Roundtable 10-20 on Rape During Civil War46 min read

What explains why some armed organizations engage in high levels of rape during civil war, while others engage in little? Why is gang rape such a high fraction of rape by organizations that do engage in widespread rape during civil war? What accounts for the participation in the rape of girls and women by female combatants in those organizations?

Dara Kay Cohen addresses these questions in her book.

H-Diplo | ISSF Roundtable, Volume X, No. 20 (2018)

Editors: Thomas Maddux and Diane Labrosse
Web/Production Editor: George Fujii 

Dara Kay Cohen. Rape During Civil War. Ithaca: Cornell University Press, 2016. ISBN: 978-1-5017-0251-8 (hardcover, $89.95); 978-1-5017-0251-8 (paperback, $26.95).

Published on 10 December 2018



© Copyright 2018 The Authors


Introduction by Elisabeth Jean Wood, Yale University

What explains why some armed organizations engage in high levels of rape during civil war, while others engage in little? Why is gang rape such a high fraction of rape by organizations that do engage in widespread rape during civil war? What accounts for the participation in the rape of girls and women by female combatants in those organizations?

Dara Kay Cohen addresses these questions in her book. She argues that members of armed organizations who forcibly and randomly recruit members—abduction in the case of rebel groups or press-ganging in the case of state militaries – struggle to create the social cohesion that is essential to survival in the midst of armed conflict. Members force new recruits to engage in gang rape of civilians, which forges social bonds among units on the ground. Rape and gang rape are not typically or commonly adopted as organizational policies, she argues, but are tolerated by commanders. Female as well as male combatants may be forced to participate. Thus rape, particularly gang rape, can be frequent without being adopted as a strategy, tactic, or weapon of war.

To assess her argument, Cohen draws on hundreds of interviews she conducted in Sierra Leone, Timor-Leste, and El Salvador, many of them with former combatants, as well as a wide range of other available data and sources. She is one of the few researchers who have spoken with combatants about rapes they themselves carried out. The interviews with former members of the rebel force in Sierra Leone (Revolutionary United Forces, RUF) confirm—often powerfully—the role of gang rape in forging social bonds among the rebel force.

Cohen also builds an original cross-national data set on wartime rape. Well aware of the challenges confronting researchers documenting wartime rape, she coded annual reports from three sources into broad categories to capture the frequency (from ‘not reported’ to ‘massive’) and nature (e.g. if reported as systematic, a tactic, or a weapon, 66). Her statistical analysis confirms that armed organizations that forcibly recruit through abduction (rebels) or press-ganging (state militaries) are more likely to engage in frequent rape. She finds that measures for alternative explanations are not significant (with the exception of state failure and contraband for rape by rebel forces). Her argument is also supported by survey data (not her own) that shows that female combatants participated in about a quarter of the incidents of gang rape during Sierra Leone’s civil war (125).

Cohen’s argument, which she does not intend to cover all cases, thus accounts for a significant part of the documented variation in the level of wartime rape across armed organizations. It also explains some specific patterns on the part of armed organizations that do engage in frequent rape: gang rape is a much higher fraction of rape during war than peacetime; women sometimes participate; the frequency of rape by different parties to a civil war are sometimes sharply different (see below). Moreover, her work uncovered patterns hitherto neglected: that state militaries are more likely to engage in widespread rape than rebel organizations, and that ethnic civil wars are not more likely to have a high prevalence of rape than non-ethnic conflicts.

This roundtable’s three reviewers concur that Cohen’s book is path-breaking. Laura Sjoberg lauds the clarity of the theoretical puzzle it addresses, its “striking” originality, and the impressive field research and data collection on which it rests. Valerie Hudson catalogues the “justified” prizes it has won. Amanda Blair describes it as “exceptional” and “transforming the field.”

The reviewers also raise two principal concerns. Both Blair and Sjoberg find that Cohen does not provide a compelling argument about why forcibly recruited combatants engage in specifically sexual violence to build cohesion, something Cohen acknowledges in the book (31). Sjoberg suggests that the problem stems from Cohen’s not having engaged enough with feminist theory on gender and violence in order to analyze “what is happening” during rape, what rapes “really are.” (Sjoberg herself addresses the “work that rape does” in her excellent book Women as Wartime Rapists (New York University Press, 2016)). Sjoberg also points out that Cohen’s argument does not account for some well-known cases of wartime rape such as Rwanda, where rape was part of the state’s genocidal repertoire and was very likely ordered and certainly promoted by commanders of state forces and allied militias.

Hudson argues that Cohen’s finding that the level of gender inequality in a country enduring civil conflict does not predict the level of wartime rape—a surprise to many scholars and advocates—reflects her use of incomplete measures of gender inequality. Drawing on her own new measure of women’s disempowerment within the household, Hudson show that a higher percentage of countries with high levels of women’s disempowerment also have high levels of wartime rape than countries with medium or low levels of women’s disempowerment. However, she does not carry out a multivariate regression similar to Cohen’s analysis. Hudson concurs with Cohen’s acknowledgement that some other factor may drive both forced recruitment and wartime rape and suggests that it may be strong norms of gender inequality that do so.

In her response, Cohen states that that of course gender inequality matters and readily acknowledges that commonly used measures, including those she uses, are imperfect proxies for gender inequality. Scholars face a trade-off between using those proxies, which cover most countries over the relevant time period (1980-2012), and new indices such as Hudson’s that do not go as far back as 1980, which limits their ability to explain violence that varies sharply over time, such as wartime rape.

However, she argues, the “most interesting and consequential” variation is not at the country level. Rather, in civil wars with high levels of rape, the parties to the war often engage in sharply different levels of rape. Thus society-wide features such as country-level gender inequality cannot explain variation across armed organizations; researchers, therefore, should focus on the organizations’ characteristics. For example, only one of several organizations, namely, the RUF, was responsible for much of the widespread rape and gang rape in Sierra Leone’s civil war. She also suggests that local social norms concerning gender also do not account for the variation across the parties to the war.

