Scholars have made significant advances in understanding conflict-related sexual violence since the turn of the century. In particular, we now understand a lot more about how sexual violence varies across conflicts and armed organizations (State or non-State), and significantly more about why it does so. Some organizations rape boys and men as well as girls and women; some target only members of a particular ethnic group, while others target more broadly. Some organizations more often engage in rape by multiple perpetrators than by a sole perpetrator. Most importantly, not all armed organizations engage in rape. The forms of sexual violence by armed organizations during conflict also vary, including sexual torture and mutilation; forced pregnancy, abortion, prostitution and marriage; and sexual slavery, as well as rape. In some organizations, women as well as men perpetrate sexual violence. While there is a lot we do not understand about this variation, in light of the ongoing suffering that such violence inflicts and the quality and quantity of recent research, it is timely to assess what we know and the implications for policy.
In this paper, I first summarize recent research (in social science, but also public health) that documents the patterns of sexual violence – its form, targeting and estimated frequency – on the part of both State actors and non-State actors (rebels and pro-government militias) during conflict, including the absence of rape on the part of some actors. In particular, whether rape by armed organizations is significantly more frequent than that by ordinary civilians varies across conflict settings. I then argue that classic explanations for conflict-related rape do not account for the full spectrum of the documented variation. I distinguish different meanings of the oft-repeated claim that conflict-related sexual violence is part of a continuum of violence between peace and war and suggest that recent research supports only some versions of this theory; many patterns of sexual violence by armed organizations during war do not reflect pre-war patterns.
Second, I summarize recent research that analyzes why armed organizations exhibit such variation in their patterns of conflict-related sexual violence, drawing on theoretical approaches that analyze the internal dynamics of armed organizations. After arguing that the distinction between “strategic” and “opportunistic” rape is insufficient, I suggest that when armed organizations engage in frequent rape, they often do so as a practice rather than as a strategy. I then briefly discuss the conditions under which rape as a strategy and rape as a practice occur. Finally, I lay out some implications for policy, advancing some principles that should guide its development.
Throughout, by “conflict-related sexual violence” I mean sexual violence by armed organizations during armed conflict. By “armed organizations” or “armed actors” (I use these two phrases interchangeably), I mean State actors (military, police, paramilitary organizations under the direct command of other State actors) and non-State actors (rebel and militia organizations). When I refer to sexual violence by civilians, I do so explicitly. By “sexual violence” I refer to sexual violence as defined by the Rome Statute of the International Criminal Court, which includes “[r]ape, sexual slavery, enforced prostitution, forced pregnancy, enforced sterilization, or any other form of sexual violence of comparable gravity”.Footnote 1
Variation in conflict-related sexual violence
Perhaps the most important finding of recent research is that sexual violence during conflict varies sharply across armed actors.Footnote 2 Many armed organizations engage in widespread sexual violence, but not all do so: 59% of 177 armed actors in the civil wars between 2000 and 2009 in twenty African countries were not reported to have engaged in rape or other forms of sexual violence.Footnote 3 There is of course severe under-reporting of conflict-related sexual violence in many contexts; however, these data reflect reporting of sexual violence after human rights and women's organizations had begun to actively document rape and other forms of conflict-related sexual violence. While under-reporting no doubt continues, the documented differences across armed organizations are very sharp. Armed organizations that were not reported to have engaged in even moderate levels of rape include some State militaries, some leftist insurgent organizations and some secessionist organizations.Footnote 4 Indeed, some armed organizations engage in ethnic cleansing – often presumed to be a setting for widespread rape – without engaging in sexual violence. The best cross-national dataset available confirms that sexual violence (including rape) varies across State militaries, insurgent organizations and pro-government militias; indeed, for all three types of armed actors, a strong majority is not reported to have perpetrated sexual violence between 1989 and 2009.Footnote 5
In some conflict settings, the frequency of sexual violence by armed actors is significantly less than that by intimate partners, acquaintances or strangers.Footnote 6 For example, according to a survey of twelve rural communities in Cote d'Ivoire, during the 2000–2007 conflict 4% of women and 2.2% of men endured forced or coerced sex by perpetrators other than intimate partners; of those men and women, less than a tenth had been forced or coerced by combatants or uniformed officials (0.3% of the women surveyed and 0.2% of the men).Footnote 7 The prevalence of forced or coerced sex by combatants during the eight years of the crisis was significantly less than the prevalence of intimate-partner forced sex in the year following the crisis, which was 14.9% among ever-partnered women.Footnote 8 According to a survey in fifteen conflicted municipalities in Colombia, 3.4% of women reported having been raped between 2000 and 2009. The reported rate of rape by family members was triple the reported rate by combatants and 50% more than the reported rate by strangers.Footnote 9 Where belligerents not only effectively prohibit rape by their members but also enforce norms against rape by civilians (including spousal rape), the overall frequency of rape during war may be significantly less than peacetime levels.Footnote 10 Of course, in many settings the rate of rape during conflict, which includes that by armed actors as well as civilians (including intimate partners), is significantly greater than that during peacetime because some civilians and some armed actors engage in more rape than they would have had peace continued. In some but not all settings, conflict-related sexual violence (usually defined as that by armed actors) is greater than that by family members during the conflict, as in the eastern Democratic Republic of the Congo (DRC), where a survey of the North and South Kivu provinces and the Ituri district found that of the 39.7% of women and 23.6% of men who had suffered sexual violence in the past sixteen years, 74.3% of the women and 64.5% of the men reported that it had been conflict-related (carried out by armed actors).Footnote 11 The reported rates of intimate-partner sexual violence (IPSV) and community-based sexual violence were much lower.
