Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, International Studies. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 15 April 2024

Civilian Victimization During Conflictfree

Civilian Victimization During Conflictfree

  • Alexander B. DownesAlexander B. DownesGeorge Washington University
  •  and Stephen RangazasStephen RangazasGeorge Washington University


Since the end of the Cold War, scholars have increasingly sought to explain the causes of civilian victimization—the intentional use of violence against noncombatants—during armed conflict. The question of the effects and effectiveness of violence against civilians, in contrast, has received less scholarly attention. One strand of research examines the impact of wartime civilian victimization on postconflict political behavior and outcomes. A second strand investigates the effectiveness of violence during the war itself. The principal question this literature asks is: Does civilian victimization “work”? Put more precisely, can intentionally targeting noncombatants help belligerents achieve their wartime objectives, whatever those might be? Civilian victimization takes different forms and serves different purposes in different kinds of conflicts; scholarship on its effectiveness is thus divided into work on irregular (predominantly intrastate) wars versus conventional (predominantly, but not exclusively, interstate) wars. No matter what its particular form or in which type of conflict it is used, however, civilian victimization tends to follow two broad logics: coercive and eliminationist. Most scholarship on irregular wars examines the effectiveness of coercive victimization, whereas studies of conventional war look at the efficacy of both. For example, a key debate in the literature on civilian victimization in irregular wars concerns whether selective or indiscriminate violence is more effective at deterring civilians from shifting their allegiances to the adversary. A broad consensus holds that violence is effective only when selective, but new studies have found that indiscriminate violence can also work under certain circumstances. Similarly, there is broad agreement (with some notable exceptions) in the literature on conventional war that coercive civilian victimization—which is almost by definition indiscriminate—is ineffective. In contrast, scholars have yet to assess systematically the effectiveness of eliminationist victimization in conventional war.


  • Conflict Studies
  • Security Studies


Although violence against civilians in wartime is as old as war itself, civilian victimization has become a central topic of study for scholars of armed conflict in both comparative politics and international relations only since the end of the Cold War. A watershed moment was the publication of Stathis Kalyvas’s book The Logic of Violence in Civil War, which brought the question of why civilians are targeted so pervasively in intrastate conflicts front and center (Kalyvas, 2006). A torrent of research followed, examining the effects of numerous factors on the causes of civilian victimization—primarily by rebels—in intrastate wars, including resource endowments and economic shocks (Dube & Vargas, 2013; Rigterink, 2020; Weinstein, 2007), external support (Salehyan et al., 2014; Zhukov, 2017), military weakness (Wood, 2010) and battlefield losses (Hultman, 2007; Wood, 2014), organizational structure (Abrahms & Potter, 2015; Manekin, 2020) and ideology (Gutiérrez Sanín & Wood, 2014; Hoover-Green, 2018; Kalyvas & Balcells, 2010), competition among rebel groups (Cunningham et al., 2012; Wood & Kathman, 2015), socialization processes (Cohen, 2016), emotions (Bulutgil, 2016; Costalli & Ruggeri, 2015; Petersen, 2002), and much more (for thorough reviews, see Balcells & Stanton, 2021; Wood, 2016).1 Other studies take the opposite approach, seeking to understand the causes of restraint with respect to civilians by states and rebel groups (Bell, 2016; Hoover-Green, 2018; Jo, 2015; Stanton, 2016; Straus, 2012). Separate but related, the study of the causes of terrorism—much but not all of which occurs in the context of civil wars—grew in quantity and scholarly stature following al-Qaeda’s attack on the World Trade Center and the Pentagon on September 11, 2001 (Bloom, 2005; Hoffman, 2006; Kydd & Walter, 2006; Moghadam, 2008; Pape, 2003, 2005; Shapiro, 2013).

In international relations, Robert Pape, although not explicitly writing about civilian victimization, demonstrated in the context of strategic bombing in interstate wars that punishment strategies, which involve the intentional targeting of civilians, routinely fail to yield coercive success (Pape, 1996).2 Studies of the causes of civilian victimization (and restraint) in interstate wars soon followed, with analysts emphasizing the importance of leaders’ specific objectives (Valentino, 2004), desperation to win and reduce casualties (Downes, 2006, 2008; Valentino et al., 2006), territorial annexation (Downes, 2006, 2008), military culture and norms (Bell, 2022), European identity (Fazal & Greene, 2015), and international law (Morrow, 2014; Valentino et al., 2006).3

No review article can do justice to the flood of research that has appeared since the 1990s. Moreover, recent literature reviews have ably covered the recent work on the causes of civilian victimization in intrastate wars (Balcells & Stanton, 2021; Wood, 2016). The focus here, therefore, is primarily on the effects and effectiveness of civilian victimization. Although there is a growing literature that investigates the consequences of violence inflicted during wars on civilians’ attitudes and beliefs after the shooting stops (Balcells, 2012; Daly, 2022; Rozenas et al., 2017; Wood, 2003), this article reviews studies that examine the effectiveness of civilian victimization on wartime outcomes such as compelling collaboration with armed actors (and deterring collaboration with other armed actors), compelling adversaries to end a war, directly weakening or defeating opponents, and consolidating control over territory.

