Ebook: Building and Using Datasets on Armed Conflicts

The so-called “hard” or “exact” sciences, with their necessary emphasis on technology and on the technical, are hardly reputed for being very human, and, conversely, the so-called “human” sciences are often pronounced as “soft” because they cannot be based on the certainties associated with the former. The search for truth which is the essential dimension of the construction of a peaceful world therefore has to navigate between considerations of a philosophical nature and the concrete data of the hard sciences. If, ever since the humanism of the Renaissance period, we have been happy to lay claim to the wisdom of one of its great writers, Rabelais, who taught a moral lesson to the young Pantagruel with the neat formula “science without conscience is the ruin of the soul”, we nonetheless stand in awe before modern scientific advances and the extraordinary achievements that they have opened up. If everything is not permissible, at least everything seems possible!

It was such an outlook which no doubt had something to do with our decision taken ten years ago here in Grenoble to give concrete shape to a utopian dream by creating in our city a “school for peace”. At that time we were reminded that a Brazilian bishop, Don Helder Camara, famous for the stands which he took in favour of human development and the promotion of peace through dialogue and a more just distribution of resources, had once remarked that Grenoble would be an ideal place to combine its advanced technological and computing research with the philosophical and humanist dimension to bring about the necessary synthesis for the promotion of the cause of peace in the world. And he had made this comment at the height of the cold war, as a way of confronting and fighting against the extremism that posed such a threat at that time. Since then, the Third World War has not happened and if one can pose the question of why the North Atlantic Treaty Organisation should be maintained, an organisation which without a doubt contributed to the prevention of such a war, we should not ignore the threats and challenges which remain, or are even greater today, in the general field of world security.

This is why we were very gratified when we realised that an Alliance, which some felt was the product of a bygone age, was reorienting its scientific programme in order to “favour research on questions linked to security and by so doing to better reflect the new environment in which NATO finds itself”. And that in order to face the complex questions associated with security, there would be “a greater interaction between the exact sciences and the human and social sciences”. By submitting a project to hold an Advanced Research Workshop on the theme of “indicators and databases for risk prevention”, we gave ourselves the chance to deepen and prolong research which we were already engaged in, based on the idea that a climate or a culture of peace needed to be encouraged and that there were effective ways in which progress could be made in this domain. We believe that this area of study can be subjected to measurement and quantification. We are now equipped with measuring instruments for natural phenomena, and in a world of increasing complexity, we are more and more concerned with questions of efficiency, of performance and evaluation in all domains, including those which are related to questions concerning the quality of life and the avoidance of social conflict. Comments on war are now inevitably accompanied by sets of “figures” concerning the loss of human lives and the “collateral damage”—all sorts of damage which come in its wake. To the sinister “body count” of open and full war, we must add the information concerning the horrors of all other forms of violence and ignored wars, if our intention is to take the real measure of human tragedies which they entail in order to awaken public conscience and encourage appropriate political action. But we must also today take into account the figures and statistics of new fields of research and action, concerning the environment and sustainable development, if there is going to be real progress on global prevention. The human development indicators of the United Nations Development Programme (UNDP) seem to us to show the way forward in this respect. Furthermore, we support the recommendation of the mathematician, historian and philosopher, Michel Serres, when he says “Since we control the whole world, we must learn to control our own control mechanisms” and we also share the belief of Martin Hirsch, president of the humanitarian organisation Emmaüs, who, just before taking up his French governmental appointment, declared on the subject of poverty indicators: “They are indispensable… To try to do without them would be like tackling road safety, without using devices for measuring and registering speed!”

We were therefore clearly aware that all sorts of research is being undertaken in this vast domain and the time had come to bring it all together and submit it to further empirical evidence. All the actors concerned, the politicians, diplomats, the military, the NGOs, as well as the heads of enterprise, need a clearer vision of the state of a world which is more and more complex. The improvement of our information systems is therefore particularly useful, both for pure research and for its applications. The sharing of means and techniques of information, the development of “synthetic” coordinated indicators and even the identification of new data which will help with risk analysis and conflict prevention in tomorrow's world, all these constituted the agenda for this Advanced Research Workshop. Our approach was to give as much freedom as possible to specialists in presenting their work, so that new ways forward and approaches could be identified and facilitated. The aim of this publication is to render account as faithfully as possible of the diversity of approaches taken by the 27 researchers and experts from 13 countries who participated in this Advanced Research Workshop. It will of course take time to evaluate the potential for progress in all this but these two figures point the way forward.

This introduction reviews some of the concepts related to the hypothetico-deductive method, especially the measurement of facts and the operationalization of theoretical concepts. It shows how these questions are crucial when dealing with datasets on armed conflicts, and how they relate to each of the articles of this volume. The causes and nature of the discrepancies among datasets are studied, and some proposals are made to try and improve scientific methods in this area.

How many interstate and internal wars have taken place recently between whom? How long have they been going on? And how severe are they? How many of these conflicts did take place in the past? And what are the trends? These are crucial questions peace and conflict research has tried to answer for some time now. The data collected for that purpose are relevant both for the scholars searching for explanations and the practitioners trying to prevent these kinds of conflicts in the first place, attempting to resolve them once they have occurred. The problem is that different data collections provide us with different representations of both kinds of conflict: interstate wars and civil wars. These differences are due to differing operational definitions but possibly as well to different theoretical conceptualizations. In order to make that research relevant for politics, replications of the different hypotheses with different data sets are necessary. If the results are robust, these differences would not matter that much. If, however, they are not, which seems to be the case at least for civil wars, the theoretical foundations have to be reconstructed. Only if that has been achieved, can peace and conflict research contribute with greater confidence to the prevention and mitigation of violent conflict.

