Abstract
Scholars of political violence often face problems concerning data availability. Research on the perpetrators of that violence is no exception. Over the past 40 years we have made great strides in understanding who joins in violent action and why, yet have rarely probed the representative nature of the subjects queried or contemplated the implications of this sampling for our conclusions. It is generally assumed that those left to ‘tell the tale’ about what transpired are representative of those who participated in the violence. In this article, we use the context of the 1994 genocide in Rwanda to probe questions about which perpetrators of violence we include in our research and subsequently, who we miss. We theorize an often overlooked group of perpetrators, the ‘murderers in the middle’, who take orders from above, mobilize others to kill, and zealously participate themselves. We contend that this group of perpetrators is potentially unique from those generally captured, identified, and studied in that they are likely to have actively and willingly engaged in violence for personal gain as well as for ideological reasons. Systematically missing groups of perpetrators has potential implications for research on participation in mass violence as well as our understanding of why this behavior occurs.
Those of us interested in understanding political violence (e.g. genocide, civil war, atrocities/massacres, and the like) face tremendous hurdles concerning the access and availability of reliable information about what happened and why (Ball, 2005; Davenport, 2009; Seybolt, Aronson & Fischhoff, 2013). This obstacle is particularly notable with regard to the study of participation in mass human rights violations 1 where researchers depend upon evidence taken from survivors, human rights advocates, journalists, and perhaps most importantly, the perpetrators themselves. Each of these sources has unique limitations for data generation. For example, individuals who suffered from the violence or bore witness to the activity may be either dead or traumatized – respectively, unable or unwilling to speak about who did what to whom. Information from those who report on relevant behavior (e.g. human rights activists and journalists) is often secondhand from those who were not present at the events. Information on participation and motives can come from those who engaged in violence themselves – a group that is potentially very aware of what took place as well as why, but potentially unwilling to talk about what they have done because of fear of prosecution or persecution. All three sources have provided useful insights into the phenomenon of interest. This said, information taken from the last source (perpetrators) has generally been viewed as being the most important for understanding mobilization for mass political violence, exploring the issue of why individuals engage in relevant activity and how they go about doing it. Research on participation is central, for if there are no perpetrators 2 (i.e. no participants) there can be no violent action.
Yet there are known limitations to the use of perpetrator testimony. While the study of mass political violence has put a premium on testimony and the act of ‘bearing witness’, the role of perpetrators in contributing to this narrative is problematic (Schmidt, 2017). We know that perpetrator accounts are often biased as perpetrators ‘self-fashion’ in interviews and in court (Straus, 2017). It can be difficult to interpret whether this self-fashioning or neutralization is a product of ‘contemporary social expectations, general tendencies to minimize moral guilt, or processes originating during the genocide itself’ (Anderson, 2017: 57). Furthermore, perpetrators may have legal incentives to lie or fabricate their stories. All of these limitations have implications for the conclusions we can draw from this research.
Despite these known limitations, the literature has rarely probed the implications of potential limitations within the sampled population for what is concluded. Theoretical conclusions assume that those perpetrators who were available and drawn upon to ‘tell their tales’ about what transpired are representative (in some meaningful way) of those who participated in the violence of interest when it occurred. But is this the case? Are there others who are not queried (i.e. those that are systematically ignored by researchers)? If these neglected perpetrators were asked questions, would the tales they tell influence the conclusions about what violence takes place and why? This is the topic of our investigation.
Within this article, we theorize an often overlooked group of perpetrators, the ‘murderers in the middle’, who take orders from above, mobilize others to kill, and zealously participate themselves. As theorized, these perpetrators are distinct from the ‘ordinary men’ and ‘big fish’ previously included in studies. Murderers in the middle are specialists in violence (Wilkinson, 2006) who voluntarily participate in mass political violence, or are already engaged in positions of violence before a conflict begins. We contend that this group of perpetrators is potentially unique from those generally captured, identified, and studied in that they are likely to have actively and willingly engaged in violence for personal gain as well as for ideological reasons. Differing from most understandings, this group is not simply ‘swept up’ into acts of violence; instead, they instigate and sustain an environment in which violence is considered a viable option and participate in significant amounts of that activity. These ‘murderers in the middle’ represent a crucial part of the perpetrator population, and in cases such as the Rwandan genocide, they may be responsible for the majority of the violent activity.
This group of perpetrators is particularly difficult to study. Generally, these individuals know they could be implicated in the violence, have the knowledge as well as the resources to escape before being arrested, and have the wherewithal to continually avoid identification and capture. To put these individuals back into the study of participation and perpetrator motives fundamentally shifts the methods, findings, and the conclusions we draw about why mass political violence takes place. Furthermore, this theoretical exercise has important implications for the study of political violence more broadly. Understanding who we sample and what conclusions we are able to draw from that sample impacts the policy proscriptions derived from our work.