In response to her argument’s alleged lacuna concerning why there is specifically sexual violence to build social cohesion, Cohen notes that she draws on the literature in criminology and psychology to strengthen her theory. And she refers to her argument in the book that gang rape is a particularly costly activity (the combatants she interviewed knew quite a lot about sexually transmitted infections and their consequences, for example), that therefore creates social bonds among participants.

More fundamentally, Cohen concurs with Sjoberg that empirical researchers and feminist theorists would strongly benefit by engaging across their distinct epistemologies and questions, in particular, what counts as an explanation (what and why, in contrast to how). She closes her response by noting that gender inequality is a necessary but not sufficient condition for wartime rape.

The roundtable demonstrates the importance of the book’s theory and findings, as well as its power to set the agenda for researchers working on sexual violence, war, and socialization. Cohen’s book does not account for all aspects of wartime sexual violence, to be sure, but by laying out an original theory, mapping patterns of reported rape, and analyzing both qualitatively and statistically original data painstakingly gathered, she has advanced our understanding of rape during war to a remarkable degree.


Dara Kay Cohen is a Ford Foundation Associate Professor of Public Policy at the John F. Kennedy School of Government at Harvard University. Her first book, Rape During Civil War (Cornell University Press, 2016), examines the variation in the use of rape during recent civil conflicts; the research for the book draws on extensive fieldwork in Sierra Leone, Timor-Leste and El Salvador. The book received the 2017 Theodore J. Lowi First Book Award from the American Political Science Association, the 2018 Best Book Awards from the International Security Studies Section (ISSS) and the Feminist Theory and Gender Studies (FTGS) Section of the International Studies Association. Her research has appeared or is forthcoming in the American Political Science Review, World Politics, Quarterly Journal of Political Science, Journal of Conflict Resolution, Journal of Peace Research, International Security, and Stanford Law Review. Her current project is focused on the intersection of political violence, public opinion and gender in Port-au-Prince, Haiti. Cohen received her Ph.D. in political science from Stanford University and an A.B. in political science and philosophy from Brown University.

Elisabeth Jean Wood is the Crosby Professor of the Human Environment and Professor of Political Science, International and Area Studies at Yale University. She is the author of Forging Democracy from Below: Insurgent Transitions in South Africa and El Salvador (Cambridge University Press, 2000) and Insurgent Collective Action and Civil War in El Salvador (Cambridge University Press, 2003), and co-editor with Morten Bergsmo and Alf B. Skre of Understanding and Proving International Sex Crimes (Oslo: Torkel Opsahl Academic Epublisher, 2012). She is currently writing a book on wartime sexual violence.

Amanda H. Blair is a Postdoctoral Teaching Fellow in the Social Sciences at the University of Chicago. Her research focuses on the intersections between violence, armed conflict, and gender, particularly in the context of Central and East Africa. In her current project, she examines how armed conflicts and protracted insecurity have contributed to the development of sexual economies and sex trafficking networks, and as a result, have transformed everyday relations of sexual exchange, intimacy, and reproduction. Amanda has published and presented her research on sexual violence in armed conflict in a variety of forums. She received her doctorate in Political Science from the University of Chicago in 2017, and holds a master’s in political science from the University of Chicago and a bachelor’s in political science from the University of Wyoming.

Valerie M. Hudson is Professor and George H.W. Bush Chair in the Department of International Affairs at The Bush School of Government and Public Service at Texas A&M University, where she directs the Program on Women, Peace, and Security. Hudson is co-PI of The WomanStats Project, and co-author of Sex and World Peace and The Hillary Doctrine. Her current Minerva-funded research project is entitled “The First Political Order.”

Laura Sjoberg is Associate Professor of Political Science at the University of Florida. Dr. Sjoberg’s work has been published in more than three dozen journals, and she is author or editor of ten books, including, most recently, Beyond Mothers, Monsters, Whores (with Caron Gentry, Zed Books, 2015) and Women as Wartime Rapists (NYU Press, 2016).


Review by Amanda H. Blair, University of Chicago

In this meticulously researched and well-argued book, Dara Kay Cohen examines why armed groups use rape during civil war. Well-entrenched, existing theories suggest that armed groups use rape either opportunistically—for example, as a reward for participation—or strategically, as part of a top-down plan to accomplish broader group goals.[1] But as Cohen points out, these theories oftentimes rely on the problematic, yet sticky, assumption that rape is ubiquitous during war. This is despite the fact that scholars have repeatedly shown that the use of rape varies across and within armed conflicts, which Cohen also provides evidence for in this book.[2]

After tending to existing theories and assumptions, Cohen suggests an alternative logic to explain rape during civil war, one that is generalizable across contexts. According to her theory, what she calls “combatant socialization,” armed groups that rely on forced recruitment are more likely to use rape during war than groups that rely on voluntary mobilization (3). The reason being that groups that rely on forced recruitment methods, such as press-ganging or abduction, have lower levels of cohesion and use group violence to create cohesion. Unlike groups that voluntarily recruit members, armed groups with low levels of cohesion tend to lack preexisting ideological and social ties that create bonds of loyalty, esteem, and trust among members of the group. Using “rape—especially gang rape, or rape by multiple perpetrators,” Cohen argues, “enables armed groups with forcibly recruited fighters to create bonds of loyalty and esteem from initial circumstances of fear and mistrust” (2). Drawing from literature across the social sciences, Cohen suggests that group violence creates cohesion in a number of ways: it both severs individuals’ ties with their community, immersing them into the norms and structure of an armed group, and reinforces intragroup bonds when individuals recount violence to other group members.