Conflict-related sexual violence varies in form and targeting as well. Armed actors appear to engage in sexual torture, sexual slavery, non-penetrating assault, sterilization, forced prostitution and pregnancy to highly varying degrees. Some target women and girls who belong to “enemy” groups during ethnic or political cleansing; others appear not to use such criteria. Some armed organizations target only females, while others target males as well, an emerging theme in research on conflict-related sexual violence.
Yet too narrow a focus on conflict-related sexual violence runs the risk of ignoring contextual differences that are essential. Knowing whether rape occurs in the context of genocide or torture, for example, is essential to analyzing why it occurs. Variation in repertoire is complex, belying any dichotomous categorization such as those that engage in all forms of terror vs. those that engage in restraint.Footnote 12 For example, the number of female Muslim Bosnian civilians raped by Bosnian Serb militias appears to be roughly the same order of magnitude as the number of male Muslim Bosnian civilians killed by them;Footnote 13 in sharp contrast, the Tamil Tigers appear to have rarely engaged in rape of civilians but killed many.Footnote 14 In short, the frequency of lethal violence may be very different from that of sexual violence.Footnote 15
An explicit concept of “pattern of violence” may help clarify this complexity. A “pattern of violence” by an armed organization is comprised of the repertoire of forms of violence in which it regularly engages and, for each element of the repertoire, the targeting and frequency of that form of violence for the specified time period and region (of course, the pattern of a particular unit of the organization can be similarly defined).Footnote 16 The organization's sexual violence repertoire is thus a part of its overall repertoire. In analyzing targeting, scholars often use a broad, qualitative distinction between selective (targeted because of an individual's behaviour) and indiscriminate targeting. Increasingly the literature on violence during armed conflict also distinguishes a third category, that of collective targeting based on identity as members of an ethnic or religious group, a political party or a village thought to represent or support the rival.Footnote 17 Of course, in analyzing sexual violence repertoires and targeting, comparison must often be qualitative or ordinal given data constraints.
Classic explanations of conflict-related sexual violence
Classic theories advanced to explain conflict-related sexual violence explain only a small part of the observed variation. In particular, theories to explain conflict-related rape do not account for its variation because they over-predict rape during war. The militarized masculinity approach, for example, argues that societies in war develop (or draw on) institutions and norms that inculcate a highly militarized masculinity based on sharp distinctions between genders: to become men, boys must become warriors.Footnote 18 The result is that combatants represent domination of the enemy in highly gendered terms and use specifically sexual violence against enemy populations. However, the argument does not explain the absence of sexual violence on the part of some very effective insurgent and State armies.Footnote 19 Similarly, increased opportunity to rape during war cannot account for armed organizations with ample access to civilians that engage in little rape.Footnote 20 Nor does the “substitution” argument (that rape “substitutes” for sex with prostitutes, camp followers, female combatants or willing civilians) account for the targeting of particular groups of women, the often extreme violence that frequently accompanies conflict-related rape, the occurrence of sexual torture, or rape by forces with ample access to prostitutes.Footnote 21
Relatedly, patriarchal culture cannot account for the observed variation as it too over-predicts conflict-related rape. Moreover, such broad cultural proclivities cannot account for asymmetric conflicts where one party to the war promotes sexual violence while the other does not, a pattern true of almost 40% of civil wars.Footnote 22 While devaluation of women may be a necessary condition for the occurrence of widespread sexual abuse of women, this general notion of patriarchy is too broad to account for the observed variation; it is not a sufficient condition.Footnote 23
Similarly, the argument that conflict-related rape occurs because the armed organization orders or promotes its use as a strategy of violence against civilians cannot account for the many armed actors that engage in other forms of violence but do not engage in rape. Of course, strategic sexual violence in various forms does occur during some armed conflicts on the part of some armed actors: as sexual torture against detainees to obtain information; as institutionalized forms of sexual slavery and forced marriage; and as a form of terror or punishment, to control resources or territory, or to “cleanse” an area of a targeted population.
Thus, many of the classic theories explain only part of the observed variation. Indeed, they generally focus narrowly on rape, predict more conflict-related rape than the already tragic levels observed, and fail to explain the fact that many armed organizations do not engage in even moderate levels of rape or other forms of sexual violence.