To do so, armed conflicts are divided into two types—conventional and irregular—that only partially overlap with the dominant categories of interstate and intrastate established by the Correlates of War Project.4 Although interstate wars (with a few exceptions) tend to be conventional, intrastate wars can take either form. Although the specific forms that civilian victimization takes in these wars differ, they have similar underlying logics in how they work their effects. First, civilian victimization can be coercive, used to compel individuals, groups, or states to take certain actions (or stop taking actions they have already started) or deter them from initiating undesirable (from the coercer’s point of view) actions. Coercion—whether it takes the form of compellence or deterrence—takes two forms: punishment and denial. Punishment threatens or inflicts pain on a target to influence a third party, whereas denial threatens or uses force to thwart an adversary’s strategy for victory (Pape, 1996). In studies of coercion, punishment by convention is equated with targeting civilians while denial is assumed to target military forces. Neither of these follows logically. Punishment inflicts pain on something the target values, which could be a variety of things, including national symbols or monuments, religious sites, or even its military forces. And thwarting an adversary’s military strategy can entail taking actions that inflict harm on civilians, such as chemical defoliation or scorched earth strategies. Here, the focus is on the effectiveness of punishment and denial strategies that target civilians or civilian livelihoods.

The second form of civilian victimization is eliminationist. While it sounds sinister, eliminationist victimization does not necessarily entail mass killing or genocide. Eliminationism is a strategy in which belligerents, whether for reasons of security or territorial aggrandizement, seek to remove a target population from a particular piece of territory. Although expulsions, cleansing, or scorched earth can be coercive in intent, in many cases they are driven simply by the desire to take and hold territory depopulated of those who might challenge a belligerent’s control. Eliminationist victimization is not necessarily implemented to help win the current war; rather, it is forward looking and about securing the peace after the war.5

Few findings regarding the effectiveness of civilian victimization are uncontested. Perhaps the clearest finding in the literature is that coercive victimization in the form of strategic bombing in conventional wars is ineffective. In irregular warfare, scholars largely agree that selective violence—which combines punishment and denial—is more effective than indiscriminate violence. Yet several studies have challenged this consensus, finding that indiscriminate violence—at least under certain circumstances—works. For many types of victimization, however, there is either no unanimity among scholars on effectiveness or simply no basis for judgment, and thus further research is necessary.

The article proceeds as follows. First, it defines the key terms used in the analysis, such as “civilians” and “civilian victimization,” “conventional” and “irregular war,” “selective” and “indiscriminate violence,” and “punishment” and “denial.” Second, it examines the effectiveness of mainly coercive forms of civilian victimization in irregular war. Third, the article turns to the effectiveness of coercive and eliminationist violence in conventional wars. Both of these sections conclude by identifying areas where further research is needed. The final section summarizes key points and addresses ethical and moral questions raised by research on this topic.

Defining Terms

This section lays out some essential definitions. First, it briefly defines “civilians and civilian victimization” and outlines the many forms that violence against civilians can take. Second, it differentiates between two types of warfare, conventional and irregular, and explains why a third type—symmetric non-conventional (SNC)—is omitted. Third, the section introduces the distinction between selective and indiscriminate violence, the dominant typology of violence in the literature on civilian victimization in irregular war. Fourth, it explains the difference between the two strategies of coercive victimization, punishment and denial.

Civilians and Civilian Victimization

Defining “civilians/noncombatants” is not as simple as it may appear at first glance. Most definitions are based on participation versus non-participation in armed conflict. Common Article 3 of the 1949 Geneva Convention defines civilians as “persons taking no active part in hostilities,” and the subsequent Additional Protocol I of 1977 (Articles 43 and 50) interprets active participation narrowly as applying only to full-time members of states’ armed forces or nonstate armed groups. Others have interpreted the phrase “taking no active part in hostilities” more broadly. Michael Walzer (1977), for example, included workers in munitions factories as combatants. Once the door is opened to exceptions, however, it becomes difficult to draw a clear line around who is in and who is out. If munitions workers count as combatants, can they be targeted at home as well as at work? What about individuals involved in producing military uniforms or growing food soldiers need to survive? The “active participation” criterion is even more complicated to assess in guerrilla warfare, where insurgents rely directly on the population for food, labor, recruits, and information, and people may serve as part-time combatants, such as by joining a village self-defense militia. The International Committee of the Red Cross (ICRC) (Melzer, 2009), for example, proposed a definition of combatants based on the concept of “continuous combat function,” which includes only full-time members of armed factions. Part-time fighters would thus be considered combatants (and thus targetable) only when they are fighting but would be civilians when they are not.