Perpetrators of mass political violence
The field of conflict studies has long sought to understand why and how people participate in mass political violence (Gurr, 1970; Lichbach, 1995; Tolnay & Beck, 1995). Conceived of as how ‘ordinary men’ become killers (Browning, 1998), what makes people ‘evil’ (Staub, 1992), and how some become ‘willing executioners’ (Goldhagen, 1997), understanding participation in violent activity is seen by many as the lynchpin for the future prevention of these atrocities (Adler et al., 2008). Indeed, if there is a mass killing planned and no one joins then there can be no mass killing. Accordingly, participation sits at the core of any study regarding political violence. Below, we present a brief overview of the main findings of the work regarding perpetrator motivations as well as an overview of how this research has historically been conducted.
Participation in mass violence
Theories of mass violence participation can be divided into those relating to social and structural factors (macro motivations) (Harff, 2003; Valentino, 2004) and individual characteristics (micro motivations) (Adler et al., 2008; Browning, 1998; Fujii, 2009; Staub, 1992; Straus, 2006).
Macro motivation theories involve institutional, cultural, and situational explanations for participation in mass violence that are unique to a given society at a given point in time. These explanations include the characteristics of the governing regime (Harff, 2003; Valentino, 2004), local political and social culture (Charny, 1982; Kuper, 1983; Goldhagen, 1997), or historical climate (Adler et al., 2008) that lead people to engage in relevant behavior. Micro or individual-level explanations focus on why a particular person would choose to participate in violent action, independent of the context. Here, explanations include psychological deviance (Staub, 1992) or latent aggressive tendencies (Urdal, 2006), fear (Adler, Loyle & Globerman, 2007, 2008), urges of vengeance or retribution (André & Platteau, 1998), and social pressure (Fujii, 2009), as well as selective incentives (Gurr, 1970; Weinstein, 2006) and personal greed (Lichbach, 1995). In line with this approach, describing youth participants in the Cambodian genocide, Hinton argues that these young men ‘sought to establish a sense of identity’ within their peer groups thus rendering them vulnerable to suggestion/influence (Hinton, 2004: 267n). New work by Kjell Anderson (2018: 7) addresses the intersection between these two levels of analysis arguing that ‘most perpetrators […] respond to their surroundings in ways that minimize risk and provide coherence’.
Emerging from this work, two findings have come to predominate the study and understanding of mass political violence: (1) almost anyone could be a participant in violent behavior when placed in the right context, and (2) fear and social sanctions are more pervasive motivations for participation than political ideology. As a result, perpetrators of mass violence are generally seen as weak followers, able to be pressured into action by a strong state and/or a dynamic social network. These predictions, of course, do not hold for the Pol Pots (in Cambodia) and Bagosoras (in Rwanda) who served as the main planners and organizers of mass violence, but rather they identify the motivations of rank and file participants who make up the bulk of the killing force.
Researching participants and participation
Research into the motivations of perpetrators has traditionally relied upon two methodological approaches. 3 This research has been conducted through either: (1) case studies of individual perpetrators, focusing mainly on prominent leaders or extraordinary individuals, and personal interviews, focus groups, and historical (archival) accounts of certain groups of individuals (e.g. particular units, villages, etc.) or (2) larger-N projects with perpetrators or the general population(s) analyzed together.
Individual studies have focused mainly on prominent perpetrators, often after they have been captured and brought to trial for their crimes. In an analysis of Nazi elite participation, for example, Hannah Arendt (1963) used evidence produced by the court and personal testimony from Adolf Eichmann’s trial in Jerusalem. Similarly, Gitta Sereny (1974) interviewed the imprisoned Commandant of Treblinka, Franz Stangl, and later published a study in which she described his personal experience and motivation for participation. Alexander Hinton (2016) offers a similarly thorough analysis of the Cambodian torturer Duch (Kaing Guek Eav) and his trial before the Extraordinary Chambers in the Court of Cambodia. This work is conducted to illuminate individual motivations for participation and leadership among high profile participants in mass violence. In these cases, the perpetrators of interest have been identified by authorities, captured, and tried for their crimes.
As conceived, the findings from individual and group studies tend to be heavily reliant on the accessibility of people and records. Consequently, we argue, this methodology is particularly susceptible to sampling bias and deception. As but one example of this, much of our understanding of the banality of evil thesis derives from Arendt’s (1963) analysis of Adolf Eichmann (Issac, 1992). New research, however, has called into question this investigation arguing that Eichmann merely took on the persona of banality to save his own life, feeding the myth of the murderer being ‘caught up’ in atrocities (Stangneth, 2014).