Conflict-related sexual violence is undoubtedly a complex phenomenon; any mono-causal theory is unlikely to account for the observed variation.Footnote 24 However, promising combinations of causes for conflict-related rape, such as militarized masculinity and opportunity together, also do not explain the variation for the same reason: the combination over-predicts rape. In light of the sharp variation in sexual violence across armed actors, on the one hand, and the limitations of theories focused either on individual incentives or on broad assertions of the strategic value of conflict-related sexual violence, on the other, the literature increasingly takes the armed organization as the unit of analysis, documenting variation in the institutions and cultures of organizations to explain variation in their patterns of sexual violence. Before assessing this recent literature, I will discuss whether conflict-related sexual violence can be seen simply as part of a continuum of violence.
Does conflict-related sexual violence fall along a continuum of sexual violence?
Do we need to account specifically for conflict-related sexual violence? The continuum thesis asserts that conflict-related sexual violence is part of a continuum of violence in general or sexual violence in particular.Footnote 25 In essence, the thesis holds that the same gender relations that drive sexual violence during peace drive it during war, and therefore patterns of sexual violence in peace and war differ in degree but not in kind. At this level of abstraction, the thesis is obviously true in the banal sense that all violence falls along some violence continuum and gender relations are integral to sexual violence (against women and girls and also against men and boys). It is also true in the specific sense that men often rape women for sexual gratification and as an expression of power and rights over women as property in wartime as in peacetime. Particular forms of sexual violence by intimate partners, family members, acquaintances and strangers are prevalent in many societies whether or not they are at war.Footnote 26 For example, researchers found no significant difference in the rates of sexual coercion by intimate partners in the year just before and that just after the conflict in East Timor, which suggests that the rates continued during the conflict as well.Footnote 27
Yet other interpretations or implications of the thesis when applied specifically to rape are false. For example, in her analysis of the best available dataset on conflict-related rape, Dara Kay Cohen found no correlation between a standard measure of patriarchal institutions and the level of conflict-related rape.Footnote 28 As argued above, such societal-level explanations are hard pressed to account for the asymmetric pattern of conflict-related rape in many civil wars. Nor does the combination of patriarchy and opportunity account for the absence of rape by armed actors in patriarchal societies that have ample access to civilians.Footnote 29 Nor does the continuum thesis explain the innovations in sexual brutality that we observe on the part of some armed organizations (rape with guns, sexual mutilation, etc.), innovations that would appear to have little precedent during peacetime. Moreover, when armed organizations engage in high levels of rape during conflict, the very high fraction of rapes that are carried out by multiple perpetrators contrasts sharply to the fraction observed during peacetime.Footnote 30 In Sierra Leone, for example, 76% of conflict-related rapes of women were by multiple perpetrators.Footnote 31 In three war-torn provinces in the eastern DRC, 73% of rapes of women and 38% of rapes of men were by multiple perpetrators.Footnote 32 Nor does the thesis account for high levels of sexual violence against boys and men during conflict on the part of some armed organizations.
A distinct version of the continuum thesis which asserts continuity between patterns of conflict-related violence (in general, not specifically sexual violence) and patterns of sexual violence in the post-war period appears to be better supported by recent research. In an analysis that combines household survey data from seventeen countries in sub-Saharan Africa with geo-referenced conflict data, Gudrun Østby shows that conflict intensity in the home region of respondents had a significant effect on the probability that the respondent had suffered IPSV after the conflict.Footnote 33 Whether the mechanism linking conflict-related violence, including sexual violence, with post-war IPSV has a causal effect by increasing the risk factors for victimization or for perpetration (or both) is not well established.Footnote 34 In support of the latter, in South Africa the rate of intimate-partner physical violence by men who have been exposed to political violence is significantly higher compared to the rate by men who had not been, and in the Occupied Palestinian Territories, the rates for physical and sexual intimate-partner violence were both higher on the part of men exposed to political violence than for those not exposed.Footnote 35
An additional, as yet little explored link between conflict-related sexual violence and post-war violence concerns conflict recurrence: periods of peace after civil conflicts with high levels of conflict-related rape are reported to be 3.5 times more likely to end in renewed conflict.Footnote 36
These diverse findings suggest that the relationships between pre-war sexual violence and conflict-related sexual violence, and between conflict-related sexual violence (and violence during conflict generally) and post-war sexual violence, vary across forms of sexual violence and probably across settings as well. The continuum thesis alone cannot account for this variation. Sexual violence by intimate partners, for example, is much better understood as part of such a continuum than is the type of multiple-perpetrator rape carried out by some armed organizations. Opportunistic sexual violence, whether by family members or strangers, should be more easily conceptualized as part of a continuum than the strategic adoption of sexual violence by an organization. In an exemplary analysis of the evolution of patterns of sexual violence in Northern Ireland, Liberia and Timor-Leste, Aisling Swaine found that while some forms of sexual violence began before the conflict and continued during and after the conflict, other forms were innovations during the conflict, with some of those carrying over to the post-conflict period.Footnote 37
Explaining variation in conflict-related sexual violence: Institutions, ideology and the culture of the armed organization
For simplicity of exposition, we will mainly focus on conflict-related rape (referring to other forms of sexual violence when relevant). Because the classic explanations when taken separately or even in relevant combinations do not explain the observed variation in conflict-related rape, as shown above, many scholars now focus on the culture, ideology and institutions of armed organizations.Footnote 38
To field an armed organization, leaders must develop institutions for the enlisting and training of recruits, for organizational cohesion, and for the control of members. To be sure, organizations vary sharply in the degree of development of such institutions, but their survival depends on them. In particular, leaders seek to control the pattern of violence (the repertoire, targeting and frequency of violence) wielded by their combatants, at least to the extent of avoiding the turning of weapons against commanders.Footnote 39 Even when an armed organization appears to embrace the terrorizing of civilians, there are decisions to be made about targeting and timing. In particular, military leaders may make explicit decisions to prohibit, promote or tolerate rape (and against which groups or individuals); if they have not made an explicit decision, they may have to do so if accusations of rape emerge.