Most empirical studies of civilian victimization sidestep these knotty problems and adopt relatively standard definitions of who is a civilian but avoid going into great detail regarding how they identify civilians in any given case. Valentino (2004, p. 13), for example, defined a civilian as “any unarmed person who does not actively participate in hostilities by intending to cause physical harm to enemy personnel or property.” Downes (2008), adopted a similar definition but excluded certain classes of people based on their proximity to active participation, including munitions workers, sailors on armed merchant ships, and contractors who service and maintain weapons (Downes, 2008, pp. 14–15 and 263n6). In the context of guerrilla insurgencies, Kalyvas (2006, p. 415) defined noncombatants simply as “individuals who are not full-time armed members of a faction,” thereby excluding part-time fighters, informants, or collaborators, and others who provide non-combat support. In studies based on fieldwork of individual killings (e.g., Balcells, 2017; Kalyvas, 2006), scholars are sometimes able to identify particular individuals and verify by their activities that they were not combatants. Studies of broader campaigns of victimization (Downes, 2008; Valentino, 2004), however, have to rely on aggregate fatality counts and assume (usually justifiably) that most were civilians.

Given this definition of civilians as persons who do not actively participate in hostilities, civilian victimization is simply “the violent and intentional victimization of civilians” (Kalyvas, 2006, p. 20). Most studies further specify that civilian victimization consists of policies of sustained and purposeful targeting of civilians rather than random or accidental violence. Downes, writing in the context of state violence in interstate wars, argued that “civilian victimization is a government-sanctioned policy or strategy, as opposed to random or uncoordinated attacks by a few military units” (Downes, 2008, p. 15). Another way of saying this is that civilian victimization is authorized by an armed actor, whether that be a state or a rebel group. Although early studies (e.g., Downes, 2008; Kalyvas, 2006; Valentino, 2004) equated civilian victimization with killing, the term encompasses any form of intentional lethal or non-lethal harm, including torture (Hassner, 2022), rape (Cohen, 2016), forced expulsion (Lichtenheld, 2020), and destruction of civilian property or livelihoods (Feuer, 2023).

Technologies of Warfare

This review divides warfare into two categories: conventional and irregular (Kalyvas & Balcells, 2010). Conventional war refers to conflicts in which belligerents, who wear uniforms, exert exclusive control over territory, and are separated by clear front lines, confront each other directly in pitched battles using heavy weapons such as artillery or tanks. It should be noted that this typology transcends the traditional interstate-intrastate divide. Although almost all interstate wars are fought conventionally, a substantial proportion of civil wars have also been fought in this way—for example, in the United States (1861–1865), Spain (1936–1939), Bosnia (1992–1995), and Syria (2011–). Similarly, numerous wars of imperial expansion have taken a conventional form, including Britain’s many wars in South Asia (e.g., First and Second Sikh Wars of 1845 and 1848) and Africa (e.g., First and Second Mahdi Wars of 1881 and 1896), France in Syria (1920), Russia (Khokand, 1864, and Bukhara, 1865), and many more.

Irregular wars are armed conflicts in which the nonstate side—owing to its material weakness relative to the state—avoids pitched battles, preferring instead to launch surprise, hit-and-run attacks at times and places where it has local superiority. Insurgents typically operate in small, mobile bands to avoid detection and destruction by state forces. Territorial control in irregular wars is often more ambiguous than in conventional wars: Although guerrillas may establish control over base areas in remote regions, there are also zones in which incumbents and insurgents exert partial control, and areas where neither side exercises control (Kalyvas, 2006).

A third category of conflict—SNC—that does not fit into the conventional-irregular dichotomy is omitted. As described by Kalyvas and Balcells (2010), SNC wars typically occur when states are very weak or have collapsed and thus do not possess a marked advantage over rebel groups. The result is a conflict that resembles a struggle between two poorly armed militias rather than a war between a state army and a rebel group. For analytical simplicity, this article sets aside SNC wars—the dynamics of which have not been the subject of much scholarship—and focuses on the conventional and irregular categories, which have garnered the bulk of the scholarly attention.6

Selective Versus Indiscriminate Violence

Since it served as the centerpiece of Kalyvas’s analysis, the distinction between selective and indiscriminate violence has dominated the intrastate war literature on violence against civilians. Selective violence is targeted at specific individuals based on their (real or suspected) actions. It is often enabled by information provided by local informants collaborating with one of the belligerents who can identify defectors—individuals fighting for, or cooperating with, the adversary.

Indiscriminate violence, in contrast, consists of violence that is not targeted at particular individuals because of their actions but rather is based on “guilt by association” (Kalyvas, 2006, p. 142) or simply being in the wrong place at the wrong time. A good example of indiscriminate violence is the reprisal massacre. Suppose that insurgents destroy a railroad line near a village. Incumbent forces arrive but lack information on who, if anyone, in the village assisted the rebels in carrying out the attack; all they know is that the rail line was blown up. As punishment for the attack, incumbent forces select 10 villagers and kill them. For all they know, the dead villagers were government supporters. Because incumbents lack information about who is who and who has done what, and that information is costly and time-consuming to acquire, they may simply kill anyone who happens to be in the area of the insurgent attack. Indiscriminate violence is thus not totally random: It arises out of some war-related activity but then is random among those guilty by association of being near that activity.7