In this chapter I first present a brief overview of the different datasets on armed conflict that are available. I then provide a preliminary typology which attempts to place various types of collective violence in context to each other (civil war, interstate war, terrorism, genocide, communal conflict, etc.). Finally, I discuss the consequences of having differing conceptual frameworks and definitions within the conflict data community. I argue that while the plethora of definitions may lead to confusion, heterogeneity amongst the different armed conflict data projects nevertheless serves as a key indicator for the validity of the statistical inferences researchers wish to draw. Without theoretical or empirical consensus regarding exactly what behavior constitutes armed conflict, the diversity of definitions and data provides a means to examine the robustness of the conclusions drawn in the quantitative study of armed conflict.

This paper presents the databases of the Observatory of Armaments on nuclear testing and the armament industry, and how they relate to NGO advocacy. The two key achievements from this work are the publication of a directory of sales of French weapons, and the Ministry of Defense's growing recognition of the existence of medical and environmental problems where France has previously conducted nuclear tests.

Arguably the most important aspect for researchers and policymakers is to have access to reliable information, especially when focusing on armed conflict. In order to know when there is greater risk for the outbreak of war, whether containment has been effective, or when post-conflict measures are appropriate, there is a need to clarify what an armed conflict consists of. The most commonly employed definitions of armed conflict are being measured through battle-related fatalities tied to the political goals of the warring sides. This definition excludes several other indicators of human suffering, such as violent crime, genocide, starvation, and forced migration. This chapter reviews some of the problems with expanding the concept of conflict and argues that it is beneficial to use a narrow conflict concept in order to study the relationship between war and other phenomena such as criminality and genocide.

This paper presents the database of Hostage-Barricade-Terrorism (HBT) situations and abductions, which were carried out by Palestinian terrorist organizations during the period 1968–1994, against Israeli soldiers and Israeli citizens. The study proposes a new methodology for describing processes adopted in HBT and abduction events.

Violent conflicts evolve in multiple stages. Academic and policy oriented conflict research seeks to identify indicators of conflict escalation and assess their relative weight in models of conflict analysis. Conflict scholarship suggests that structural factors condition the initial conflict dynamic. Alternatively, dynamic factors are highlighted as basic in times of crisis. This paper draws a distinction between the concepts of “long term structural risk assessment”, and “short term early warning”. It is argued here that conceptual clarity improves the predictive power of conflict analyses. By reviewing databases hosted at the Center for International Development and Conflict Management at the University of Maryland this paper introduces four models predicting conflict escalation. Three of the models are concerned with structural crisis factors, while one suggests dynamic factors are present during the escalation of violence. The paper reviews the methodologies of indicator development used within these models.

The use of scales and indexes in order to provide precise measurements of the relative state of physical and social phenomena is quite common. Despite the popularity of indexes there is considerable debate regarding their reliability and validity in the case of social research, and many researchers find scales of social phenomena to be misleading. A number of possible solutions to these problems will be suggested for measures of conflict and cooperation.

Datasets on armed conflicts are built through complex procedures involving categorizing and classifying single events, aggregating raw data and making changes as new information becomes available. They are usually designed to be shared. This paper reviews each of these steps and proposes methods to enhance both the openness and reliability of these datasets and their accompanying procedures. It aims at helping put into practice the general guidelines of scientific methodology and bridge the gap between epistemology and the management of conflict data.

The process of bringing together information (data) from different databases faces a variety of practical challenges. These make it difficult to readily access, merge, and present data collected and maintained by different researchers and organizations. Practical challenges range from the need for matching “keys” on which to merge information in an automated manner, to different data organizational structures in different databases, to issues of obtaining permission to link data through a single source. A wide range of users (e.g. journalists, military analysts, laypeople, quantitative data analysts) also want to access information in different formats, display modes, and with varying levels of precision. This paper discusses some of these practical issues with meshing data from different sources. It will also discuss the EUGene software program. This software was designed for (typically academic) analysts, and pulls together data from a number of widely-used international relations data bases and makes data merging and access to the resulting data easy for end users. EUGene has confronted some of these issues, and the solutions suggested by that experience are presented.

This paper is reporting on the ACLED (Armed Conflict Location and Event Data) project which records civil war events by precise location and by day. First the paper explains how the dataset can support and further ongoing research on civil war. Then the paper briefly presents a summary of how the dataset is defined. Lastly some of the methodological issues related to high resolution geographical coding and analysis are discussed.

Visualization is a powerful tool to derive insights from massive, noisy, and possibly inconsistent datasets. We propose a method for the visualization of conflict networks that show a set of actors together with hostile or conflictive relations on the systemic level. Our method highlights the most involved actors, reveals the opposing groups, provides a graphic overview of the conflict structure, and allows for smooth animation of the dynamics of a conflict. The visualization technique can deal with potentially complex network structures and distinguishes visually between bilateral and multilateral conflicts.