In order to test her theory, Cohen combines statistical analysis of original cross-national data on rape perpetrated during civil wars from 1980 to 2012, with three comparative case studies. She selects cases based on two types of variation: group responsible for perpetrating the majority of rape (state or non-state) and level of rape reported (mass or low). The cases include Sierra Leone (non-state/mass), Timor-Leste (state/mass), and El Salvador (low), and draw from an array of data and primary sources, including 270 interviews Cohen conducted (the large majority in Sierra Leone) with armed combatants, civil society members, local and national government representatives, researchers, and diplomats. (102, 142, 174) In addition to testing her theory of combatant socialization, Cohen outlines a number of alternative hypotheses that permeate the existing literature: opportunism/greed, ethnic hatred, and gender inequality. For example, in so-called ‘ethnic’ conflicts or in countries with high levels of gender inequality, these theories predict that we will observe higher levels of sexual violence. An excellent summary of the hypotheses at the macro and micro levels can be found on pages 54-55.

Generally, Cohen finds support for the combatant socialization and opportunism/greed hypotheses, but not for the other two. The statistical analysis, for example, shows that both forced recruitment methods and state collapse/failure are correlated with higher levels of wartime rape, lending support to the combatant socialization hypothesis and a strand of the opportunism/greed hypothesis. And the cases of Sierra Leone and Timor-Leste similarly provide support for the combatant socialization hypothesis and variants of the opportunism/greed hypothesis. But the case of El Salvador intentionally poses some interesting questions about the limitations of her theory. In El Salvador, neither the state forces nor the primary rebel group (Farabundo Martí National Liberation Front) used high levels of rape, despite having both relied on forced recruitment methods at various times. Cohen suggests that the case of El Salvador “illuminates the power of intensive external and internal pressures to curb such violence.” (189) On the one hand, the Salvadoran state forces faced external pressure from the United States to limit its abuses against civilians, including rape, and on the other, the FMLN not only capitalized on the fear of/protection from sexual violence to recruit new members, but also paired these norms against sexual violence with severe penalties, including execution. Meaning, a group’s use of sexual violence may be reduced—even when it relies on forced recruitment methods—when there are strong norms, pressures, and/or penalties for using sexual violence.

While this book greatly exceeds expectations in some regards, in comes up slightly short in others. Scholars and policymakers who are interested in group violence—as opposed to individual or organizational violence—will walk away with more than they anticipated; whereas, those who are interested in why sexual violence, in particular, is used during war, may leave the book feeling a bit underwhelmed. That being said, this book offers compelling insights into some of the most enduring questions regarding violence, armed conflict, and gender, and is an impeccable example of mixed-methods social science research.

Turning first to the strengths, it must be said that Cohen’s contributions to the literature and data on sexual violence are exceptional. The book is truly an interdisciplinary effort as it engages literature across the social sciences, and the rich, original quantitative and qualitative data Cohen offers transforms the field as we know it. Moreover, Cohen’s theoretical contribution—group violence as a mechanism for combatant socialization—bolsters a burgeoning area of scholarship on wartime sexual violence that seeks to move beyond the dichotomy of opportunistic, individual-level violence versus strategic, organizational-level violence, to focus on group-level violence and processes of socialization.

Cohen’s theory of combatant socialization also offers many fruitful directions for future research. While the book focuses primarily on the intergroup use of violence, we know sexual violence is also used internally, within a group, not only in cases of forced marriage and sexual slavery, but also against fellow service members, including in groups with high levels of cohesion, such as the United States Armed Forces.[3] With Cohen having set up the foundation for theorizing group violence, it would be fruitful for scholars to examine the types of ‘work’—in terms of cohesion, norms, structure, etc.—intergroup violence does compared to intragroup violence.

An additional direction for future research revolves around the ambiguity associated with the causality or logic behind combatant socialization. For example, Cohen writes, “gang rape is committed with the (likely subconscious) goal of creating social bonds among members of an armed group” (36). While I do not think that this creates any problems for Cohen’s theory, I would like to see her (and others) lean into the ambiguity of the subconscious, bottom-up aspects of combatant socialization in order to flesh out the implications of this theory for existing theories of human action, particularly theories of violence that all-too-often rely on ideas that violence is rationally, consciously, and/or intentionally used towards particular end(s).

Now on to a weakness. After refining the scope of the argument, Cohen examines why (male and female) armed groups use (gang) rape (against female non-combatants) during civil wars (from 1980-2012). But I take the book to be primarily about how group violence creates cohesion, in which gang rape serves as a lens to explicate the theory. As such, the book overstates its claims about why armed groups use rape, gang rape, sexual violence, and/or sexualized violence, in particular, during war. Cohen is aware of this problem when discussing existing literature, as she writes, “Why sexual violence is chosen thus remains uncertain,” though she does not go so far as to address this issue in the framing of the text (31). To me, this is the primary weakness of the text, as the reader expects an exploration of why rape, in particular, is used during war, but instead reads a text that examines how armed groups use violence, including gang rape, to create cohesion.

Beyond this limitation, I would have liked to see Cohen compare across the case studies, nothing too extensive as the book is already quite rich, but a cross-case comparison may yield additional insights. One cannot help but notice in both cases with mass rape—Sierra Leone and Timor-Leste—that the largest spikes in sexual violence came towards the end of the war. One wonders whether this can be observed in other cases. Do other factors contribute to low cohesion other than recruitment method that could account for the use of gang rape? The book does an excellent job testing alternative explanations—opportunism/greed, ethnic hatred, and gender inequality—in the statistical analysis and within each case, but I think Cohen missed an opportunity to compare across the cases.

All in all, it should come as no surprise that this book won the 2017 Theodore J. Lowi Award for “Best First Book” written in any field in Political Science. It examines one of the most pressing questions facing scholars and policymakers today, and offers interesting insights into why we see higher levels of sexual violence used by some groups and not others, as well as speaking to approaches to mitigate the use of sexual violence during armed conflict.