Punishment Versus Denial

In the literature on coercion in international relations, two broad categories of strategies are recognized: punishment and denial. Successful coercion—whether in the form of deterrence or compellence—is a function of costs and benefits. Punishment, Pape wrote, “operates by raising costs or risks to civilian populations,” whereas denial “operates by using military means to prevent the target from attaining its political objectives or territorial goals” (Pape, 1996, p. 13). Punishment thus increases the cost side of the equation while denial reduces the benefit side.8 Pape was careful to acknowledge that punishment is not limited to civilians (Pape, 1996, p. 13) but proceeded to treat it as if it were. Indeed, civilians as the target of punishment forms the basis of the mutual assured destruction (MAD) school of nuclear deterrence. MAD holds that as soon as each side in a rivalry possesses a nuclear arsenal sufficient to destroy its adversary’s society, and which cannot be eliminated in a first strike, stability (and peace) should prevail (Jervis, 1989). The assumption that punishment targets civilians is thus firmly cemented in the literature.

However, just as punishment need not always be directed at civilians, neither is denial necessarily targeted at military forces or objects. Instead, denial effects can be achieved by victimizing civilians. In conventional war, for example, states sometimes aim to make certain strategic regions untraversable by consuming or destroying all sources of sustenance such that armies on the move cannot feed themselves. Similarly, in irregular war, states have used population concentration to “drain the sea” of the people that insurgents rely on as crucial sources of supply, recruits, labor, and intelligence (Downes et al., n.d.). These are examples of denial strategies aimed at reducing the adversary’s likelihood of winning through civilian victimization.

Effectiveness of Civilian Victimization in Irregular Wars

Kalyvas (2006) organized civilian victimization during irregular wars into violence that seeks to coerce civilians and an eliminationist logic with no goal of governing the targeted civilians. Although work on the latter is growing (Schwartz & Straus, 2018; Steele, 2017; Straus, 2015), most of the scholarship on the effectiveness of civilian victimization focuses on the former.9

Two categories of effects of coercive civilian victimization are commonly studied: consolidating control over civilians in an area and weakening an opposing armed actor. With regard to control, scholars tend to find that precisely targeted victimization, commonly referred to as selective violence, is generally more effective at coercing civilians to collaborate with an armed actor than is collective or indiscriminate targeting. Less precise methods of targeting are at best ineffective at obtaining collaboration and may even counterproductively drive civilians to cooperate with an opposing armed actor. With regard to weakening an adversary, despite notable exceptions, many scholars have found that the use of indiscriminate violence by states generally increases rather than weakens an opposing insurgency’s strength. Insurgent strength is usually measured as either an armed actor’s immediate strength, such as its ability to hold territory or launch attacks, or broader conflict outcomes, such as victory or defeat. Similarly, an increasing body of work has found that the use of terrorism lowers insurgents’ chances of winning a conflict or gaining concessions (Abrahms, 2006; Fortna, 2015), though again there are notable exceptions to this finding (Pape, 2003, 2005). Even if there is far less consensus on the effects against opposing armed actors, the common finding on civilian victimization during irregular conflicts is that less rather than more violence against civilians is what tends to “work.”

Effects on Consolidating Territory

Kalyvas (2006) famously argued that armed actors employ civilian victimization in areas where they are consolidating their territorial control by using selective violence to coerce civilian collaboration. According to Kalyvas, selective violence combines a coercive punishment logic to deter civilians from collaborating with the adversary while also eliminating specific civilians who acted against the armed actor, thus contributing to denial (Kalyvas, 2006, p. 27). By using coercive threats to deter others from defecting, selective violence is also less costly than using violence to directly eliminate every possible threat that arises. This is particularly useful for armed groups when they lack other means of ensuring civilian collaboration in their areas of control (Wood, 2010). Kalyvas (2006) argued that civilian collaboration will follow territorial control as civilians falsify their preferences and comply with armed actors’ demands to ensure their survival. Once an armed group fully consolidates its control over an area, the need for civilian victimization declines as civilians refusing to collaborate have been eliminated or coerced.

Gilbert (2022) extended Kalyvas (2006) by exploring insurgent groups’ use of a nonlethal form of violence, kidnapping, to ensure civilian collaboration with rebel taxation. Gilbert found that the coercive threat of selective kidnappings—by punishing those who failed to pay while deterring shirking from others—led many elites and businesses in Colombia to pay taxes to leftist insurgencies. Similarly, Souleimanov et al. (2022) found evidence that the Russian state’s selective targeting of the kin of known insurgents was more effective at deterring collaboration with the Chechen insurgency than less precise methods of targeting, such as indiscriminate bombing. Interestingly, both Gilbert (2022) and Souleimanov et al. (2022) raised the question of whether there are long-term reputational costs of using selective violence despite its short-term efficacy but left the exploration for future work.

These works on selective violence emphasized that to be effective at coercion, armed actors must possess the ability to collect sufficient information on those refusing to comply and selectively punish them to credibly communicate the risks of noncompliance to other civilians in the area. When armed actors lack the information to selectively target civilians, they lose the ability to effectively communicate coercive threats because even collaborating civilians may be targeted for violence (Kalyvas, 2006). Accordingly, Kalyvas (2006) argued that the use of indiscriminate violence is rooted in non-strategic explanations and will tend to decline over the course of a conflict as armed actors learn of its ineffectiveness. Even worse, by punishing indiscriminately, armed actors may drive civilians to support or join their opponents if they can offer civilians more protection than those who attempt to stay neutral (Kocher & Kalyvas, 2007; Mason & Krane, 1989) or through moral outrage (Wood, 2003).

