it into perspective in a concluding chapter.
Each part of the project relied on specific methods. Since COVID-19 made it impossible to immerse the team’s researchers in the laboratories that make up the databases, particularly in the United States, the first axis changed its methodology. It is based on the use of descriptive statistics to report on the links between the use of databases and the more or less alarmist interpretation of the outbreak of armed conflicts in several scientific journals representative of the field of strategic, security, and peace studies. The second axis used several methods: QDR miner software applied to a corpus of press articles compiled by the team; 70 semi-directive interviews with practitioners (diplomats, military, humanitarian, journalists) in the three selected countries; and administration of an online questionnaire in September 2021 with a class of auditors from the Centre des Hautes Etudes Militaires in Paris. The third axis is based on participatory seminars with civilian and military practitioners. These focus groups, led by members of the team, were enriched by role-playing exercises with students from Sciences Po Lille to test a simulation of armed conflict using databases.
Based on the DATAWAR research programme, this book seeks to explore the question of how social practices of data collection and analysis in quantitative conflict studies influence researchers’ and practitioners’ representations of armed conflict. Can you give us some insights about which practitioners are concerned, and why?
In recent years, we have witnessed a growing quantification of war-related dynamics—whether in terms of casualties, military expenditures, displacement figures, or other indicators. As databases documenting these phenomena become increasingly diverse and widely disseminated, one of our initial hypotheses was that their use by practitioners would also expand. This assumption appeared to be corroborated by the discourse of database producers themselves, who frequently emphasise that these resources are intended to inform practitioners’ decisions or to enable them to track specific trends. This prompted us to ask a series of questions: do practitioners in fact make use of these datasets? If so, how? And if not, why not? More specifically, our investigation focused on three groups of actors: political and military decision-makers, journalists, and NGO professionals.
The first step of our analysis was to determine whether practitioners drew on armed conflict datasets in the course of their professional activities—whether in journalistic reporting, NGO annual reports, or policy documents such as thematic studies, geographical reports, and doctrinal papers published on government websites. One striking finding was that, although political institutions, NGOs, and the press do occasionally rely on insights from quantitative studies to forecast conflict and shape their analytical or normative positions, these datasets are not mobilised as frequently as one might expect. In the fields of defence and foreign policy, quantitative methods—such as conflict databases or early warning systems—are often regarded with scepticism, viewed less as tools for genuine decision-making than as managerial instruments. In many cases, such data are employed more instrumentally, primarily to reinforce expertise grounded in qualitative analysis. Journalists, for their part, rarely incorporate these datasets into conflict reporting, hindered by constraints such as limited training and time pressures. Nevertheless, newsrooms often encourage recourse to figures, recognising their cognitive and rhetorical impact in engaging readers. Humanitarian workers also tend to rely more on numerical data to communicate with their audiences but also, and more crucially, to secure funding from donors.
This reluctance to fully embrace quantitative resources can be explained in part by sociological variables, such as training deficiencies or professional cultures that are more or less receptive to statistical approaches. Political factors also play a role: diplomatic networks often provide decision-makers with information perceived as more directly relevant to conflict analysis, while conceptions of the “national interest” remain tied to strategic and digital dimensions beyond the scope of quantitative indicators. Yet despite the shared scepticism observed across these professions, numerical data nonetheless circulate among them. This circulation, however, is not without risk: it can lead to distorted representations of conflicts and, consequently, to misallocation of policies and resources.
Is there a way to make practitioners aware of the risks and limitations of the use of large data sets?
Given the limited and often instrumental use of conflict databases by practitioners—together with their explicit distrust of numerical evidence—it remains challenging to articulate a precise and comprehensive framework for raising awareness of the risks and limitations associated with large datasets. Nevertheless, we have sought to explore several avenues that may help to complicate and enrich our understanding of these dynamics.
Upstream, at the stage of data production, dialogue with database producers themselves appears crucial. Such exchanges not only allow producers to better grasp how their data are actually mobilised by practitioners, but also facilitate greater transparency regarding the methodologies on which these datasets rest. Downstream, at the level of data use, we call for heightened reflexivity regarding the often-overlooked biases and unintended consequences that accompany data-driven and algorithmic analyses of human behaviour. The aim here is to make visible—and more readily accessible—an epistemological stance that explicitly acknowledges the positivist assumptions underpinning data production. In doing so, we have sought to document and communicate the implications of dataset use for the representation of conflicts and the political responses formulated by decision-makers and NGOs.
