In April 1994, the Clinton administration found itself in a difficult situation. Standing U.S. policy at the time was to work in concert with the United Nations to actively intervene to prevent or halt any genocide that occurs in a country or countries. This policy evolved in reaction to the failure of the U.S. and others to take a more active stance in halting the Jewish genocide perpetrated by the Germans during World War II. But Rwanda in April, 1994, was a long way from Poland and Germany in 1944. The mass killings in Rwanda in 1994 differed from previous genocides in scale, location, speed, perpetrators’ intent, victims’ culpability, and method. The Clinton administration’s hesitance to intervene in Rwanda in 1994, while likely driven by domestic political concerns, reopened the debate about what exactly is genocide, and how is it best understood, as a singular event or as an ongoing strategy of organized violence that exists as part of a larger more explicitly political process.

This project aims to develop a more nuanced understanding of the processes by which mass killings in the Great Lakes region of central Africa occurred during the 1990s. Specifically, we seek to understand how the process of genocide varied across space and time. We hope to be able to begin to explain two main issues. First, at the group level, what accounts for the behaviors of those that chose to retain their decency amid the violence around them. While literally hundreds of thousands of people actively or passively participated in the killings and violence, others made the deliberate choice not to; amidst genocide there were oases of humanity and resistance. The decision to opt out varied across both space and time, but we have little understanding of why in some prefectures little killing took place, and why in others the compliance with orders to commit violent acts were heeded regularly. Second, following this analysis, we hope to be able to better understand the reconciliation process that is now being used in Rwanda by tapping the attitudes of individuals inclined to forgive their neighbors versus those who are not, while controlling for the differences in their experiences before, during and after the events of 1994. The effectiveness of reconciliation and national healing programs after large-scale mass-killings depends in no small part on detailed understandings of who the perpetrators were, what the specific crimes were and the victims’ role in placing themselves at the scene. Treating all genocides alike, as a binary event/non-event (like that employed within the existing quantitative literature), hinders our ability to tailor specific reconciliation and judicial processes such as the current Gacaca process now underway in Rwanda, which is a local, community-oriented effort, being used throughout the country. In order to design and implement political and judicial processes to allow the victims to move forward with their lives as individuals and to develop as nations, we need to better understand the process by which genocidal activities take place and to be able to understand the variance of the process over both space and time.

In an effort to better understand genocide and those factors that influence individuals’ decision to opt out of and explicitly to defy this behavior, we have already collected data on several identifiable activities related to the Rwandan genocide from April 6th to July 13th, 1994. In Figure 1, for example, we track the number of killings, by day and by prefecture. In some areas of the country we find that the killings continued for several months, in others the killings ended after roughly three weeks. For other activities such as individual acts of killing, torture (beating and raping), other violent action (e.g., shooting, burning, spearing) and non-lethal activity (e.g., abducting, detaining, and looting), we find even greater variance from province to province over time. In Figure 2 (see Appendix 1) we illustrate the national daily distribution of killings. In Figure 3 (also see Appendix 1) we illustrate that the instances of individuals proactively saving potential victims, or otherwise resisting the genocide, also varied by location and date.

Figure 1. Number of killings by day and by prefecture.

In our existing data, we compiled our first sets of dependent variables – various acts of political violence and resistance. The balance of the proposal lays out our hypotheses seeking to explain the variance in these acts. Our hypotheses are drawn from two competing theoretical approaches that hope to explain and predict which members of society will chose to participate in violent acts and those who will choose to opt out. As we will show, we plan to utilize both individual level as well as aggregate data. For example, in our coding of the violence we record in our data who was involved (the organizational affiliation of the perpetrator), where the violence took place (the cell, district, secteur and prefecture) and when it took place (the day) – see codebook. While we have collected data at the individual event level, in order to protect the identity of both victims and the accused attackers, we aggregated the data to the district and prefecture-day.

We used three sources with nation-wide coverage of group actions for the measurement of genocidal activity and resistance: African Rights (a Rwandan human rights organization), Human Rights Watch (the international human rights organization), and a Rwandan government report issued by the Ministry of Youth, Culture and Sport. We seek support for the collection and formatting of additional data to supplement data already collected at the individual level (descriptions of victims and perpetrators of individual violent acts) to be used as independent variables within an analysis of behavioral variation during the of genocide: e.g., measures of power concentration, crisis, and numerous contextual factors that influence the willingness of individuals to participate or to opt out. We rely upon three sources for this: the Rwandan census of 1991, country mapping data complied by the Centre for Conflict Management and the GIS Centre at the National University of Rwanda, in Butare, and survey data from several Rwandan, French and Belgium authorities each representing a particular politically interested viewpoint on the genocide.

Our proposal is divided into three sections. We begin with an overview of the existing scholarship on genocide. Following this, we will present our research design and then provide a brief concluding statement.