Empirical studies focusing on civilians’ reactions to violence generally support the argument that indiscriminate violence may backfire. Civilians punish armed groups that use violence indiscriminately by withdrawing support for the group or viewing opposing armed actors more favorably (Lyall et al., 2013; Pechenkina et al., 2019), or supplying information to the other side (Schutte, 2017; Shaver & Shapiro, 2021).10 There are significant caveats to this finding, however. Lyall et al. (2013), for example, found evidence of in-group bias in civilian responses to violence in Afghanistan: The negative effect of violence on civilian support for the International Security Assistance Force versus the Taliban was much greater after violence by the former than the latter. Pechenkina et al. (2019), using evidence from Ukraine, reported that individuals who directly experienced indiscriminate artillery shelling at the hands of the government were likely to hold the government responsible even when the violence was provoked by the insurgents. In contrast, bystanders were willing to attribute blame to insurgents if they believed government attacks were rebel-provoked. Moreover, recent research suggests that civilians are less likely to punish armed groups for unintentional collateral damage if the preparator materially compensates those affected (Blair, 2022; Lyall, 2019; Silverman, 2020). Research has yet to determine whether these programs mitigate civilian responses to intentional indiscriminate violence, however.

In contrast to these approaches, several scholars have found that armed actors may successfully use indiscriminate violence to coerce civilian collaboration to their side. Zhukov (2023) argued that indiscriminate violence may ensure civilian collaboration but only at massive levels that successfully out-terrorize the opposing side’s coercive threats. Similarly, to avoid further indiscriminate reprisals by the state, civilians may join pro-government militias as a way of signaling their loyalty (Schubiger, 2021). Finally, Daly (2022) showed that in postwar elections civilians do not always punish armed actors who inflicted significant levels of violence on noncombatants during the war, suggesting that long-term costs may be minimal.

Effects on Weakening Armed Opponents

As opposed to scholars who focus on groups that consolidate territorial control to ensure civilian collaboration, other scholars seek to evaluate the effect of civilian victimization on defeating armed opponents. Armed actors in irregular wars often use civilian victimization to gain concessions or drive their opponent to surrender through the coercive mechanisms of punishment or denial (Pape, 1996). However, there is little consensus on the effectiveness of civilian victimization for coercing an armed opponent.

Much of the work on the effectiveness of indiscriminate violence by states focuses on the dependent variable of armed actors’ immediate strength, most commonly measured as the ability to launch attacks and secondarily as territorial control. Several works have found that indiscriminate violence by states weakened insurgencies’ ability to operate through the punishment mechanism of coercing communities into pressuring the insurgents to cease attacks (Lyall, 2009) or through a denial logic of depriving the insurgency the population’s support (Downes, 2007; Valentino, 2004; Zhukov, 2015). Downes (2007), for example, argued that British use of indiscriminate violence during the Second Anglo-Boer War isolated the insurgents from the population, resulting in their ultimate defeat. From this case, Downes theorized a set of conditions under which indiscriminate violence as a denial strategy is likely to be more effective at defeating insurgents: the smaller the insurgency’s population base, the smaller and more constrained the geographic scope of the insurgency, and the less external supply available to the insurgency.

In contrast, other scholars have found the opposite result: that indiscriminate violence strengthens insurgencies. Kocher et al. (2011), for instance, showed that villages indiscriminately bombed by the United States during the Vietnam War were more likely to shift toward Viet Cong territorial control compared to those not targeted for bombing. Similarly, using two sophisticated regression discontinuity designs, Dell and Querubin (2018, p. 701) found that bombing during the Vietnam War “increased the military and political activities of the communist insurgency, weakened local governance, and reduced noncommunist civic engagement.” These studies interpreted their results in line with the theoretical work previously mentioned that indiscriminate violence or overwhelming firepower leads to counterproductive civilian mobilization, which in turn strengthens the opposing side’s ability to operate. Separate but related, Souleimanov and Siroky (2016), using qualitative evidence from Chechnya, challenged Lyall’s (2009) result that random artillery shelling suppressed insurgent attacks. They argued that any reduction in attacks was only short term, as it took insurgents 6–9 months to recover and respond. Finally, Condra and Shapiro (2012) found that collateral damage, although not intentional indiscriminate violence, increased insurgent attacks in Iraq.11

The literature on the effectiveness of insurgent indiscriminate violence is also quite mixed. Pape (2003) argued that the spread of suicide bombing—a form of punishment—was due to its effectiveness in coercing democracies to make concessions. Thomas (2014), using data on civil wars in Africa, found that the use of terrorism increased the chances of insurgents being invited to negotiate and increased the probability of being granted concessions. Wood and Kathman (2014), also using data from African civil wars, provided evidence that a moderate level of violence against civilians increases the probability of settlement for insurgents. In contrast, other work has found that the use of terrorism lowers insurgents’ chances of ultimately gaining concessions or winning conflicts (Abrahms, 2006, 2012; Fortna, 2015). These studies argued that terrorism against civilians is far less effective at securing favorable conflict outcomes for insurgents than attacks against military targets.

