Representation and Realities : Investigating Roadside Bias in Coded Civil War Data

Abstract

This project builds on prior disaggregated studies of civil conflict by discussing the utility of news media as a source for coded conflict data, with Liberia serving as a case study. The Armed Conflict Location Event Data (ACLED), made available by the International Peace Research Institute Oslo and the Center for the Study of Civil War, breaks civil war into individual conflict events at the local level. This offers an exciting opportunity for small scale analysis of rebel movement. However, there is a striking tendency for events to occur near roads, which raises questions about a possible roadside bias in news coverage of war, as news accounts are the primary source of event data. By characterizing ACLED event locations through spatial analysis, and by qualititatively assesing news reporting during Liberia's war through personal interviews and engagement with historical records, this paper examines a potential under-representation of rural areas in data of this kind, and explores the implications.