Predicting the End of Recurrent Civil Wars

Abstract Chapter 3 outlines the process of selecting empirically grounded hypotheses on the factors that best explain the end of recurrent civil war. First, it presents the novel Conditions of Recurrence Dataset (CoR-D), which captures fourteen peace processes that experienced conflict recurrence before achieving the negotiated and stable end of large-scale conflict-related violence between the same adversaries. Second, it explains the use of Classification and Regression Trees (CART) to identify patterns predicting conflict recurrence and non-recurrence. Third, the chapter describes the CART analysis and a range of robustness tests (including alternative machine learning techniques and sample modifications). It concludes by presenting three robust and empirically grounded hypotheses on the factors best predicting the end of protracted civil war: UN leadership of the mediation process; plural justice provisions; and measures to include women in post-conflict society.