Larger-N work has been done in the context of more recent conflicts. For example, Adler, Loyle & Globerman (2007), Adler et al. (2008), Brehm (2017), Fujii (2009), Mironko (2004), McDoom (2013), and Straus (2006) have all published research on the Rwandan genocide based on perpetrator interviews in the country. This work has generally been conducted through surveys of the Rwandan prison population or through snowball samples 4 of ex-combatant organizations. Similar work on the Holocaust includes studies by Christopher Browning (1998) and Jan Gross (2001), who used archival material on a German police battalion in Poland and the Polish village of Jedwabne to profile participants in state violence. More often implicitly than explicitly, this work is designed to be representative of the average participant in violence, but again it has focused almost exclusively on those perpetrators who have been identified or captured and, frequently, confessed to their alleged crimes. Hypotheses about the participation of ‘ordinary men’ in mass political violence generally follow from this methodology.
The missing murderers
Despite recent attention to sampling issues within conflict studies (Price & Ball, 2014), empirical investigations of participation remain focused on a very specific subset of the perpetrator population. Even if we assume that the perpetrators in question are telling the truth about their participation (a gross assumption, to be sure), a problem of sampling bias arises from existing approaches. On the one hand, we are unable to study people who were killed during the conflict and, on the other hand, we do not include people who are hiding or people openly denying their participation in violence – either inside or outside the country of interest. Based on how we research perpetrators, who could we be missing and what impact would these missing perpetrators have on the conclusions our research draws? This is addressed below.
Those left to tell the tale
To begin to systematically theorize about the characteristics of those who got away, we turn to an overview of those who were not so lucky. People who are captured and subsequently available for investigation/questioning/study generally fall into three categories.
First, there are individuals who were ‘big fish’ or were of such a high profile during the violence that they were not permitted to get away. The Nuremberg Trials, the International Criminal Tribunal for Rwanda (ICTR), the International Criminal Tribunal for the Former Yugoslavia (ICTY), the Extraordinary Chambers in the Courts of Cambodia (ECCC), and now the International Criminal Court (ICC) had/have investigative branches tasked with tracking down persons of interest once indictments have been made. This work is facilitated by the will of the international community through existing extradition treaties. Furthermore, organizations such as the Simon Wiesenthal Center and Impunity Watch specialize in documenting human rights atrocities and locating those identified individuals responsible for these crimes.
Second, there are the people who tried to escape the country, but didn’t get far. This could be individuals who were captured by the state or international community in POW camps, refugee camps, or neighboring countries. This group also includes individuals who voluntarily returned home after a certain period of time and were subsequently identified. While these individuals may not have been high priority for the international community or their own government, this group lacked the resources, knowledge, and motivation to go into hiding and stay hidden, at least for very long.
And third, there are the individual perpetrators who never left in the first place. This group is interesting because it is theoretically composed of individuals who thought that participating in mass violence was not wrong or at least was morally justified in some way, those who thought their actions were not punishable when compared with the actions of others, or those individuals who potentially misjudged the political climate in the post-violence period particularly in regards to accountability. In this category, perpetrators were identified and/or captured in proximate locations to where their crimes were originally committed or returned to those locations shortly after the violence/conflict ended.
Missing murderers in the middle
As identified above, our current understanding of perpetrator behavior is derived almost entirely from those individuals who were captured and available to tell their tales. These are not the only perpetrators, however. What of those individuals killed during the conflict, those people in hiding, or those openly denying their participation in the violence? Surviving individuals have two important characteristics which may be potentially different from those currently highlighted in perpetrator studies. First, these individuals knowingly fled, suggesting knowledge of guilt or personal threat and, second, these individuals had the material resources and/or knowledge to make an effective escape and to remain hidden.
We refer to this category of perpetrators as ‘murderers in the middle’. While the precise characteristics of this group likely varies across conflict, we theorize some potential universal features. Theoretically, murderers in the middle are perpetrators who rank somewhere above low-level ‘ordinary men’ who engaged in very little killing, but yet below high-profile ‘big fish’ who masterminded the overall policy of execution. Evasion isn’t easy as it requires knowledge of escape routes, often fabricated documentation, and years of silence. For this reason, murderers in the middle are likely to have a more robust personal network than ‘ordinary men’, often represented by higher levels of education or personal resources. The decision to evade capture suggests that these perpetrators may have a more sophisticated political acumen. In some cases, these perpetrators are specialists in violence or have previous experience in violence generation, perhaps as members of the military or militia forces, which would give them the access to resources and political knowledge needed to escape and remain hidden. In this case, ‘murderers in the middle’ could likely be responsible for large-scale, systematic and intentional mass killing, often following orders and coordinating others to do the same. These perpetrators are also likely to be politically connected through their personal networks to the group espousing violent ideology.
