LAUSR

Abstract How does political violence affect popular support for peace? We answer this question by examining Colombia, where in 2016 the people narrowly and unexpectedly voted against a peace agreement designed to end a half century of civil war. Building on research on the impact of political violence on elections as well as research on referendum/initiative voting in the United States, we argue that local experiences with violence and the political context will lead to heightened support for peace. We test these expectations using spatial modeling and a municipal-level data on voting in the 2016 Colombian peace referendum, and find that municipal-level support for the referendum increases with greater exposure to violence and increasing support for President Santos. These results are spatially distributed, so that exposure to violence in one municipality is associated with greater support for the peace referendum in that municipality and also in surrounding areas. Our findings have implications not only for Colombia, but for all post-war votes and other contexts in which referenda and elections have major and/or unexpected results.