Review by Valerie M. Hudson, Texas A&M University

Dara Kay Cohen’s book Rape During Civil War has won numerous awards, such as the 2017 Theodore J. Lowi First Book Award from the American Political Science Association, the 2018 Best Book Awards from the International Security Studies Section (ISSS) and the Feminist Theory and Gender Studies (FTGS) Section of the International Studies Association. Such accolades are justified. Cohen brings a careful mixed-methods design to an important theoretical (and policy relevant) issue. The book retains its dissertation-linked attention to issues of reverse causality, alternative hypotheses, perturbing variables, etc., which is refreshing, as is the construction of her original dataset on rape in armed conflict. For these reasons, it might well be used in an introductory research design course, in addition to courses related to the subject matter. (And Cohen’s field fortitude is noteworthy; see, for example, Chapter 3, note 7 and Chapter 4, note 45).

Cohen’s primary argument is that mass rape/gang rape is used where soldiers are ‘press-ganged’ into service, for such a recruitment strategy creates non-cohesive units that cannot be relied upon to fight effectively. Gang rape serves, then, as a means to create group cohesion through group public performance of taboo acts associated with virility. Cohen is quick to observe that this is not the only explanation of the phenomenon, but it is an important one with clear policy implications. She further notes that she can find no linkage between the use of mass rape/gang rape and greater fighting unit effectiveness, which an important finding in and of itself.

Cohen is right to explain that sexual violence is preferred because “this is the form of violence which most clearly communicates masculinization and feminization” (quoting Inger Skjelsbaek; 31).[4] This makes it all the more puzzling for Cohen to conclude, as she does, that gender inequality is not empirically supported as an explanation for mass rape/gang rape in civil war. After all, if gang rape is chosen precisely because of its gendered meaning, surely part of that meaning is that inequality between the sexes is somehow ‘right.’ Since Cohen also asserts that one must be “trained” to overcome “innate reluctance” to rape through means such as “desensitization,” “moral disengagement,” etc. (40), being raised and/or living in a culture when levels of gender inequality are high would presumably provide just such training. On what grounds, then, does Cohen rule out the ‘gender inequality’ hypothesis?

Cohen probes that hypothesis in aggregate statistical analysis as well as in the case-study portion of the design. In the three cases examined, Cohen looks for evidence that the group had a significant proportion of women in its ranks and that it recruited women, trained them, and armed them. However, I think a conceptual case could be made that these indicators may not be sufficient; that is, these might be indicating exploitation of a subordinated group groomed to be submissive and obedient to male authority, and not gender equality. A more nuanced set of indicators—such as proportion of women in positions of decision-making authority within the group—may be necessary to supplement Cohen’s set of indicators in order to address this alternative explanation more straightforwardly.

At the macro level of analysis, Cohen chooses the country’s fertility rate as the indicator of gender inequality, and in the multivariate probit analysis, fertility rate proves not to be significant. But fertility rate may not exhibit a good one-to-one mapping with gender inequality; consider that the fertility rate of contemporary Iran is sub-replacement (1.68 in 2015). What if we used a more comprehensive multivariate indicator of gender inequality? In recent research, my co-authors and I have used an eleven-variable integrated index of women’s disempowerment at the household level, examining phenomena such as age of marriage, patrilocality, polygyny, cousin marriage, property rights for women, brideprice/dowry, inequitable family law, etc.[5] If we divided Cohen’s data into the two subsets used in her analysis (those civil wars coded 0 or 1, and those coded 2 or 3), and trichotomized our eleven-variable index (low, medium, and high levels of household disempowerment), what would we see in cross-tabulation?

Observed frequencies
Low   Medium   High Women’s Household Disempowerment
  0       1      2             Total
Rape 0 or 1 3 15 14 32
Rape 2 or 3 1 10 48 59
Total 4 25 62 91


The chi-square result for this cross-tab clocks in at the .001 level of significance. Rape levels of 2 or 3 (widespread and widespread/systematic rape) are far more likely to be found in countries where women’s household level of disempowerment is high. While I do not have Cohen’s complete set of independent variables to work with, it would be very interesting to include this eleven-variable index or one like it in the multivariate modeling exercise and examine the results. It is also noteworthy that Timor-Leste and Sierra Leone—where Cohen found high levels of rape—also have high levels of women’s disempowerment at the household level, while El Salvador, Cohen’s case lacking mass rape, just barely reaches the medium level of household disempowerment.

Cohen has convinced me through her admirably creative and careful research that press-ganging is a risk factor for high levels of rape in civil war.[6] But she also suggests (and I agree with her) that “there are other factors—such as norms—that may be causing both forced recruitment and higher levels of rape” (95). Given the recent findings of Erin Bjarnegard, Erik Melander, and their co-authors about how gender unequal attitudes are significantly associated with hostility towards out-groups (minorities and foreigners) as well as actual engagement in political violence, it may be that norms of strong gender inequality within a society may predispose armed groups to both forced recruitment as well as higher levels of rape in civil war.[7] This stance would not undercut Cohen’s findings in the least, but would contextualize them in a way that might have both theoretical and policy significance.



Review by Laura Sjoberg, University of Florida

Logics of Rape (?) and Gender

Dara Kay Cohen’s Rape During Civil War is a meticulous and impressively clear study of an important and timely topic. Inspired by a curiosity about why some armed groups engage in ‘conflict sexual violence’ more frequently than others,[8] Cohen’s multi-method study includes both quantitative work using an original cross-national dataset on rape during civil wars between 1980 and 2002 and qualitative field interviews in Sierra Leone, El Salvador, and Timor-Leste. In her qualitative case selection, Cohen was careful to vary type of war, length of war, and levels of conflict sexual violence. The description of the construction and use of the quantitative dataset appears to be just as rigorous, and the book presents the sources of the data, the coding schema, the coding logic, and data limitations. In my view, the presentation of the choice and use of methods in this book should serve as a model for others – it is clear, consistent, and modest.