This group of missing murderers is extremely important for existing perpetrator research which arguably overrepresents ‘ordinary’ perpetrators (i.e. those who were fearful, morally confused, and easily manipulated by their leaders) as well as ‘extraordinary’ planners (i.e. those who were hateful ideologues and overly dogmatic). While many scholars will agree and have identified that as a research community we are missing a particular category of perpetrators (Gross, 2001; Straus, 2006), we have failed to theorize the effects of this omission on existing theories of perpetrator mobilization and participation, as well as the impact of this bias on research, understanding, and preventative policymaking. On this point, we have two considerations.
On one hand, it is possible that omitting the missing group has no effect on the conclusions of existing participation theories. It may be that excluding people who were killed during the conflict and immediately after, as well as people in hiding following the violence, will not significantly affect the findings of existing research. This option suggests that there are no measurably different characteristics between those who died or successfully fled and those who lived and were subsequently identified and/or captured.
On the other hand, by focusing on people who survived and did not (successfully) flee the area after violence ended, we could be overrepresenting individuals with particular experiences and motivations such as those who were morally confused or overwhelmed by the situation once violence began. For example, it may be that individuals killed during the conflict were more engaged or energetic fighters, likely to be on the front lines of violence, therefore increasing the chances that they were killed. In Rwanda, Scott Straus calls this group ‘hardcore killers’. It may be that those killed were more likely to be targets of retribution violence once the state-sponsored killing had ended. In these cases, we would be systematically missing particularly willing or visible killers who were known and subsequently targeted for their participation. In essence, when we sample from perpetrators who survived the conflict we could be giving too much weight to a group of individuals who were less complicit and deliberate in their actions.
People in hiding following mass violence may introduce additional bias into our findings. These perpetrators have several distinct characteristics from those currently included in our research. They were able to get away – suggesting either luck (which is not theoretically interesting) or a certain level of resources, both economic and intellectual (which is). Perpetrators in hiding may be those that knew that they needed to get away and did so in a timely manner. And, finally, perpetrators in hiding may be those whom the international community (and other interested actors) has not made a significant or successful attempt to find. By missing individuals who intentionally and successfully chose to evade capture once the violence ended, current research is failing to address the motivations of individuals who potentially deliberately and systematically made the choice to participate in mass violence, realized the potential consequences of their actions and had the wherewithal to escape.
Perpetrator classification
There may be no real way of knowing whom we are missing and the potential impact of this missing group on existing perpetrator samples across different periods of violence. This said, it is essential to theorize the existence of these groups and their potential effect. Missing certain murderers could indeed be neglecting a significant part of the puzzle that we wish to explain.
In order to explore our argument regarding the missing perpetrators of mass political violence and the potential bias that is introduced, we use data on the 1994 genocide in Rwanda. This is an important case for our theory for three reasons. First, Rwanda is a case of contemporary genocide allowing for our most sophisticated methodological techniques to be brought to bear on these questions. Second, there has been detailed work on perpetrators and participation in this violence allowing us to critically evaluate extant research and findings. Third, and more specific to the conflict itself, the Rwandan genocide was terminated by the victory of the Rwandan Patriotic Front, and the full weight of that government and the international community (through the International Criminal Tribunal for Rwanda) has been directed towards finding genocide perpetrators, increasing the likelihood that all perpetrators were identified and brought to account. In short, if ever there were a case in which all the perpetrators should have systematically been collected and studied, this would be it.
Political conflict in Rwanda
In the first weeks of April in 1994, the Parish of Nayange, in Kibuye, Rwanda filled with frightened refugees. In those days of terror, close to 2,000 people fled to and occupied the main church building. A week later the church was surrounded by the Rwandan Army. The church was bulldozed to the ground using two Caterpillar tractors owned by a foreign building firm. Refugees, who were not killed in the wreckage, were shot trying to flee. No army official has ever been tried for this massacre. The Caterpillar operator was arrested in February 1995 and remains in prison today. (African Rights, 1995) Across the country in Butare prefecture, the local government organized a bureaucratic plan for ‘self-defense’. This plan organized mandatory patrols or killing squads for every man in the community. Failure to participate resulted in harassment and often death. These groups moved throughout Butare hunting people in hiding, raiding houses that were sheltering Tutsi, forcing neighbor to confront and often kill neighbor. There are over 2,000 perpetrators on trial and in prison for these crimes. (Human Rights Watch, 1999)
Like other genocides of the 20th century, the Rwandan genocide took place in a time of existing armed conflict (Harff, 2003). From 1990 to 1993, a war raged between the Rwandan government (the Armed Forces of Rwanda; hereafter, FAR), largely associated with the Hutu ethnic group, and the Rwandan Patriotic Front (hereafter, RPF), largely comprised of Tutsi. The RPF took up arms against Rwanda after a series of unsuccessful negotiations regarding the right-to-return for a large group of Tutsi refugees who had fled previous ethnic pogroms. The war ended with a UN-negotiated ceasefire and the resulting power-sharing agreement between the RPF and FAR contributed to the rise of extremist factions within the Hutu political elite and army (Prunier, 1997). These extremist factions used dehumanizing propaganda and fear of the security threat posed by the RPF to incite violence against Tutsi and Hutu political moderates and to consolidate political support throughout the country (Uvin, 1999).