However, exerting control over violence is challenging for two reasons. First, combatants in general differ from commanders in their preferences for patterns of violence, where by “preferences” I include reasons for action such as norms, tastes, ethical commitments, emotions, affective ties to others and psychological propensities (e.g. conformity). For example, combatants may prefer to engage in more or less rape than commanders would prefer them to engage in, a contrast that may be particularly sharp when commanders (sincerely) prohibit or order it (of course, combatants differ among themselves in their preferences as well). Second, commanders often do not know what their combatants are doing on the ground – that is, in what pattern of violence they actually engage (as opposed to that ordered). These differences between commanders and combatants – in preferences and information – mean that armed organizations suffer from what social scientists term a “principal–agent” problem. As a result, many scholars currently focus on variation across organizations in the ideologies and institutions through which organizations attempt – to varying degrees – to mitigate or at least manage the tensions between the “principal” (the commander) and the “agents” (the combatants under his or her command).Footnote 40
There are two fundamental origins of the differences in preferences. First, most recruits must be taught to overcome an initial aversion to killing. To forge combatants who are willing to fight, if not on behalf of the organization in the abstract then in defence of their brothers in arms, organizations must reshape combatant preferences to allow the wielding of violence. Most armed organizations do so initially through the induction of combatants into the organization through formal institutions such as boot camp and informal ones such as initiation rituals.Footnote 41 In many State militaries, the powerful experiences of endless drilling, dehumanization and degradation at the hands of the drill sergeant and then “rebirth” as organization members through initiation rituals mould recruits into combatants whose loyalties to the organization may be experienced as stronger than those to family.Footnote 42
Second, combatant preferences may evolve dramatically during active deployment. Both the suffering and wielding of violence often bring profound changes to the combatant's own norms, commitments and tastes concerning violence. The desensitization of combatants to violence, the dehumanizing of victims, the anxiety and uncertainty of combat and the threat of violence – as well as the displacement of responsibility not only onto the organization but also onto the enemy, who “deserve what they get” (blame attribution) – are all powerful wartime processes of moral disengagement that tend to widen the repertoire (possibly including sexual violence), targeting and/or level of violence.Footnote 43 Collective responsibility for atrocities can itself become a source of organization cohesion and a bulwark against betrayal.Footnote 44
In light of the challenges that leaders face in fielding an armed organization, Amelia Hoover GreenFootnote 45 argues that there are two ways in which armed organizations resolve this “commander's dilemma”, as she terms the tension between needing both to produce and to control violence. The first way in which organizations may in principle resolve the commander's dilemma is through institutions that indoctrinate recruits so strongly that they internalize the commander's preferred pattern of violence (and perhaps even the commander's reasons for that choice), a level of indoctrination stronger than the “secondary cohesion” analyzed in the military sociology literature.Footnote 46 Some organizations attract members who are already committed to the organization's ideology, while others attract opportunistic recruits; indoctrination of the former is significantly easier.Footnote 47 In the case that commanders prefer a pattern of limited targeting and a narrow repertoire, the organization must instil an understanding that some forms of violence undermine the organization's purpose through ongoing, intensive political education, argues Hoover Green.Footnote 48 In the ideal case for the commander, combatants thus come to internalize the leadership's choices about violence and to implement them willingly, with no need for discipline. Of course, the leadership of many organizations does not pursue narrow repertoires and limited targeting, and armed organizations inculcate their ideology to highly varying degrees. But some armed organizations – for example, some Marxist organizations that understand conflict as likely to continue over many years or perhaps decades – go to impressive lengths to inculcate their ideology long after the initial training period. Similarly, those State militaries that seek to avoid targeting civilians need strong institutions for the ongoing socialization of soldiers if the psychosocial dynamics of war are not to override the leadership's preference.