The presentation of the empirical and theoretical puzzle in the book is equally clear. Cohen is interested in why rape during civil wars varies in frequency and type, and finds previous accounts of the causes of wartime rape inadequate to the cause of explaining this variation.[9] Cohen tests arguments about opportunism and greed, ethnic hatred, and gender inequality, suggesting that none of them account for actual patterns of rape during civil war. Instead, Cohen argues that understanding war rape as a tool of combatant socialization accounts for differences in the level and type of rape in armed groups across a wide variety of civil wars. She predicts, then, that rape during civil wars will be more prevalent when there are high degrees of forced recruiting, and that gang rape will be the most prevalent type of rape in civil wars. She concludes that: “one essential, but previously overlooked, factor is the level of internal cohesion within an armed group: groups that struggle with cohesion are more likely to perpetrate rape … than those that do not.” (191)

The argument is justified empirically by impressive field research and data collection; it is justified theoretically by significant and impressive reference to literatures as diverse as military sociology, criminology, psychology, and politics. The theoretical argument proceeds in Chapter 2 as follows: a characterization of wartime rape as group violence, an account of the need for socialization into armed groups when those groups recruit forcibly, an explanation of how group violence (particularly gang rape) can improve social cohesion, an account of when and why sexual violence is chosen over other cohesion mechanisms, and an exploration of how trends of rape in civil war begin. Across several readings of Rape During Civil War, I found very few quibbles with the theoretical or empirical accounts of rape during civil war.

Still, I do not think that I could come to the same conclusions that Cohen does, even on the same empirical evidence. My point of departure from the book is not with what it says, it is what it does not say, and what it does not explore. While Rape During Civil War includes several theoretical and empirical nods to feminist work on conflict sexual violence, it is not (and does not claim to be) a work of feminist theory. This is a choice, and a valid one, made by the author—one which, at many points in the text, does not take away from either the empirical or theoretical contributions that the book makes.

The place that the lack of investment in feminist theorizing manifests the most clearly, however, is in the section that discusses why sexual violence is the tool that armed groups use to gain cohesion (28-31). Cohen cites research showing that participants in gang rape “gain and maintain social status” (28), that demonstrates “the connection between social cohesion and sexual violence empirically” (29), and that characterizes “rape as a bonding experience among perpetrators” (30). Still, Cohen admits that “precisely how cohesion is created through rape is the source of some debate” (30). The analysis that follows cites the possibility that the mechanism for the creation of cohesion is norms of masculinity, and/or the desire to achieve recognition of one’s masculinity (30), as well as the feminizing aspects of rape (31). Cohen ultimately comments that “why sexual violence is chosen remains uncertain” (31). Cohen finds sufficient conditions for wartime sexual violence: the need for cohesion, the lack of resources to fund more expensive avenues for cohesion, and the lack of a command structure sufficient itself to drive cohesion. But, despite the title of chapter 1, the book does not get to the logic of wartime rape. It provides the conditions of variation of wartime rape, but it never really explains what is happening when wartime rape is committed, and how those seeking cohesion come to use wartime rape to the exclusion of (or in preference to) other potential mechanisms.

If I were looking at Cohen’s puzzle and the data that is analyzed in Rape During Civil War, I could not satisfy myself with a mention of masculinity and feminization that led to an indeterminate conclusion about the dynamics that make gang rape a mechanism of socialization in armed groups. To Cohen’s credit, she finds this argument in a wide variety of places, from psychology to criminology. But the logic behind it—what the rapes really are—is thinly presented. To me, the question of why rape is and has to be is bound up in gender analysis. Feminist work has explored the performance and signification of conflict sexual violence in a wide variety of ways. Cynthia Enloe and others have looked at the ways in which sexual images are used by military organizations to ‘prepare’ soldiers for battle because those organizations associate (individual) sexual arousal with (individual) bravery and energy in combat.[10] Aaron Belkin and others have demonstrated that homoeroticism is not incidental to practices of militarism, but a necessary and always-present constitutive other to hegemonic military masculinities—such that soldier homoeroticism itself may be a crucial source of cohesion.[11] Spike Peterson has argued that conflict sexual violence is an expression of feminization (and therefor devalorization and dehumanization) not only of the victim (male or female) but also of the state, nation, or group that the victim is understood to represent.[12] Annica Kronsell has discussed sexual violence by male members of a particular military group towards female members of the same group.[13]

In this short review, I do not have the space to unpack these or other gender-based readings of conflict sexual violence. That said, even this brief sampling suggests some answers to the question of why sexual violence that might include both more thorough logic than Cohen presents and potential interventions in/with the explanations about military cohesion in Rape during Civil War. First, this work suggests that identifying the victims of conflict sexual violence might be necessary to fully understanding why that sexual violence is committed. Second, it suggests that there is a deeper relationship between militarization and sexualization than can be accounted for by a desexualized understanding of what cohesion is. Third, the work on intra-military sexual violence suggests that if rape during civil war serves a cohesive function, the cohesion is partial and masculinized.

Above and beyond these confounding issues, Rape in Civil War leaves out that rape in civil war can be and has been an end in itself. While Cohen finds few central commands directing rape, the widespread nature of rape also is not fully disaggregated. Rape in civil war can be and has been a tool of ethnic cleansing and genocide. For example, in the ‘Rwandan Civil War’ there was a low level of forced recruiting and a very high level of conflict sexual violence. While I will not rule out the possibility that conflict sexual violence had a cohesion function for perpetrators, conversations with perpetrators and victims (as well as the relevant literature) suggests that, in Rwanda (as I suspect it is in many other places), rape was a tool of genocide—punishment for Tutsi women for their perceived privilege, an assertion of ethnic superiority, a tool of emasculation for (watching) family members, a weapon to force mixed-ethnicity pregnancies, and a weapon for killing (either directly or with HIV/AIDS) the ethnicity targeted for extermination. Still, the answer both to questions about variation in frequency and about why conflict sexual violence is ‘chosen’ over other means of improving social cohesion, it seems to me, must address rape as a (gendered) tool of genocide, whether by central command or not.

In sum, I think a lot of this book. It is incredibly well-written. It is strikingly original. It is impressively humble. It is the product of an amazing amount of data collection and field research. It is impressively versatile across a wide variety of literatures. It stands on its own. That said, I think a gender analysis seriously applied to the work in Rape during Civil War could open it up to more: more about the times cohesion takes a back seat to genderings, more about the genderings of cohesion, more about why conflict sexual violence is the chosen tool of cohesion, and more about variation in conflict sexual violence.