Genocide began on 6 April 1994 and the violence lasted for approximately 100 days. The sheer scope and pace of the killing was startling and patterns of violence varied across the country. Over 800,000 people, both Tutsi and Hutu political moderates, were killed in this three-month period (upper bound estimates are closer to 1.2 million) (Commission pour le Mémorial du Génocide, 1996; Davenport & Stam, 2009; Straus, 2004, 2006; Verpoorten, 2012).
An estimated 175,000–210,000 people participated directly in the violence (Straus, 2004) with projections exceeding one million (Amnesty International, 2002). Early political killings were undertaken by the National Police and Rwandan Army (see the example of Nayange Parish above). These actors are viewed as generally being responsible for the majority of large-scale massacres as well as the targeted political killings of elites and eminent persons (African Rights, 1995). Later, local killing squads, both roving and community-based, were responsible for a new wave of deaths. These squads, made up of trained local militia (e.g. Interahamwe) and community recruits, manned roadblocks, performed door-to-door searches, and traveled through the countryside inciting violence in neighboring communities (see the Butare example above) (Fujii, 2009). These actors are viewed as generally being responsible for smaller-scale activities and a lower death count.
Participation in mass violence in Rwanda
To date, theories of participation in the Rwandan genocide have almost exclusively supported the ‘ordinary Rwandan’ hypothesis. This hypothesis privileges a narrative of neighbor-on-neighbor violence and family members who are mobilized to kill other family members, and has been further disseminated through highly popularized personal accounts (Gourevitch, 1998). This explanation promulgates the conjecture that the majority of violence was undertaken by individuals in civil society, a claim we will challenge in our discussion below.
We argue that current research on genocidaires (i.e. those who engaged in mass violence against ethnic Tutsi and politically moderate Hutu) fails to capture the experience of those who knowingly and intentionally participated in the majority of violent activity. Up to now, researchers have either conducted interviews (and in some cases surveys) with low-level functionaries who have little blood on their hands (some currently in prison and some who have been recently released), or they consult the legal testimony of a few high-level bureaucrats, currently in custody at the International Criminal Tribunal for Rwanda who also interestingly have little blood directly on their hands. As a result of this sampling, current research has missed those who deliberately carried out the bulk of the violence, but were able to effectively evade capture.
Following from this, most research on Rwanda argues that perpetrators were, in part, motivated by the fear that failure to participate in mass killings would be sanctioned either by the government or by Hutu political extremists in the area (Adler et al., 2008; Straus, 2006). Because of the strong ‘enemy from within’ vocabulary in Rwanda at the time, participants used their actions to distinguish themselves as active supporters of the current government and willing to ‘work’ in support of the nation (Adler, Loyle & Globerman, 2007). The situational context of the concurrent war created a need to defend self, family, and nation that was brought on by a fear of the RPF invasion and its potential military victory (Mamdani, 2002). Additional research maintains that participants in the Rwandan genocide were motivated by adherence to national leaders and community officials. For example, some local government officials actively supported the genocide while others tried to pacify their communities and resist local violence as long as possible, contributing to variation in participation rates across the country (Human Rights Watch, 1999).
Personal edification and greed are commonly cited motivations. Once the violence began, the general state of lawlessness was easy to exploit and some Rwandans participated in order to materially benefit from the violence (Human Rights Watch, 1999). Economic depravity and individual frustrations were voiced through large-scale looting of homes and community spaces (Human Rights Watch, 1999). High levels of looting are substantiated by the gacaca courts’ convictions of 1,260,000 perpetrators of property crime (Nyseth Brehm, Uggen & Gasanabo, 2014). Beyond the personal benefits of looting, some individuals were also directly rewarded for their participation in killings. For example, in the Gikongoro area, Lt Colonel Simba is purported to have made personal payments to individuals who assaulted Tutsi (Human Rights Watch, 1999: 309n). Furthermore, some perpetrators used the situational context of social breakdown for personal retribution (Fujii, 2009).
