Visualizing Ethnic Fractionalization, Conflicts and Power Relations

Why do ethnic conflicts arise? This question has been a persistent one in the field of political science, and has generated various theories and approaches attempting to explain conflict proliferation from different angles. Consulting current political science literature and Prof James Fearon, an expert on ethnic conflicts, we identify two main approaches to understanding the problem: one has ethnic fractionalization as the unit of analysis, while the other puts ethnopolitical configurations of power in focus. In our visualization, we aim to synthesize these two approaches with group-level data by lending ethnic power structures, ethnic fractionalization, linguistic distances and conflict incidence visual meaning. Having merged datasets from multiple sources and generated a new dataset on linguistic distances, we create an integrated visualization system using multiple sources. Methods employed include the use of normalized stacked bars, animated orbital structure and the novel application of MDS on visualizing linguistic distances. At least from preliminary user-testing, our final product has enabled users to relate historical contexts to visualizations, facilitated trend spotting and helped analysts evaluate quality of dataset in new ways. INTRODUCTION Recent quantitative scholarship on ethnic politics and conflicts has generated a rich trove of data pertaining to ethnic fractionalization, polarization, power relations and conflicts. Using these data, political scientists have come up with various theories to answer the perennial question of why civil conflicts emerge. Our research reveals two key approaches to understanding ethnic politics and conflicts: one examines ethnic diversity and its relation to armed conflict, as represented by works of Easterly, Alesina, James Fearon & David Laitin (2003) and Alberto Alesina et al (2003); the other approach puts ethnopolitical configurations of power in focus and employs a more recent global dataset by Cederman, Wimmer, Min (2009). While there have been visualizations that visualize countrylevel data and map conflicts, no attempt has been made to visualize group-level data that can be useful for analyzing the structure of ethnic diversity and power relations. The objective of our system design is to synthesize the aforementioned two approaches in understanding ethnic conflicts, spatializing ethnic diversity and ethnic power relations in one data exploration platform. More specifically, our visualization seeks to examine why civil conflicts arise by: 1. Revealing internal ethnopolitical configurations of power within countries and how they change over time; 2. Visualizing civil conflict incidences and their relations to internal power structures of countries; 3. Lending “ethnic diversity”, a key variable in the debate on civil conflicts, greater visual meaning by merging and visualizing data pertaining to ethnic fractionalization, polarization and linguistic distances; 4. Facilitating cross-country comparisons helpful for spotting patterns and testing hypotheses, by enabling user-directed custom selection of countries of interest. We worked with James Fearon from the Political Science Department, a leading scholar in the study of ethnic conflicts and civil wars, who provided us with a replication data from his famous paper “Ethnicity, Insurgency and Civil War”, as well as invaluable feedback on our visualizations. We also consulted Jeremy Weinstein, a scholar known for his work on ethnicity and development, from whom we learnt a great deal more about the subject matter and ongoing studies in the field amenable to visualization. After multiple need-finding sessions and in-depth research on current conflict literature, our group decided to visualize three variables of interest: ethnic fractionalization, a metric calculated using linguistic distances, conflict onsets and ethnic power relations. This paper presents relevant research on the problem at hand, related work that led us to making the design decisions that we did, visualization techniques and finally an assessment of results and insights that our visualization platform has generated. RELATED WORK Visualization of conflicts is not new. The need to preempt conflicts in order that appropriate measures can be taken before an outbreak has driven defense agencies and intelligence communities across the world to devise tools and visualizations that can help analysts do trend-spotting. The demand for visualizations that can help us better understand the nature of ongoing and future conflicts has been growing, especially post Arab-Spring. Ongoing civil and ethnic conflicts in the African continent have also prompted renewed interest – both in the political science discipline and visualization field – in visualizing conflict data. Preliminary research led us to examining several visualizations that share a similar motivation. Visualizing Complex Vulnerability in Africa: The CCAPS Climate-Conflict Mapping Tool Figure 1: CCAPAS Climate-Conflict visualization In a bid to identify regions in Africa that are most vulnerable to climate change, the Climate Change and African Political Stability Program (CCAPS) created a mapping tool that visualizes a combination of datasets on international development projects, national governance indicators, incidences of conflict and climate vulnerability. 1 Interestingly, this tool allows users to interact and customize visualization by assigning different variables different weights (e.g. if a user thought governance should have more weight within the model it could re-weigh baskets of indicators) so that the resulting visualization can disaggregate data into its component parts. With an intuitive interface and compelling visuals, the tool can potentially be a valuable resource for policy analysts and researches to assess the complex interactions that take place among environmental, political and social factors. Advanced filters allow the user to identify a subset of conflicts and aid projects. World Conflict Visualizer 1 The Robert S. Strauss Center for International Security and Law. “Climate Change and African Political Stability Program”. Accessible here: http://ccaps.aiddata.org/dashboards Figure 2:Conflict History Visualizer (www.conflicthistory.org) This interactive web interface allows users to learn about conflicts that have occurred throughout history from the 18 th century all through to present data. In terms of visualization technique it is quite straightforward, involving a simple geographical mapping of conflicts around the world across time. The information panel with a comprehensive list of conflict-related information is useful for a user who would like to engage with historical information on demand. Zeroing-on group-level dynamics Most platforms available today display country level data on conflict-relevant indicators such as the frequency of protests, quality of governance, incidence of conflicts. While these are certainly useful for trend-spotting across countries, they offer little insight into underlying ethnic structures, diversity and power relations that drive most civil conflicts in history. A simple, straightforward mapping of civil wars and ethnic conflicts may reveal certain clustering effects across geographical boundaries but do not reveal group membership of actors that drive interactions in ethnic conflicts. Our interactions with current political science literature and consultations with Professor James Fearon and Jeremy Weinstein have prompted us to visualize group-level data to uncover group dynamics across time, something that has not yet been attempted. Conflict networks As far as we are aware of, the only study that looked into group-level dynamics is a paper by Brandes and Lerner (2008) in which they proposed a general method for visualizing conflict networks. 2 Using an array of visualization techniques, their method highlights the most involved actors, reveal the opposing groups and provide a graphic overview of the conflict structure. For instance, given a list of events, their method visualizes the resulting conflict network in which distances are driven by the severity of conflict relationship between parties of conflict. In contrast to pure dyadic analysis, such networks give additional information about indirect ties (e.g. enemies 2 Ulrik Brandes, Jurgen Lerner. “Visualization of Conflict Networks”. University of Konstanz. 2008 [Accessible here: http://www.inf.unikonstanz.de/algo/publications/bl-vcn-08.pdf] of enemies), density complexity and structure of actors’ network environment. Figure 3: Visualization of conflict network constructed from events related to War in Bosnia. The nodes represent political actors, edges represent conflictive relations) We find this an interesting visualization technique applicable to our problem at hand, specifically in visualizing linguistic distances. NEEDFINDING Consulting James Fearon and Jeremy Weinstein from the Political Science Department and conducting research on the subject matter led us to focusing on two dominant approaches to understanding the rise of civil conflicts: ethnic fractionalization and inter-ethnic group power