Author’s Response by Dara Kay Cohen, Harvard University

I am grateful to Amanda Blair, Valerie Hudson, and Laura Sjoberg for their very generous reviews of my book, and for their thoughtful critiques of the argument and analysis. Thank you also to Elisabeth Wood for writing the introduction to this exchange. I am a great admirer of all four of these scholars, and I appreciate the opportunity to engage with them on issues of gender and violence, about which we all care deeply. In this response, I begin with a brief summary of the main argument and my goals for this book. I then turn to two main questions that are raised in the three reviews: Why do some but not all armed organizations engage in wartime rape? And what is the role that gender inequality may play in explanations for wartime rape?

Argument in Brief

In the book, I seek to explain variation in the use of rape during wartime, particularly recent civil wars. In order to explain variation, I focus mainly on the armed group as the most important unit of analysis, rather than the country or the conflict (I say more about this below). Beginning with the puzzling pattern that the most commonly reported form of wartime rape is gang rape, I ultimately argue that one important and overlooked factor is the level of intragroup cohesion in an armed group. Armed groups with low cohesion, such as those that randomly recruit their fighters by force, can participate in gang rape as one method of increasing bonds between fighters, a process I call “combatant socialization” (2). I argue that gang rape as combatant socialization does not imply an organizational policy of rape on the part of the armed group, but rather a tolerance for the practice by the command.[14] Indeed, one of the most important conclusions from the book is that rape, even on a massive scale, does not imply that it was ordered by a commander as a ‘weapon of war.’ In the statistical analyses in the book, I show that abduction and pressganging (its equivalent by state groups) is associated with increased reports of rape. I also trace the linkages in the argument—between abduction and pressganging to low cohesion, and then low cohesion to rape—in three post-conflict case studies of state and rebel forces in Sierra Leone, Timor-Leste and El Salvador.

The book builds on and is indebted to feminist and gender studies scholars who have long studied questions of wartime sexual violence. In addition, I test some of the most powerful conventional wisdoms about the use of rape in wartime, including, for example, the assumption that it is commonly a tool of ethnic cleansing or genocide. But the book is also a reaction to the fact that some of these arguments cannot help us to develop a general understanding of why rape happens during civil war. For example, and simply put: ethnic hatred is too frequent and genocide is too rare to serve as generalizable theories for the occurrence of rape.[15]

The book is not, as Sjoberg writes, an exhaustive accounting of all of the ways rape is and can be used, including as a tool of genocide. As Blair notes, the book is primarily an argument about how groups engage in (one form of sexual) violence to create cohesion; since the publication of the book, I have extended this argument in an article about armed groups in the Democratic Republic of the Congo, and consider some of the factors—such as the age of the recruits—that might make mitigate the process of violent socialization.[16]

Why sexual violence?

Both Blair and Sjoberg focus on the question of why sexual violence is used by armed groups, and both cite the same line from a section of Chapter 1 called “Why Sexual Violence?” (28-31) where I write “why sexual violence is chosen remains uncertain.” As I note in the beginning of that section, “Reasons for selecting rape over other, non-sexual forms of violence are debated in the literature and findings from a variety of fields are suggestive—but not conclusive” (28). While Blair argues that the puzzle of why rape in particular is selected is left unclear in the theoretical section, Sjoberg writes that the logic of “what is happening when wartime rape is committed” remains unexplained.

These are justified critiques. The problem of why sexual violence is a vexing one, and I ultimately relied mainly on research about perpetrators of and motivations for gang rape from psychology and criminology. These include, for example, the experimental studies of “precarious manhood” by Joseph Vandello, et al., which find that threats to masculinity can provoke public demonstrations of aggressive forms of masculinity.[17] Applying this research to the contexts I study helps to resolve the puzzle of why gang rape increases during wartime: as Hudson notes, it is because gang rape is a (1) public performance of (2) a form of group violence (3) that is associated with a particular form of masculinity. As I discuss in the book, rape is also a particularly costly form of violence from the perspective of both the armed group and the individual. For instance, rape can spread sexually transmitted infections that often remain untreated, rendering individual combatants unable to fight. The costs associated with rape may increase its utility as a tool of violent socialization.

Sjoberg is right to point out that feminist and gender studies scholars such as Cynthia Enloe, Aaron Belkin, and Spike Peterson offer additional insight into the dynamics of both sexualized violence and the type of masculinized cohesion that is produced through such violence. Whether these scholars’ particular arguments may be used to explain variation in wartime rape, however, is debatable. For example, as Wood has argued, armed groups that are not reported to have perpetrated frequent rape often draw on the same forms of militarized masculinity about which Enloe writes.[18] In addition, Peterson’s argument about dehumanizing the nation, state, or group of the victim may not apply in a non-ethnic civil war context, where perpetrators and victims share many of these attributes.

Because much of the feminist work on wartime rape is also interpretivist/post-positivist, the “lack of investment in feminist theorizing” as Sjoberg notes, reflects the fact that my book is a work of a positivist/empiricist scholarship. As Sjoberg, Kelly Kadera, and Cameron Thies demonstrate, these different epistemologies not only tend to use distinct methods, but also ask fundamentally different questions.[19] For example, instead of asking “what happened and why” some armed groups engage in rape, as I do in my book, many of the feminist/post-positivist works instead explore how, or “the constitution and discursive treatment of the research subject.”[20] That said, and as Sjoberg, Kadera, and Thies convincingly argue, I strongly agree that scholars can make more progress through engaging more readily with others’ research that is not in the same epistemological tradition, enriching both theories and research designs.

Does gender inequality really not “matter?”

Hudson’s comments are centered on the non-significance of the findings about gender inequality from the cross-national regressions in Chapter 3 of the book. In this chapter, I find that for the universe of cases I study—major civil wars between 1980-2012—and after controlling for other factors correlated with civilian victimization, the relative level of gender inequality is not associated with increased reports of rape.