Group associations played a key role in determining which individuals participated in the genocide (McDoom, 2013). For example, Fujii (2009) finds that participation was often the result of social networks that were able to mobilize individual participants. This finding suggests that social sanctions, group pressure, and legitimation were useful and, indeed, essential tools in mass mobilization.
Finally, some perpetrators admit to feeling overpowered and confused by the situation. Adler et al. (2008) refer to this as the ‘tsunami effect’. In this context, individuals were no longer certain of right and wrong. Morality was reversed and participating in the killing became ‘good’ (Staub, 1992). Rather than stop to critically evaluate their actions, some perpetrators of the Rwandan genocide were caught up in the momentum of the time.
As presented, these explanations draw a picture of authority-bidding Rwandans afraid of the violence and manipulated by their peers. In this work, a group of ‘ordinary’ people participated in horrific crimes. The question remains, however, what portion of the perpetrators and the killings during the genocide does this group represent and does the focus on this group obscure an understanding of what actually took place?
Identifying those left to tell the tale
Before turning to a discussion of the implications of potential sampling bias for our understanding of participation in the Rwandan genocide, we attempt to identify the universe of cases from which our research on this case is being drawn. This is not an easy task for three reasons. First, the conflict is characterized by a high level of violence and variation in the type and form of that violence. Beginning in 1990, there were killings which took place as the result of the interstate and civil wars, genocide, petty crime, and personally motivated violence, and later following the RPF takeover, revenge violence (Verpoorten, 2012). Second, the civil war and genocide in 1994 caused one of the largest population movements in UN history (Prunier, 1997). At one stage in the conflict, UNHCR estimated that there were four to five million internally displaced persons and refugees in the Great Lakes region out of a population of approximately eight million. In addition, there was a massive migration of people outside of East Africa. Reyntjens (2004) estimates that there are still over two million refugees abroad. Third, there has been little systematic analysis of the patterns of arrests following the conflict. Wagner (2003) states that the transition government in Rwanda detained over 800 genocide suspects by mid-August 1994. The arrest rate by the end of 1994 was as high as 1,500 people per week (Rose, 2007). By early 1995, over 6,500 suspected genocidaires were being held; by 2005, that number was close to 72,000 (USIP, 1995). Furthermore, human rights organizations have suggested that many alleged perpetrators were subject to extrajudicial trials resulting in execution (Amnesty International, 2001). By 2000, 348 people convicted of genocide crimes through the national courts were sentenced to death (Schabas, 2005). 9 The number is potentially much higher.
For those that fled the country, capture involved the international community. One legal response took the form of the International Criminal Tribunal for Rwanda (ICTR). At the time of its closing, the ICTR had indicted 93 suspects of genocide and political violence and sentenced 62, while three suspects remain at large. 10 As a result of the efforts by the Rwandan government and the international community, three main sites of detention and, consequently, research came into existence:
The ICTR, which held alleged perpetrators in facilities in Arusha, Tanzania; the prosecuted were sent to prison (initially in Tanzania, but later in Mali);
Rwandan prison facilities, maintained by the government throughout the country; 11
Diverse communities in Rwanda, as those accused of low-level crimes of genocide in gacaca courts are often permitted to remain at home before sentencing and return to their communities after time served.
Perpetrator research in Rwanda has focused exclusively on those who have been held in the facilities/locales identified above. Once identified, researchers have been able to gain access to these perpetrators through prisons, perpetrator support organizations, and re-integration facilities in order to conduct their research – mostly under the permission of the Rwandan government (Pottier, 2002). 12 For example, Adler, Loyle & Globerman (2007), Adler et al. (2008) and Mironko (2004) conducted their research exclusively in Rwandan prisons. In contrast to this, Fujii (2009) interviewed perpetrators in a single village over multiple site visits. While the unprecedented access to perpetrators of violence facilitates an interaction with some of those allegedly involved in mass killing, it does not resolve the question of who is and who is not being studied. Given the nature and variation of the violence, the extreme movement of people which took place in Rwanda at the time, and the large-scale, yet haphazard, attempts to hold individuals accountable for their participation in the violence, we should be critical of our ability to develop representative samples of the perpetrator population. Below, we explore these questions by looking at how the violence in Rwanda was perpetrated as a means for theorizing who got away and what this means for our understanding of Rwandan political violence.