The second way is through strong disciplinary institutions: combatants obey orders because they are punished if they do not. In this case, the ability of the organization to enforce the commander's chosen pattern of violence depends on the flow of information concerning the actual patterns wielded on the ground up the chain of command and on the willingness and ability of superiors to hold those below them accountable. Maintaining discipline through the vagaries of combat thus requires the development of strong internal intelligence institutions to ensure the flow of such information. For example, the LTTE insurgency in Sri Lanka deployed a parallel chain of command dedicated to internal intelligence.Footnote 49
If indoctrination of combatants is complete, preferences of superiors and combatants will be consistent, and no unordered violence will occur. If there is a conflict between the preferences of the commander and those of the combatants, but disciplinary and internal intelligence institutions are sufficiently strong, then combatants will follow orders despite their own individual preferences. So in both these cases, if the leadership chooses to promote rape of civilians, for example, combatants will rape with high frequency against the chosen target, and if the leadership chooses to prohibit rape, combatants will not rape (except in isolated instances). In short, if the organization's internal institutions are strong, it is possible to conclude that if sexual violence occurs, it is ordered, except for isolated incidents.
But what happens when the orders of superiors and the behaviour of combatants concerning violence collide? Indeed, often the organization's institutions are not so strong, with the result that the organization is unable to deter or facilitate behaviour that it would rather prevent or promote. In this case, individual and unit norms concerning rape will determine the pattern of rape by combatants. They will also do so both when the organization does not have a policy concerning rape, and also when such a policy exists but individual commanders do not enforce it.Footnote 50 Many organizations appear to formally prohibit sexual violence but do not build the institutions or exert the will to effectively do so – with the result that rape, if it emerges, is neither ordered nor punished but is tolerated, an observation to which I return in the next section.
Several recent works on conflict-related sexual violence confirm this focus on the armed organization. The absence (or presence) of sexual violence against civilians on the part of some organizations reflects their ideologies and institutions.Footnote 51 During El Salvador's civil war, differences in patterns of sexual violence across State forces and insurgent organizations corresponded to different institutions, and when institutions changed, the pattern of violence changed as well.Footnote 52 Distinct branches of the State military (which have distinct institutions) engaged in different patterns of sexual violence during Peru's civil war.Footnote 53 Among armed organizations that develop institutions for military training, those that also develop institutions for reiterated political indoctrination are significantly less likely to engage in high levels of rape, according to tentative findings by Hoover Green.Footnote 54 Hoover Green also tentatively finds that rebel groups who follow communist ideology are less likely to engage in wartime rape.Footnote 55
However, this focus on the armed organization runs the risk of ignoring causes of conflict-related sexual violence stemming from interactions with other organizations, including combat dynamics (the pattern of violence may change with imminent defeat, for example) and diffusion of patterns of violence from organization to organization. And of course, the approach begs the question: from where do internal institutions come? One source is ideology: some ideologies include strong blueprints for institutions and may also proscribe certain forms of violence against certain targets on either strategic or normative grounds.Footnote 56 Armed organizations also copy the institutions of other organizations in an ad hoc fashion.
Between strategic and opportunistic: Rape as a practice
The difference between ordered and unordered violence is often approached in the literature on conflict-related sexual violence through the contrast between “opportunistic” and “strategic”. Again focusing on rape for simplicity of exposition, let's consider “opportunistic rape” to be rape carried out for private reasons rather than organization objectives, and “strategic rape” to be instances of rape purposefully adopted in pursuit of organization objectives. In its extreme form, “strategic rape” is ordered (not necessarily by top commanders).
However, the distinction as used in the literature is often confusing. “Strategic” sometimes appears to be used as a synonym for “massive”, which conflates whether or not violence is carried out for organization purposes with its frequency. The existence of a strategy is sometimes inferred, rather than demonstrated, as when widespread rape is followed by massive flight, and the consequence – flight – is presumed to also be the purpose without supporting evidence.Footnote 57 Similar concerns arise when rape is claimed to be a “weapon”, a “tactic” or a “tool” of war without further evidence that it was in fact purposefully adopted in pursuit of organization objectives. Such simplifying assumptions overlook the distinct mechanisms that contribute to a high incidence of rape.Footnote 58
Moreover, the distinction begs several questions: what should we conclude when commanders consistently fail to punish certain forms of violence despite their being against organization norms and/or rules? Or when combatants are drawn into forms of violence by the exceedingly strong forms of peer pressure present during war, rather than by individual opportunism or their superiors' orders?