To this end, and drawing on our experiences with disseminating the “datawar” initiative, we developed five pedagogical principles designed to encourage practitioners not to reject quantitative data outright, but rather to engage with them in a way that foregrounds their constructed nature. This is particularly relevant in the training of political science and international relations students, who often constitute the future cohorts of practitioners.
The first principle is that a pedagogy of conflict data should be pluralist: it should present a spectrum of epistemological perspectives on data, without prescribing a single interpretative lens from the outset. Second, it is important to highlight the symbiotic relationship between theoretical assumptions, the design of datasets, and their modes of visual representation. Third, data should be promoted not as objective or uncontested reflections of reality, but as tools for deliberation and normative debate. Fourth, pedagogy should encourage students to experiment with improvised or ad hoc uses of data, demonstrating that—and this is the fifth principle—meaningful insights into conflict dynamics can emerge without advanced training in statistics or specialized software. Overall, emphasis should be placed on reflexive engagement—acknowledging both the potentials and limitations of quantitative approaches—so that data are approached critically, yet without dismissing their possible contributions to understanding and addressing conflict.
There are numerous and frequent debates opposing quantitative and qualitative research methods, in general. Why does choosing either method have political consequences when it comes to studying conflicts and war?
There are two ways of answering this question.
The first is that this choice has political consequences because methods are not just technical choices, they reflect and reinforce different assumptions and representations of what war is and how we think about war. While our objective was not to revive the so-called Second Great Debate in International Relations between quantitative and qualitative approaches, many of our arguments resonate with those put forward by qualitative scholars. After all, one raison d’être of the programme was to challenge the adequacy of quantitative approaches and to highlight their limitations in fully capturing the complexities of war.
In the book, we show that the preference for quantitative methods— which aim to identify patterns, correlations, and causal relationships through statistical tools—has, at least, two main political implications. First, as Thomas Lindemann argues, this approach introduces a bias toward a particular kind of interpretation—namely, alarmism. Second, as Louise Beaumais points out, it contributes to framing war as something rational and manageable that can be “solved”. As Mathias Delori has shown in his book Ce que vaut une vie , policies shaped by technical models lead to cold, utilitarian decisions with their very own vocabulary (“acceptable losses”, “surgical strikes”, or “collateral damage”) that further contribute to de-humanising war.2
Qualitative methods, on the other hand, prioritise lived experiences through in-depth, local studies. While often dismissed as “anecdotical”, they help us understand not just what causes war, but also how it is experienced and remembered. This, in turn, emphasises factors such as identity or emotions, in ways that challenge dominant, rational, state-centric conceptions of international politics.3
The second way to answer this question is to put forward this choice as already political: as the book shows, methodological choices are also tied to funding sources and institutional incentives. Even actors who distrust or criticise the reductionism of quantitative approaches (diplomats, militaries, humanitarian workers, or journalists in our book, but the observation definitely also works for scholars) often feel compelled to adopt them. This is not necessarily because they believe in the epistemological superiority of quantitative approaches, but because they feel that numbers are required. It often stems from structural pressures (such as funding requirements, institutional preferences, or the desire for policy impact) which reward what is measurable over other forms of knowledge.4
Data production is never neutral, especially on an international scale. Does it contribute to the so-called acceleration of the world, to use the term coined by Harmut Rosa? Can you expand on this?
The quantification of the world—whether in terms of the production of numerical data on the one hand, or their mobilisation as a means of accounting for reality on the other—constitutes a fundamental trend of modernity. The explosion of digital data, much of which also takes the form of numbers, contributes to this quantification. Acceleration is tied to quantification: it is its very engine, since, as Rosa points out, modern societies find their stability in dynamic growth, in the drive to bring the world within our reach, to extend our access to it, and to exploit its resources. The production of data stems from this will to control. One of the aims of this book is not to explore in depth the relationship between quantification and acceleration, but rather to invite a reflexive gaze on the ways in which we use quantitative data. Such data are insufficient for decoding actors’ behaviour, especially when it comes to clarifying the strategic orientations of great powers. For instance, observing the evolution of military expenditures is a first step in analysis. But inferring from this evolution repertoires of action, reliable scenarios, or anticipations of the nature of conflict tends to obscure other explanatory factors.