Avenues of Future Research on Civilian Victimization During Intra-State Wars

There are several directions for future research on the effectiveness of civilian victimization in irregular wars. First, more work is needed to refine measurements of civilian victimization. For instance, a gap exists between theoretical arguments and measures of indiscriminate violence. Much of the theoretical literature follows Kalyvas’s (2006) original definition that considers indiscriminate violence as the intentional targeting of civilians to advance an armed group’s political goals. However, much of the empirical literature operationalizes indiscriminate violence in ways—such as bombing or shelling—that include substantial amounts of unintentional or accidental violence (i.e., “collateral damage”; see Dell & Querubin, 2018; Kocher et al., 2011; Lyall, 2009; Lyall et al., 2013; Newton & Tucker, 2022; Schutte, 2017). This operationalization is sufficient for studies that seek to explain how civilians react to the imprecise use of violence; however, indiscriminate violence conducted for a strategic purpose and collateral damage from “troops in contact” or accidents likely arise from different logics that generate different effects. The further leveraging of data from archives on historical intra-state cases (Balcells & Sullivan, 2018) may offer ways to operationalize intentional indiscriminate violence more precisely. For instance, Sullivan (2012) used data from declassified archives from the Guatemalan Civil War (1960–1996) to operationalize indiscriminate violence as massacres that pose less risk of including collateral damage.

Second, much of the existing literature focuses on the coercive effects of violence through the logic of punishment. To use the example of Kalyvas (2006), there is more attention paid to the broader coercive effects of selective violence on civilian audiences than to the direct effects of punishing a specific civilian. However, armed actors may benefit from the effects of eliminating individuals or destroying their property even when not using strategies aimed at group destruction or genocide (Steele, 2017). During the contexts of irregular wars, the effects of eliminating specific individuals significantly affect political outcomes. For instance, at the beginning of insurrection, the selective killing of public officials commonly drives much of the breakdown of a state’s local monitoring capacity, creating a vacuum of authority prior to an insurgency gaining sufficient strength to gain territorial control of an area. Similarly, states often leverage the effects of indiscriminate violence against civilians as part of a denial strategy for coercing an armed opponent, not the targeted civilians (Downes, 2007; Pape, 1996). States using these strategies hope that the systematic destruction of property or population removal (Zhukov, 2015), types of violence often neglected in the existing literature, overpowers any gains to the insurgency from the mobilizing effects of violence previously discussed. Further investigation of armed actors’ use of strategies leveraging the direct effects of violence may help resolve existing theoretical puzzles and open additional avenues of research.

Finally, there is a lack of systematic testing of the claims of effectiveness made by many of the theory-building studies presented here. For example, although each study presents detailed anecdotal evidence from a specific intra-state conflict, claims that selective violence is effective in ensuring civilian collaboration (Gilbert, 2022; Kalyvas, 2006; Souleimanov et al., 2022) have largely not been tested using cases outside of those where the theory originated. Similarly, future work should pursue new sources of data to test mechanisms of civilian responses to indiscriminate violence where there is either a lack of testing of mechanisms presented (Kocher & Kalyvas, 2007) or tests of attitudinal rather than behavioral shifts in the local population (Lyall et al., 2013; Pechenkina et al., 2019). Additional research, such as Schubiger (2021), that evaluates the mobilizing effects of violence will be difficult but essential for resolving debates on the conflicting findings on the effectiveness of indiscriminate violence for weakening opponents.

Effectiveness of Civilian Victimization in Conventional Wars

The literature on the effectiveness of civilian victimization in conventional wars is sparser than that in irregular conflicts. Much of the work that does exist is case-specific. The debate over the relative influence of the atomic bombs on Japan’s decision to surrender in 1945, for example, is voluminous (and polarized).12 There are many reasons for this imbalance. Most interstate wars are conventional wars, and International Relations scholars have devoted far more attention to studying the causes of such wars than to how they were conducted, and have given almost no attention at all to their impact on noncombatants (for exceptions, see Downes, 2008; Valentino, 2004).

Unlike in irregular wars, access to an adversary’s civilian population is not given because territorial control is strictly segmented and under the exclusive control of a single belligerent. The implication of this territorial division is that civilian victimization can occur only if belligerents have the military capability to reach the enemy’s population. The Spanish Civil War illustrates the two ways this can happen (Balcells, 2017). When Republican or Nationalist forces took territory previously under the control of the other, they gained immediate access to a population that contained enemy supporters on whom they could inflict “direct,” face-to-face, violence. In areas under firm enemy control, however, the belligerents could use only “indirect,” outside-in, violence, such as strategic bombing of enemy-held cities (Balcells, 2017, pp. 21–24). Direct violence can thus be either selective or indiscriminate; indirect violence is almost always indiscriminate.

