Dealing with data on peace and conflict: Experiences and lessons learned

Abstract

Researching conflict and peace has never been as driven by large-N data as it is now, nor has it ever before seen such abundance of data and data-production methods. However, the analytical conundrums in the discipline have not abated, as the abundance of data brought forth further questions concerning data generation processes, research design, research ethics, and appropriate methodological approaches to studying the dynamics of war and peace. Furthermore, the potential for cooperation across research institutions and disciplines involved in studying conflict and peace, and often embedded in international research networks, is hampered by underdeveloped standards regarding data comparability in various types of analysis. For example, legal scholars may study the texts of peace settlement documents, and political researchers may study the manner in which rebel groups approach peace negotiations, but merging the data would depend on a shared set of definitions concerning what constitutes a rebel group, how the data are to address rebel group splits and mergers, and similar. In addition to that problem, two more issues are often found in peace and conflict research. First, the data are often collected at differing levels of analysis, so that aggregation and disaggregation of data require conceptual clarity, but most of all, good knowledge of particular cases and locations. For example, we know that ceasefire arrangements often refer to limited geographic areas. However, the data on scope of ceasefires and instances of violence often do not cover the same area, and instead overlap to a greater or lesser degree. This may be relatively simple to parse out for a country specialist but becomes a problem when such decisions need to me made in hundreds of cases, spanning decades and continents. Second, there is the issue of bias in data collection, stemming primarily from the nature of phenomena being studied and the locations in which conflict occurs. We know that data on violence, fighting, casualties, deaths are more likely to be reported in some areas as opposed to others (richer, urbanized, better connectivity, etc.) and the next set of challenges to the study of intervention in peace and conflict processes will have to grapple with these sources of bias in order to produce valid research and policy recommendations. This presentation outlines these issues in more detail and presents the work of the University of Edinburgh’s Political Settlements Research Programme (PSRP). The PSRP and particularly the team around the PA-X Peace Agreements Database and Dataset (www.peaceagreements.org) have aimed to bridge the divide between conflict data and data on peace negotiations, by considering the manner in which peace negotiation features may be matched to events, organizations, persons, groups that are taking part in documented organized violence. In relation to this, we are developing a suite of PeaceTech abilities, based on the idea of using peace negotiations information in conjunction with other conflict-related data to assist researchers, citizens of countries in conflict, policy specialists, mediators, and diplomats.