The missing murderers in Rwanda
The significant variation in violence in Rwanda is not well reflected in those who were identified and held accountable for their participation after the violence stopped. Specifically, very few members of the former Rwandan Army (FAR) were ever officially tried for their actions. While members of the army were responsible for some of the largest massacres, as discussed below, they are not represented in the perpetrator literature. Furthermore, there is very little research on the community organizers responsible for mobilizing villagers and coordinating participation at the largest massacre sites.
From existing survivor testimony (Gourevitch, 1998) and human rights reports (African Rights, 1995; Human Rights Watch, 1999), it seems that the largest episodes of killing during the genocide involved massacres by the army and organized Interahamwe. These massacres took place primarily in parishes, but also in ‘hospitals, schools, football stadiums, local government offices, in open ground or river banks’ (African Rights, 1995: 258). Large-scale massacres followed similar patterns. The interahamwe surrounded the buildings in order to mow down those who attempted to escape while soldiers, often accompanied by gendarmes, military reservists, communal policemen and prison wardens entered the rooms and compounds. Exits were often blocked either by militias or communal policemen [….] Sometimes tear gas was used to disorient the victims and undermine their capacity for resistance. Many of the massacres were well planned military operations. (African Rights, 1995: 258)
In an interview with African Rights, medical doctor Kamoso Pie argues, ‘The idea that the killings were triggered by popular anger is laughable [.…] Those of us who lived in Kigali did not see enraged peasants looking for these targets. What we saw were Presidential Guards, soldiers, gendarmes, and interahamwe attacking the homes of people who were known to be unpopular with extremists. (African Rights, 1995: 248) In all of the largest massacres, which were organized as military operations, soldiers played a key role in organizing and carrying them out. Soldiers not only lent their skills and weaponry – and also their authority – to the killers […] Such a large number of people would never have died in Rwanda if it was not for the role played by the army. (African Rights, 1995: 48) Soldiers and National Police, whether on active duty or retired, killed civilians and they gave permission, set the example, and commanded others to kill. Although fewer in number than civilian killers, the military played a decisive role by initiating and directing the slaughter. In the first hours in Kigali, soldiers of the Presidential Guard and the paracommando and reconnaissance battalions, along with some National Policemen, carried out the carnage in one neighborhood after another. Soldiers, National Police and the communal police also launched the slaughter and organized all large-scale massacres elsewhere in the country. (Human Rights Watch, 1999: 318–319)
If the army and paramilitary groups played the leading role in the violence as we argue, then the individuals who have been interviewed, written about, and subsequently whose tales informed research and discussion about what took place in Rwanda are associated with the smallest number and the least severe activities. For example, consider the group of perpetrators examined by the well-received work of Straus (2006). In addition to employing some national-level statistical information, this research engaged in a random sample survey of those who were arrested, sentenced, and pled guilty for genocide (210 individuals across 15 prisons). These were lower-level participants (i.e. individuals who admit to killing a single person, or being involved in a beating that resulted in a person’s death) and for a later phase of the research, Straus focused on 19 participants he designated more active ‘either for their attempts to stem the killing or in their active promotion of it’ (Straus, 2007: 128n). While identified as being more active, even this latter group did not engage in much violent behavior; indeed, the worst of these individuals only admitting to killing a few people. 16
Straus acknowledges the potential limitations of this sample. He concludes that his subjects well represent the confessor (prison) population and that his sampling technique underrepresents ‘hardcore killers’ who would have been more likely to have fled the country for good, been killed in revenge killings, or to have chosen not to confess because of the scale of their crimes (Straus, 2007: 102n). While Straus acknowledges this underrepresentation in his research, he does not theorize the potential effect that the missing ‘killers’ may have on his findings. We maintain that researchers have not been led to the modal killing or killer who participated in the elimination of the majority of Rwandan victims. Instead, they have been led to Rwandans who killed on the sidelines while the larger and deadlier campaign was underway.
Theorizing the tales we are missing
If the existing theories of participation in mass violence are limited to those individuals who have been identified and captured, then who are we missing? Perpetrators of the Rwandan genocide still at large likely fall into two distinct categories: those perpetrators actively participating in resistance to the current Rwandan government in the Democratic Republic of Congo, and those perpetrators in exile or hiding abroad. 17
Original estimates suggest that in 1994 between 30,000 and 40,000 Hutu militia men and soldiers fled to DR Congo, most of whom were involved in the genocide (Prunier, 2009). Members of this group went on to form the Rwandan Liberation Army (ALiR) and later the Democratic Forces for the Liberation of Rwanda (FDLR). These rebel groups originally consisted mostly of Rwandan Hutu, possibly Interahamwe units, who had escaped the country following the RPF victory and military personnel from the former Rwandan Army (ex-FAR) (Stearns, 2012; Stearns, Verweijen & Baaz, 2013).