The distinction, if it is to be useful, must be supplemented by an intermediate category, that of “practice”.Footnote 59 Violence that is not ordered (even implicitly) but is tolerated by commanders, let us term a “practice”.Footnote 60 A practice differs from opportunistic violence in that it may be the product of social interactions, not individual preferences – for example, the combatant's desire to conform to the behaviour of others in the unit. Such social pressures are very strong during training and combat, as is evident in combatant memoirs as well as the military sociology and history literature.Footnote 61
This set of concepts – strategic, opportunistic, and as a practice – are distinct from the dimensions that comprise a pattern of violence (repertoire, and for each element its frequency and targeting). A practice of rape could be more or less frequent, and targeted more or less narrowly. An organization's pattern of sexual violence could be narrow in terms of its target (a single social group, for example), but with either a high or low frequency, and with either a wide or narrow repertoire. A particular organization may engage in rape both as a practice and as a strategy during the same period, and rape as a practice may be more frequent than rape as a strategy as in the following example: in a village under occupation, the frequency of rape that is tolerated but not ordered by commanders (a practice) may be significantly higher than rape that is ordered against a small fraction of political prisoners as a form of torture (a strategy).Footnote 62
Whether a given pattern of rape is strategic or opportunistic or occurs as a practice may not be readily observable. If an instance or pattern of rape is punished by the chain of command, it is clearly opportunistic (unless it is a “show trial”). Institutionalized forms of sexual violence are clearly adopted for organization purposes and are therefore strategic (see below). Organizations that explicitly order combatants to rape are probably rare (but do exist).Footnote 63 Probably more common are organizations where some form of sexual violence by combatants is a strategy authorized not by explicit orders but by “total war” or other permissive rhetoric.
With these considerations in mind, I will now analyze the conditions under which rape in particular is likely to be a strategy or a practice of war of the armed organization. I will bring in other forms of conflict-related sexual violence as needed.
Rape as a strategy of war
Commanders may adopt rape as a strategy of war against particular populations as in the case of rape as a form of sexual torture of political prisoners, the public rape of members of particular groups as they are “cleansed” from an area, as a form of collective punishment (usually in the context of orders to terrorize civilians), or as a signal of the organization's resolve. In some settings, rape is an institutionalized form of compensation or reward, as when combatants are rewarded for exemplary service with civilians to victimize (or sex slaves, or wives in forced marriage). In such cases, commanders appear to perceive the benefits as outweighing the costs, which include less disciplined troops (who might come to engage in rape in contexts where it is not strategically beneficial), decreased civilian loyalty and cooperation, violation of domestic and international norms, and negative publicity possibly in international as well as domestic media.Footnote 64 Rape – including multiple-perpetrator rape – appears to have been a strategy in, for example, Bosnia, Guatemala and Rwanda: perpetrators were almost never punished, and gang rape occurred in the context of campaigns of ethnic cleansing or genocide (or torture) that were clearly ordered.Footnote 65
Michele Leiby analyzes rape as a counter-insurgency strategy on the part of States engaged in irregular warfare.Footnote 66 She suggests that State forces engage in sexual torture and rape (as well as other forms of violence) where and when rebel forces are visibly active but not strong enough to engage the State in frequent combat, using rape against communities of purported insurgent supporters as well as sexual torture against captured insurgents (and relatives) to extract information but also to punish and terrorize. She shows that sexual violence on the part of State forces during the 1980–2000 civil war in Peru conformed to the pattern predicted by her theory and was thus probably strategic.Footnote 67
Some armed organizations engage in other forms of conflict-related sexual violence as strategies of war. When an organization institutionalizes sexual slavery or forced marriage, the organization has purposefully adopted that form of sexual violence in pursuit of organization objectives, and it is therefore a strategy as defined above. For example, while still in Uganda (before being pushed into neighbouring countries), the Lord's Resistance Army forced many of the girls and women it abducted to marry combatants on terms defined by the organization, which regulated and monitored compliance with its rules.Footnote 68 In mid-2014, the Islamic State (Daesh) reportedly “abducted hundreds (perhaps thousands) of Yezidi men, women and children”,Footnote 69 subjecting many of them to rape and sexual slavery (and some to forced marriage). Forced marriage and sexual slavery are clearly strategic: they are strongly institutionalized within the group, which has issued rules for their implementation.Footnote 70
Rape as a practice of war
When rape occurs as a practice, it is not ordered (even implicitly) or institutionalized, but is tolerated for a variety of reasons. Upper-level commanders may think effective prohibition too costly: it may require the disciplining or dismissal of otherwise effective subordinates; it may divert scarce resources to an issue seen as unimportant; it may lessen the respect of subordinates for their superiors (in a unit dominated by those who see nothing wrong in rape of civilians, the commander who would attempt to prohibit it may be seen as weak) and thereby undermine vertical cohesion; or it may simply be too much trouble. Commanders may tolerate rape or sexual slavery as a form of “compensation” to combatants (see below) if the costs of ending the practice are seen as too high. In short, “too costly” is socially constructed. An individual commander may tolerate rape if it is in his interest (for example, when he himself engages in rape).