The two main types of civilian victimization in conventional wars are coercive and eliminationist.13 Much coercive victimization, such as siege, blockade, and aerial bombardment, is indirect, but it can also be direct, as when belligerents carry out scorched earth policies in enemy territory to compel an adversary to surrender. Coercive victimization can follow either the logic of punishment or that of denial. In the former, violence aims to compel an adversary to stop fighting by inflicting pain on its population. Because it is indirect, this type of violence is almost inherently indiscriminate because it cannot identify and target individual supporters of the enemy regime. When coercive victimization takes the form of denial, it aims to thwart the enemy’s military strategy, for example, by destroying civilian property and livelihoods as a means to create zones in which enemy armies are unable to operate. It is typically done directly but one instance in which it was done indirectly with airpower is discussed here. Eliminationist victimization, in contrast, is almost always direct, as cleansing or expulsion operations are carried out by troops on the ground.14

Coercive Civilian Victimization

Coercive victimization in conventional wars most often follows a punishment logic. The aim is to compel the opponent to come to terms without having to achieve a brute force victory.15 Siege warfare is probably the oldest form of coercive victimization. Sieges combine the tactics of food deprivation and bombardment in an effort to induce defenders to capitulate, allowing attackers to avoid a costly assault.16 Other forms of coercive victimization can be carried out without ever setting foot in the enemy homeland. Naval blockades, for example, follow a similar punishment logic but are implemented from the outside-in against entire countries rather than cities. Blockades aim to inflict deprivation, hunger, and disease by cutting off imports of food, fertilizers, fuel, or medical supplies from the outside. Similarly, aerial bombardment can be used to punish an adversary’s civilian population without a permanent presence on the ground in enemy territory. In some recent conflicts, such as the 2020 war between Armenia and Azerbaijan, unmanned aircraft (i.e., drones) have been used to drop ordnance on enemy targets (Dixon, 2020). Drones, however, can also be used in kamikaze mode, with the added benefit that unlike an actual kamikaze, no pilot is killed in the attack. Russia, for example, began using Iranian-made Shahed drones in its war against Ukraine in 2022 in “suicide” attacks against Ukrainian energy infrastructure (Kramer, 2022). Long-range artillery (including naval bombardment), ballistic missiles, and cruise missiles are also capable of targeting civilians without ever entering enemy territory.


There is little systematic evidence on the effectiveness of civilian victimization in siege warfare. Historical studies (e.g., Bradbury, 1992) often discuss sieges extensively but have little to say about the frequency with which they obtained the surrender of cities without having to resort to taking the target by storm. Downes and Cochran (2018), in their assessment of the contribution of civilian victimization to victory in wars of the 19th and 20th centuries, counted the siege of Paris during the Franco-Prussian War (1870–1871) as a rare success for punishment. Contemporaneous evidence allowed them to differentiate between the relative contributions of Prussian bombardment and food scarcity and hunger; they concluded that the latter had a far greater impact than the former. On the futility of the cannonade, one observer commented: “if Marshal von Moltke took the ‘moral effect’ of his projectiles into his calculations to accelerate the surrender of Paris, he might have gone on shelling Paris for a twelvemonth without being one whit nearer his aim” (quoted in Richardson, 1982, pp. 109–110; see also Horne, 1965, p. 217; Kranzberg, 1950, p. 133). In contrast, evidence suggests that hunger played a key role in bringing about surrender: “Faced with the imminent prospect of famine on a horrifying scale, the Government decided that it could not delay negotiating an armistice any longer” (Baldick, 1964, p. 222).

One of the few systematic studies of siege warfare suggests that it is not effective. Data on 20th-century sieges gathered by Lionel Beehner, Benedetta Berti, and Michael Jackson suggest that “siege warfare is less militarily effective, especially in cases of civil war or asymmetric conflict,” and “the longer a siege drags on, the more it favors the side under siege” (Jackson et al., 2016, p. 5). The authors argued that sieges succeed only when attackers are “willing to use overwhelming force” and are supported by intervention by an outside power (Beehner et al., 2017, p. 80). Beehner, Berti, and Jackson, however, did not say how common such circumstances are, and the case studies they examined (Aleppo in Syria and Grozny in Chechnya) include massive violence, external intervention, or both, and both were counterinsurgent successes.17

Besiegers historically have also used retributive/reprisal massacres against the inhabitants of one city for coercive purposes against other cities. According to the customs of siege warfare, it was lawful for besiegers forced to storm a city to plunder, pillage, rape, and murder its inhabitants. According to Bradbury (1992, pp. 317–318),

Storm gave the attackers complete control over the lives and fate of the defeated, almost any atrocity was given the cloak of legality: rape, enslavement, killing, in addition to the seizure of homes and property. No mercy need be given, and none could be expected.18

Prawdin (1940) and Grousset (1970), for example, noted Tamerlane’s successful use of this tactic in his invasion of southern Persia in 1387. The city of Ispahan, which had submitted to Tamerlane’s invading armies, rebelled and killed the occupying garrison. Tamerlane returned and stormed the city, commanding his forces to bring him 70,000 severed heads. As Prawdin (1940, p. 448) recounted,

The assize of Ispahan had its desired effect. Shiraz, the capital of southern Persia . . . did not attempt resistance, and, within a few days, produced the required indemnity. From all the local princedoms, the rulers hastened to pay homage to Tamerlane.

