The Indicators of Dissension: Using Big Data to Assess Armed Conflicts and Political Instability

While many of our discussions on “big data” and enhanced computational capacity have focused on the structures and dynamics attending today’s innovation-driven knowledge society, relevant features and relationships are realized under the various socio-political and cultural conditions that mark societal interactions.   Of these, unfortunately, armed conflict and political instability continue to plague human existence around the world.  Accordingly, relevant data needs continue to grow in relation to the pursuit of critical research and policy analysis.  Based on the extant literature and theoretically-informed approaches, we investigate the use of various big data to construct and assess indicators for a wide range of analytical correlates and predictors of conflict and instability over time in countries and groups categorized in terms of allied and/or hostile positional identities.  Using and integrating such data is not only a question of volume, but also, importantly, of data veracity, especially given varying collection platforms and the extensive re-purposing required for analysis, synthesizing, and modeling for relevant application and research questions.  For example, what biases or limitations are present when synthesizing indicators across varied datasets?   Also, what are possible correlates and/or predictors of conflict or instability in a country or region?  To what extent do these indicators explain the presence of conflict or instability when modeled?  What are methodological best practices for synthesizing complex indicators for use in analytical modeling?  Drawing upon data from a wide range of international and regional organization sources (e.g., the United Nations, World Bank, North Atlantic Treaty Organization, and European Union) and from selected world and comparative research projects (e.g., the Armed Conflict the Armed Conflict Event and Location Data Project and the Correlates of War Project), we also develop data and tools to investigate relevant characteristics and conditions, taking advantage of the ability to manipulate them in silico to assess predictive efficacy.