These results do not mean that gender inequality does not matter when we are seeking to explain the occurrence of rape. It would be nonsensical to make such a claim, if for no other reasons than rape—unlike nearly every other form of wartime victimization—is a form of wartime violence with majority female victims. It is patently clear to all scholars who study rape—in wartime and during peace—that norms about gender, sexuality, power, and violence are quite important to understanding the question of why rape.

However, the findings from my book show that given that a war has begun, the relative level of gender inequality in a particular country (the unit of analysis at which proxy measures are readily available) does not help us distinguish between countries that have wars with high levels of rape and countries that have wars with low levels of rape. In other words, gender inequality in conflict-affected countries is too common to serve as the main reason why rape is massive in some cases but more limited in others.

Hudson correctly argues that widely-used country-level indicators of gender inequality tend to be poor proxy measures, including the three standard measures I employ in the book: fertility rates, female labor force participation, and women’s political, social, and economic rights. Scholars like Kara Ellerby are sharply critical of these standard measures; as Ellerby asks about the use of fertility rates as a proxy measure, “Are women with fewer children equal to men?”[21] Hudson instead advocates for the use of more sophisticated measures that can better capture individual women’s lived experiences—including her new index of women’s equality at the household level. One issue with more sophisticated measures of gender inequality, such as the OECD Development Centre’s Social Institutions and Gender Index (SIGI), is they are rarely available at the level of aggregation needed for statistical analyses (in this case, country-year) or for the necessary period of time (in this case, starting in 1980). In sum, I concur that as the scholarship continues to develop, and better and more detailed proxy measures are created, it will be worthwhile to reexamine regression results using these new—and hopefully more accurate—variables.

More broadly, however, I would argue that the most interesting and consequential puzzles related to gender inequality are not on the level of the country at all. Consider the country of Sierra Leone, where I spent months doing fieldwork for the book. It is regularly ranked as one of the “worst countries to be a woman,” due to low female literacy, low rates of female primary schooling, and high maternal mortality rates.[22] Sierra Leone’s civil war was marked by mass rape, and is coded as a ‘mass rape’ war in every study of the phenomenon. But this dubious distinction obscures some important facts about how rape actually occurred on the ground. The vast majority of the rape was perpetrated by only one of the numerous active armed groups. We know from others’ studies (especially from foundational work by Macartan Humphreys and Jeremy Weinstein[23]) that these groups are remarkably similar in terms of the demographic backgrounds of the fighters: ethnicity, party affiliation, education, age, previous occupations, and religion. The fighters all lived in the same highly gender unequal society. So how can we explain why the RUF committed rape on a mass scale but the other groups did not?

I sought to assess arguments about norms regarding gender inequality driving rape during wartime. However, rigorously testing these arguments—defining the variables, selecting the appropriate units of analysis, seeking variation across relevant units—is very challenging. One factor I considered during my fieldwork in Sierra Leone (which did not make it into the book) was local gender norms and beliefs linked to the ethnic identity of the fighters; these are type of norms that Hudson suggests (and I agree) deserve greater scrutiny. The Mende ethnic group, located primarily in the Eastern and Southern Provinces in Sierra Leone, has a long and unique history of strong political female leadership and female warriors, and a contemporary system of respected female paramount chiefs. I had initially hypothesized that armed groups with majority Mende fighters may be less likely to use rape, due to presumably stronger norms about gender equality than other ethnic/tribal identities. However, the RUF was comprised of majority Mende fighters—as were each of the three main armed groups, two of which were not reported to rape on a large scale. Assuming that ethnicity is a plausible proxy measure for views on (one form of) gender equality, these norms do not seem to explain variation in which armed groups committed rape in Sierra Leone. This example also illustrates how difficult it is to measure and assess norms of gender equality on the micro level. This would be a fruitful avenue for future research.

More broadly, no country-level variable—including gender inequality—can adequately account for this variation. Instead, and returning to my previous point, the answer must lie in something about the armed group itself. As I and my coauthors have argued elsewhere,[24] gender inequality is a necessary but not sufficient condition for mass rape.

Thank you to Thomas Maddux and H-Diplo for organizing this exchange and for helpful feedback, and my four colleagues for their challenging critiques and questions.



[1] For example, see: Claudia Card, “Rape as a Weapon of War,” Hypatia 11:4 (1996): 5-18; Lisa Sharlach, “Rape as Genocide: Bangladesh, the Former Yugoslavia, and Rwanda,” New Political Science 22:1 (2000): 89-102; Bülent Diken, and Carsten Bagge Laustsen, “Becoming Abject: Rape as a Weapon of War,” Body & Society 11:1 (2005): 111-128; Doris E. Buss, “Rethinking ‘Rape as a Weapon of War’,” Feminist Legal Studies 17:2 (2009): 145-163.

[2] For example, see: Elisabeth Jean Wood, “Variation in Sexual Violence during War,” Politics & Society 34:3 (2006): 307-342; Michele L. Leiby, “Wartime Sexual Violence in Guatemala and Peru,” International Studies Quarterly 53:2 (2009): 445-468; Amelia Hoover Green, “The Commander’s Dilemma: Creating and Controlling Armed Group Violence,” Journal of Peace Research 53:5 (2016): 619-632; Dara Kay Cohen, and Ragnhild Nordås, “Sexual Violence in Armed Conflict: Introducing the SVAC Dataset, 1989-2009,” Journal of Peace Research 51:3 (2014): 418-428.

[3] Jessica A. Turchik and Susan M. Wilson, “Sexual Assault in the US Military: A Review of the Literature and Recommendations for the Future,” Aggression and Violent Behavior 15:4 (2010): 267-277.

[4] Miranda Alison, “Wartime Sexual Violence: Women’s Human Rights and Questions of Masculinity,” Review of International Studies 33:1 (January 2007): 35-90; Inger Skjelsbaek, The Political Psychology of War Rape: Studies from Bosnia and Herzegovina (New York: Routledge, 2011).

