Hoover Green, 2011

Citation: Hoover Green, Amelia. 2011. Repertoires of Violence Against Noncombatants: The Role of Armed Group Institutions and Ideologies. Ph.D. diss., Yale University.

Abstract: Political science literatures on violence against noncombatants during armed conflict have generally considered either "violence" as an under-specified general category or limited their consideration of violence to lethal violence. Yet non-lethal episodes of violence, including forced displacement, detention, torture, and sexual violence, make up a majority of the episodes of violence against noncombatants in most conflicts. This dissertation defines the "repertoire of violence" as the subset of types of violence that an armed group uses regularly against noncombatants, and their relative proportions. It shows, first, that there is considerable variation in repertoires of violence: armed groups that commit similar levels of one type of violence frequently do not commit similar levels of other types of violence. Some armed groups, for example, eschew rape while committing high levels of other violence, such as killing. Other groups commit high levels of many types of non-lethal violence, but kill comparatively few civilians. What explains this variation? I argue that repertoire variation is indicative of institutional differences between armed groups. In particular, armed group leaders face a "commander's dilemma:" they must recruit and train combatants to use violence near-automatically, and desensitize combatants to high levels of violence, yet they must retain operational control over combatants' use of violence. Relying on results from social psychology, military sociology and behavioral economics, the thesis argues that traditional "discipline" is insufficient to retain control, and that therefore only those armed groups that build institutions that valorize controlled violence and work to change combatant preferences over violence (e.g., indoctrination mechanisms, political education) will display narrow repertoires of violence. The theory is tested using a mixed-method, sub-national comparative investigation of armed groups and repertoires of violence during civil war in El Salvador. To measure institutional characteristics I use qualitative data from several dozen in-depth interviews with Salvadoran ex-combatants from five insurgent organizations and several state groups, as well as archival documents. To measure repertoires of violence, I employ a demographic statistical method, Multiple Systems Estimation, to obtain statistically sound measures of several types of violence over time, space and armed group. Comparing these data, I show that institutional theories predict repertoire variations more effectively than do existing theories of violence against civilians, such as those emphasizing strategy or resource endowments.

Document Access: I am happy to provide copies of the full dissertation in .pdf format. Please contact me via email (myfullname at drexel dot edu or myfullname at gmail dot com).