The literature has identified at least two contexts in which combatants come to engage in rape as a practice. First, Dara Kay CohenFootnote 71 argues that gang rape reinforces cohesion in organizations that rely on forced recruitment (and thus have to create cohesion among hostile and bewildered recruits). Drawing on the literature on urban and prison gangs, she argues that gang rape effectively builds cohesion because it is an act understood by participants to be uniquely costly, not only breaking local social norms and recruits' ties to their communities and cementing new ones to the organization, but also likely to result in sexually transmitted disease, which might go untreated. Rape in at least some of these cases – for example, the Revolutionary United Front in Sierra Leone, she shows – is not ordered or purposefully adopted by commanders; rather, senior members of small units participate and often insist that all members (including women) do so as well.Footnote 72 Cross-national data confirm that conflict-related rape is more likely on the part of organizations that forcibly recruit,Footnote 73 and interviews with former combatants in Sierra Leone confirm the underlying mechanism.Footnote 74 As Cohen herself emphasizes, gang rape is a deeply social activity and (apparently) for some perpetrators also a sexual one, which begs the question: how in the context of such terror (on the part of the recruit) can rape nonetheless be sexual? What accounts for high levels of gang rape by those organizations that do not forcibly recruit, and low levels by some that do?
The second context of rape as a practice is when it is an unordered form of compensation that is broadly tolerated by commanders (as long as it is not institutionalized). Maria Eriksson Baaz and Maria Stern analyze how soldiers of the DRC State military understand the widespread rape of civilians by the organization.Footnote 75 In the context of deeply inadequate salaries that often go unpaid for extended periods, many of the 200 soldiers interviewed by the authors linked their organization's high rates of rape with the frustration and anxiety occasioned by their failure to live up to masculine ideals of establishing and providing for a family. Soldiers also distinguished (but not sharply, and with some ambivalence) what they saw as “lust” rapes – that is, rape involving forced sexual intercourse born out of frustration – from what they termed “evil” rapes, which included mutilation and gratuitous violence. The former were rapes that were “somehow more ‘ok,’ morally defendable, ethically palatable and socially acceptable, and [the latter were] those that are ‘evil,’ and not acceptable – but still ‘understandable’”.Footnote 76
Implications for policy
In light of recent research on conflict-related sexual violence, particularly its variation in repertoire, targeting and frequency across armed organizations, what are the implications for more effective policy to address it? The question is all the more urgent for the fact that a wide range of political, religious and social actors are implementing policies, many of which are not well informed by recent research. The discussion below lays out principles to guide policy rather than recommendations for specific policies (which must be tailored; see principles 4 and 5).Footnote 77 It focuses on policies to prevent sexual violence by armed organizations (not by civilians), and not on those to address the many needs of victims. While some of these principles apply specifically to conflict-related sexual violence, some may also apply to other types of violence against civilians by armed organizations.
1. The observed variation in conflict-related sexual violence strengthens the case for holding commanders accountable for sexual violence by their combatants if the usual criteria for effective command are met. The demonstrated fact that armed actors can build institutions that inculcate and enforce norms against rape and other forms of sexual violence of civilians if they care to do so should strengthen efforts to hold accountable those who do engage in rape. Commanders exercising effective command should be held responsible, whatever analytical category best describes the organization's pattern of violence, whether strategic (including institutionalized forms of sexual violence), opportunistic or as a practice.Footnote 78
2. Policy-makers and practitioners can learn from those organizations that do not engage in conflict-related sexual violence. In the case of an armed organization seeking to minimize sexual violence by its members, strengthening its institutions for the socialization of combatants against sexual violence, including the reasons for its prohibition (rather than only emphasizing disciplinary institutions), would contribute to its effective prevention. However, it may not be easy to “graft” specific institutions (for example, ongoing training for officers that emphasizes the organization's respect for and dependence on civilians) onto the armed organization if there is little resonance with its existing organizational culture.
3. Policies should be informed by a sophisticated understanding of gender rather than treating conflict-related sexual violence as a women's issue. Footnote 79 In particular, policy-makers should seek to analyze how combatants understand their engagement in sexual violence, exploring, for example, how conceptions of failed or compensatory masculinity may drive the social dynamics of rape of girls and women as a practice.Footnote 80 Moreover, such an understanding would also illuminate the conditions under which an armed organization targets men and boys with sexual torture and rape, those under which female combatants perpetrate sexual violence, and those under which organizations target members of sexual minorities.Footnote 81 Efforts should be made to adopt a gender-neutral definition of rape, such as that used by the International Criminal Court.Footnote 82
4. Policy will be more effective if tailored to the organization's particular pattern of violence, taking into account its repertoire and targeting. Policies designed to address rape are unlikely to address forced abortion; policies to address forced abortion are unlikely to address the sexual torture of men. Moreover, documenting the organization's complete pattern of violence – the repertoire of each sub-organization (including sexual violence) and the targeting of each element of the repertoire – may strengthen efforts to pressure commanders to limit violence against civilians.