In the media sphere, our research highlights a clear relationship between the accelerationist trend and problematic journalistic practices in the use of numbers. In a context where newspapers compete with the abundance of freely accessible news, the added value of a story is increasingly tied to its clickbait potential. Audiences tend to value speed above all, as well as originality of a story. Both speed and originality, however, hinge on how rapidly a newspaper can access, process, and publish a story. In this competitive environment, numbers play a significant role in amplifying the appeal of stories. For journalists, the rush to publish generates two recurrent problems: approximations in numerical reporting and an overreliance on a narrow circle of trusted sources.
You write that a number of recent theoretical shifts in the study of conflict, such as the “pragmatic turn” or the even more recent “emotional turn” have not really been taken up by quantitative conflict analysis compared to traditional, “objective” considerations, such as the distribution of military power or the scarcity of economic or ecological resources. Why is that?
This reluctance to take sensitive dimensions into account stems from a broader trend within classical strategic studies, which today appear to be experiencing renewed interest in an international system marked by a sharp increase in military spending and the intensification of great power rivalries. The emphasis placed on measurable power capabilities—numbers, and nothing but numbers—has shaped an entire body of conflict literature. In this book, we seek to rethink our relationship to numbers in the study of armed conflicts. Such an undertaking is not unrelated to certain so-called realist approaches, such as that of Morgenthau, who emphasised the non-material and non-quantifiable foundations of power, for example the quality of diplomacy. Demonstrating that behind the use of numbers lie representations that are not necessarily made explicit or objectified seems to us essential, both for scholars and for practitioners.
Finally, can you reflect on Ned Lebow’s comment in the conclusion of the volume: “On the whole, quantitative studies need to connect themselves to more sophisticated theoretical assumptions and theories, and in the process, move away from treating states as unitary actors, recognize that underlying causes are a necessary but insufficient cause of war, acknowledge the interdependence of cases, and find ways of folding in agency?”
Ned Lebow’s comment is a valid and important observation that echoes our answer to question 4. His critique highlights a broader tension within the field, especially pronounced in the United States. What we see, in fact, is that despite many authors’ own awareness of the methodological and theoretical limitations of purely quantitative approaches when it comes to studying war and conflict, there remains a persistent proliferation of such studies (or at least of mixed-method approaches). This dynamic is partly driven by a logic of self-referentiality within the academic community. Researchers engage in quantitative methods because funding bodies and institutions demand it. Publishing in top journals, securing grants, or building a “good” academic profile increasingly hinges on demonstrating quantitative expertise. This creates a paradox: even as many scholars recognise the epistemological limits of quantitative methods, the structure of the discipline continues to incentivise their use. Once again, quantitative methods might be chosen not because they are best suited to answer the research question, but because they are legible within dominant academic and funding structures. Quite interestingly, this means that the predominance of quantitative methods should not be mistaken for a consensus on their appropriateness. Acknowledging that is already a first step in improving current quantitative studies.
Further, while not all quantitative research is designed for direct application, many studies are framed—or funded—with the explicit goal of informing policy. Yet our work shows that these tools are often underutilised in practice. Even if scholars are not fully aware of this gap, their persistence suggests that the function of quantitative research in IR is not about producing practically useful knowledge.
If the dominance of quantitative methods cannot be explained by their appropriateness, or justified by their practical utility, then what sustains their centrality? What appears to be at stake is the role that quantitative methods play in securing disciplinary legitimacy and sustaining a particular epistemic order. Producing large datasets and complex statistical analyses is still widely perceived as a way to display disciplinary capability and scientific credibility, contributing further to defining (and reinforcing) what counts as the “right knowledge” in International Relations.
Interview by Miriam Périer, CERI.
Illustrations:
- Figures on a screen, by Pawel Michalowski for Shutterstock
- Book cover
- Digital content concept, by metamorworks for Shutterstock
- 1. Editors are Louise Beaumais, Iris Lambert, Thomas Lindemann, Sami Makki, Frédéric Ramel, and Eric Sangar
- 2. See also Gibson, J. W. (2000). The perfect war: Technowar in Vietnam. Atlantic Monthly Press, as well as Bousquet, A. J. (2022). The scientific way of warfare: Order and chaos on the battlefields of modernity, Oxford University Press.
- 3. In the realm of policymaking, the first chapter of Gilles Dorronsoro's Le gouvernement transnational de l'Afghanistan: une si prévisible défaite (2021, Karthala Editions) is extremely convincing in showing the benefits of a qualitative approach over a quantitative one.
- 4. Joel Glasman has done a very good job at historicizing this process for humanitarian workers notably, see Glasman, J. (2019). Humanitarianism and the quantification of human needs: Minimal humanity. Routledge.