To the best of our knowledge, the rebel groups in DR Congo have never been systematically interviewed regarding their participation in the Rwandan genocide. Ostensibly, this group would have similar characteristics to former Rwandan army members (who have also not been studied). Differing from ordinary Rwandans subject to investigation thus far, these are well-trained individuals who had ideological and political motivations for participating in the fighting as well as expectations of professional advancement/gain. Far from ‘ordinary’ Rwandans, this group chose to and continues to fight against the Tutsi minority in Rwanda. These individuals do not appear to be motivated by fear and uncertainty. Rather, they appear to be driven ideologues and fighters who 25 years on are still fighting the war against their ethno-political opponents. 18 Furthermore, these are the individuals that seem to approximate the community that was involved with the bulk of the killing. Ex-FAR members potentially have a different rationale as they may have been acting in the line of duty, committed to the genocidal government, and unwilling to break ranks. The group of ex-FAR who fled to DR Congo potentially represents a more ideologically committed fighting base.
Perpetrators in hiding pose another theoretic concern. No longer actively resisting the government, this group got out of the country early (before, or shortly after, the RPF takeover) and continue to hide. How did these people get away? We do not have a definitive answer to this question, but we can speculate that perpetrators who escaped capture and later prosecution had both the knowledge of their own guilt and the resources to leave the country (i.e. they had a reason, the ability, and the network to flee). This group is important because these are also not ‘ordinary’ Rwandans but rather they are members of a class who were able to relocate and resettle in neighboring countries or further abroad, possibly well-resourced and educated. Furthermore, their guilt and subsequent fear of capture suggests that their activities during the genocide surpasses that of the single killing or beating variety.
How would these groups of perpetrators (i.e. military personnel, devoted Interahamwe, and those hiding) inform the greater discussion regarding political violence in Rwanda? We theorize that this group – the murderers in the middle – were likely not subject to the same fear and overwhelming emotions that caught up ‘ordinary’ Rwandan perpetrators. This group of individuals supplied the genocidal government with the intermediary functionaries needed to carry out their plans. What is particularly interesting about this group is that they left, while the majority of Rwandans in prison today were caught in Rwanda in the immediate aftermath of the conflict. The murderers in the middle fled the country demonstrating guilt by omission as well as resources either actively taking up arms or passively relocating. Failing to account for the motivations of this group skews theories on political violence to account only for those who either believed they would not be punished or did not ‘know’ how to get away. In Table II, we return to our perpetrator classification adapted for the Rwanda case.
Suggestions for research moving forward
Perpetrator classification during the Rwanda genocide
The restriction of the contemporary perpetrator narrative and approach that accompanies it informs the conclusions that we draw about the Rwandan genocide in particular and other political violence more generally. Counter-narratives regarding intentional and ideological participation have been systematically excluded. While both the masterminds and the flunkies have been accounted for, it is the intermediaries who willingly killed and acknowledged their culpability that are falling through the cracks in our analyses and understanding. There are still further biases which confound our studies, such as a focus on male perpetrators of violence (Brown, 2018). These restrictions limit our research and muddle our findings.
So, in light of these limitations, how should we proceed? In order to effectively collect information on and study participants and participation in political violence, we must understand the sample of the relevant perpetrator population across the case of interest. In other words, understanding the scope of participation and involvement in political violence requires a broader understanding of the conflict itself. By focusing on the variation in patterns of violence across a given conflict, we will be better able to see what took place and take a step back to better understand who was involved in those activities as well as who should be the subject of research.
Before we begin to analyze why individuals committed mass atrocities, we must make a detailed accounting of the atrocities we are trying to explain. It is not a matter of simply asking who could participate, but rather who did participate and in which events. Ostensibly, the answer to the first question is everyone, while the answer to the second is a significant and meaningful population. As perpetrators are notoriously difficult to track down, the viability of the sample left available to researchers after the fact may only be determined through an understanding of the events of a given conflict and the patterns of violence that resulted. Once we have identified the relevant universe of perpetrators based on the scope of violent events, we can then draw meaningful conclusions about their motivations and participation. Furthermore, through a more thorough analysis of the participant population we will be able to draw more nuanced conclusions about the multifaceted motivations of individuals across various perpetrator groups.
If research as well as advocacy are to progress, then these communities of perpetrators – the missing murderers – must also be addressed. Indeed, without a greater understanding of this group we weaken the ability to implement effective policy to stop or reduce the severity of large-scale political violence. To do this, however, scholars will have to modify how they conduct their research, and consumers of the information that they produce will have to modify how readily they accept the often compelling as well as heart-wrenching stories that such work delivers.