[5] Valerie M. Hudson, Valerie M, Donna Lee Bowen, Perpetua Lynne Nielsen, with Kaylee Hodgson and Rebecca Nielsen, “The First Political Order: Sex, Governance and National Security: Report of Findings on Political Stability Measures,” paper presented at the annual conference of the International Studies Association, 5 April 2018, San Francisco, California.

[6] However, I would like to see a more in-depth exploration of the hypothesis found in note 42 of Chapter 5—”A third alternative, which is not considered here in depth, is the simple hypothesis that groups that require civilian support will abuse the population less frequently” (240). Cohen mentions that this factor cannot explain temporal variation, but such reliance on civilians is not necessarily a static variable, either. Also, it is noteworthy that Roman Catholic groups that could assert divine prohibition of rape, and which also prescribed death for rapists (presumably as an outgrowth of that religious norm), were far less likely to commit mass rape in Cohen’s case studies. It would be useful to further investigate the role of such religious norms in explanations of rape during civil war. Last, given that several informants in the cases where mass rape occurred stated that their leaders led the way (e.g., “the boss would go first” (127; see also 123, 147, 156, 167)), it is probably unclear whether we should believe those who said they would not be punished if they did not follow their leaders. There may well be a tactical element at work here.

[7] Elin Bjarnegård, Karen Brouneus, and Erik Melander, “Honor and Political Violence: Micro-Level Findings from a Survey in Thailand,” Journal of Peace Research 54:6 (2017): 748-761, DOI:; Elin Bjarnegård and Erik Melander, “Pacific Men: How the Feminist Gap Explains Hostility,” The Pacific Review 30:4 (2017): 478-493, DOI:

[8] Cohen’s first articulation of this is that some groups “never do” (1), which I think is an exaggeration. That said, the extreme version of this claim is nowhere inherent in or necessary to Cohen’s actual argument, which I treat as comparing relative (in)frequency.

[9] Cohen cites work like Miranda Alison, “Wartime Sexual Violence: Women’s Human Rights and Questions of Masculinity.” Review of International Studies 33:1 (2007): 75-90; Maria Eriksson Baaz and Maria Stern, “Why Do Soldiers Rape Masculinity, Violence, and Sexuality in the Armed Forces in the Congo (DRC),” International Studies Quarterly 53:2 (2009): 495-518; Joshua Goldstein War and Gender. Cambridge: Cambridge University Press; Lene Hansen, “Gender, Nation, Rape: Bosnia and the Construction of Security.” International Feminist Journal of Politics 3:1 (2001): 55-75; Catharine MacKinnon (1994) “Rape, Genocide, and Women’s Human Rights,” in Mass Rape: the War against Women in Bosnia-Herzegovina, edited by Alexandria Stiglmayer. (Lincoln: University of Nebraska Press, 1994): 73-81; Emile Ouédraogo, “Advancing Military Professionalism in Africa.” Research Paper No. 6 (2014), Africa Center for Strategic Studies, Washington, D.C.:; Jeremy Weinstein, “Resources and the Information Problem in Rebel Recruitment,” Journal of Conflict Resolution 49:4 (2005): 598-624.

[10] Cynthia Enloe, Bananas, Beaches, and Bases: Making Feminist Sense of International Politics (Berkeley: University of California Press, 1989).

[11] Aaron Belkin, Bring Me Men: Military Masculinity and the Benign Façade of American Empire, 1898-2001 (New York: Columbia University Press, 2012).

[12] Spike V. Peterson, “Gendered Identities, Ideologies, and Practices in the Context of War and Militarism” in Laura Sjoberg and Sandra Via, eds. Gender, War, and Militarism: Feminist Perspectives (Santa Barbara: Praeger Security International, 2010), 17-29.

[13] Annica Kronsell, Gender, Sex, and the Postnational Defense: Militarism and Peacekeeping (New York: Oxford University Press, 2012).

[14] See also Elisabeth Wood, “Rape as a Practice of War: Toward a Typology of Political Violence,” Politics & Society 46:4 (2018): 513-537.

[15] And what of the many conflicts where ethnic divisions are not a salient cleavage but rape nonetheless is perpetrated by armed fighters? Ethnic cleavages also do not result in a clear prediction about whether and how rape takes shape: in some cases, ethnic hatreds are so strong that sexual contact (including rape) is strictly taboo.

[16] Dara Kay Cohen, “The Ties That Bind: How Armed Groups Use Violence to Socialize Fighters.” Journal of Peace Research 54:5 (2017): 701-714.

[17] Joseph A. Vandello, Jennifer K. Bosson, Dov Cohen, Rochelle M. Burnaford, and Jonathan R. Weaver, “Precarious Manhood.” Journal of Personality and Social Psychology 95:6 (2008): 1325-1339.

[18] Elisabeth Jean Wood, “Variation in Sexual Violence During War.” Politics & Society 34:3 (2006): 307-342.

[19] Laura Sjoberg, Kelly Kadera, and Cameron G. Thies, “Reevaluating Gender and IR Scholarship: Moving Beyond Reiter’s Dichotomies Toward Effective Synergies,” Journal of Conflict Resolution 62:4 (2018): 848-870.

[20] Sjoberg, Kadera, and Thies, 859.

[21] Kara Ellerby, No Shortcut to Change: An Unlikely Path to a More Gender-Equitable World (New York: NYU Press, 2017), 43.

[22] “The List: The Worst Places to Be a Woman,” Foreign Policy, 23 May 2008,

[23] Macartan Humphreys and Jeremy Weinstein,“What the Fighters say: A Survey of Ex-combatants in Sierra Leone, June–August 2003,”, 2004.

[24] Dara Kay Cohen, Amelia Hoover Green, and Elisabeth Jean Wood, Wartime Sexual Violence: Misconceptions, Implications, and Ways Forward (Washington, D.C.: United States Institute of Peace, 2013),