5. Policy will be more effective if informed by whether conflict-related sexual violence occurs as a practice, as a strategy, or opportunistically. If rape (or other forms of sexual violence such as institutionalized sexual slavery) occurs as a strategy, then persuading or forcing organization leaders to countermand the strategy may be sufficient to end it. In the case of opportunistic sexual violence or sexual violence as a practice, policy faces the challenge of persuading individual commanders to no longer tolerate practices that are already formally prohibited, and doing so without counterproductive consequences. In both cases, the armed organization's cultural dynamics – its informal forms of initiation, ostracism and punishment – may prove quite resilient to change.
6. Policy-makers can learn from policies that succeed in combating sexual violence in peacetime. Examples include social norms marketing campaigns and some male-to-male peer counselling programmes.Footnote 83 However, such policies may need to be radically adapted to the armed organization's structure, its culture and its particular pattern of sexual violence.
7. Policy-makers can learn from successful campaigns against violence during conflict. As many have pointed out, the campaign against conflict-related rape that began in the 1990s was very successful in that it led to the international criminalization of sexual crimes and the adoption of a series of UN Security Council resolutions.Footnote 84 Yet implementation is at best uneven, and consequences may include unintended ones such as the conditioning of health services to women in conflict areas on a claim to have been raped. What can analysis of the successes and failures of other campaigns (such as those against land mines, child soldiers and “blood diamonds”) teach us about policy design and implementation?
8. Policy-makers should be aware of settings that are at high risk for conflict-related sexual violence. Recent research identifies a number of such settings, where indicative factors include recruitment by abduction or press-ganging,Footnote 85 the torture of detainees (which often takes sexual form), refusal to give International Committee of the Red Cross delegates access to detainees,Footnote 86 the separation of female and male detainees during ethnic violence, and inadequate provisioning of troops, particularly if it makes having a family impossible.Footnote 87
9. After war, policy should be informed by the risk of increased sexual violence but also by the potential for enduring change. Sexual violence may increase after conflictFootnote 88 because norms proscribing it have weakened over the course of the war, because potential victims are denied status in their community and may therefore be further targeted with impunity, or because protective family, religious and gendered networks have disappeared. And if some armed organizations re-mobilize, they may return to their wartime patterns of sexual violence. Nonetheless, the changes wrought by war may make possible more just gender relations, as when women have assumed new roles in the economy and in the leadership of displaced communities, victims' associations and political organizations.Footnote 89
10. Policy-makers should beware the unintended consequences of their efforts,Footnote 90 including an over-emphasis on gathering and publicizing statistics that are inaccurate or that stigmatize victims.Footnote 91 For example, in preliminary results from a project to assess the impact of international prosecution on levels of conflict-related sexual violence, Michael Broache finds that prosecution may not have a deterring effect and, under some conditions, may even be followed by increased levels of conflict-related sexual violence.Footnote 92
Conclusion
Despite the advances summarized above, there is much we still do not understand about conflict-related sexual violence.Footnote 93 Perhaps most troubling is the fundamental question: why is violence sometimes sexual and sometimes not? More specifically, what is the relationship between rape and sexuality? This “uncomfortable subject”Footnote 94 remains the question at the heart of this field. How precisely does gender matter in constructing variations in specific patterns, especially rape? We have seen that broad notions of patriarchy cannot explain the full spectrum of observed variation in rape by armed actors, but surely gender matters deeply. Relatedly, under what conditions does torture include sexual torture, sexual slavery take the form of forced marriage, and targeting include boys and men at high levels as well as girls and women?
As discussed above, many scholars currently focus on the social construction of gender, sexuality and the costs of sexual violence within the armed organization to analyze its pattern of sexual violence, both its specific strategies and particular practices. There remains much that we do not as yet understand about the origins of both strategies and practices. What accounts for commanders' perceptions and beliefs about the strategic utility of different patterns of violence and the institutions to implement them? Do the same psychological mechanisms and dynamics that undergird rape in the context of university campuses, youth gangs, sports clubs and prisons operate in armed organizations that rape as a practice (but not in those that do not)? To what extent do ideologies, institutions, strategies and practices emerge independently, and to what extent do they diffuse from organization to organization through imitation of other organizations (perhaps across different conflicts) or the desertion of combatants from one to another? Is rape as a practice more common during war than as a strategy? Several researchers suggest that the answer is yes, but the claim has not been explicitly explored.Footnote 95
Rape is not inevitable during war.Footnote 96 It is not an unavoidable collateral damage of war – its victims, women and men of all ages, were not brought down by crossfire or an errant missile but were intentionally violated. As Neil Mitchell has emphasized, “rape is not done by mistake”.Footnote 97 Nor is it an inevitable consequence of patriarchy: many armed organizations – non-State actors as well as State militaries – often choose to prohibit rape by their members, and do so effectively.
What is not inevitable can be ended. Policy informed by recent research on conflict-related sexual violence should be better able to prevent or mitigate its occurrence. Policy guidelines of the type sketched above will perhaps contribute to this shared effort.