Rogers (2002, pp. 61–62) similarly noted the successful use of this practice in Europe in the 14th and 15th centuries.19 Little is known, however, beyond evidence from a few cases about the general efficacy of such massacres in producing future surrenders.


The evidence regarding the efficacy of starvation blockades is similarly spotty. The most analyzed cases are the reciprocal British and German blockades during World War I. The German campaign of unrestricted U-boat warfare in 1917, although the more notorious of the two, was ineffective at knocking the British out of the war and had only a minor effect on food supplies (Moyer, 1995, pp. 186–187). This is not to say it did not produce near panic among British leaders (Terraine, 1989, pp. 47–48); indeed, the enormous tonnage sunk by German U-boats during the spring and summer of 1917 prompted a frantic search for effective countermeasures (for the numbers, see Tarrant, 1989, pp. 148–149). Once Britain and the United States instituted the convoy system, sinkings began to decline and the danger passed (Tarrant, 1989, pp. 148–149). Still, the German plan was based on faulty assumptions, and British wheat supplies—the key metric in measuring the effectiveness of the campaign—were never in serious danger (they actually increased over the course of the campaign; Offer, 1989, p. 366).

The British blockade of the Central Powers, on the other hand, contributed to widespread food shortages over the course of the war, which in turn increased the prevalence and deadliness of certain diseases. Excess deaths attributable to the blockade range from 400,000 to 750,000 in Germany (Offer, 1989, p. 34; Vincent, 1985, p. 141) and about 450,000 in Austria-Hungary (Grebler & Winkler, 1940, p. 147).20 Whether and to what extent civilian suffering affected the German government’s decision to seek an armistice, however, is disputed. What is clear is that the increasing deprivation of its population did nothing to alter German war aims until the country’s military fortunes took a dramatic turn for the worse in the summer and fall of 1918. Only once military defeat loomed did German leaders begin voicing concerns about the effect of hunger on civilian morale, suggesting that punishment of civilians can contribute to victory only when combined with military defeat.21 As Offer (1989, p. 72) put it, “The Allied offensive was the hammer, the home front provided the anvil.”

If these historical cases provide scant support for the effectiveness of starvation blockades, neither do more contemporary cases. In the Biafran War of Secession (1967–1970), Nigerian forces from the fall of Port Harcourt in May 1968 until the end of the war implemented a total blockade of the Biafran enclave. A few months after the blockade became effective, the ICRC estimated that between 8,000 and 10,000 people were dying every day in Biafra in the resulting famine (Wiseberg, 1975). Yet despite a civilian death toll that has been estimated in the 1 million–2 million range (Wiseberg, 1975), the Biafrans continued to launch offensives through 1969 and had to be defeated on the battlefield by Nigerian forces before the enclave fell. Similarly, although the Saudi blockade of Yemen, in effect since 2015, has contributed to a massive humanitarian disaster in the country, it has done nothing to persuade Houthi leaders to give up power (Almosawa, 2022; Sheline, 2023).

Strategic Bombing

Finally, in no realm is the futility of coercive civilian victimization by punishment in conventional wars more established than in strategic bombing. In the interwar period, air power theorists, most notably Giulio Douhet (1921/1942), argued that future bloodbaths like World War I could be avoided by taking war directly to a state’s weakest link: its civilian population, whose fragile morale would surely buckle in the face of sustained aerial bombardment. In the best available study of strategic bombing, however, Pape (1996) argued that punishment fails because conventional bombing is unable to inflict enough pain to compel states to relinquish important objectives. Moreover, states can take measures to reduce the vulnerability of their populations, such as evacuating cities, building air raid shelters, or investing in air defense. Finally, the central idea of punishment—that it turns the target population against its government—gets it exactly wrong: Bombed populations turn their ire against the bomber.

The evidence from a century of bombing is clear: when put to the test of war, strategic bombing of civilians has repeatedly failed to produce decisive effects. Maeda (2009), for example, documented Japan’s failure to compel Chongqing, the provisional capital of China, to submit despite 218 raids and nearly 12,000 deaths from 1938 to 1941. Nor did the German “Blitz” in 1940–1941 induce Britain to seek terms from Hitler despite inflicting four times as many deaths (Pape, 1996, p. 343). Not even casualties an order of magnitude higher still could coerce Germany or Japan to surrender. British and American urban area bombing killed more than 300,000 Germans, but the Allies had to fight all the way to Berlin before Germany capitulated (Pape, 1996, p. 272).22 Against Japan, the incendiary campaign conducted by Twenty-first Bomber Command under the leadership of General Curtis LeMay burned down 178 square miles of 66 Japanese cities and towns, killing between 241,000 and 330,000 Japanese in a mere 5 months (Werrell, 1996, p. 227). Eighty thousand were killed in a single night when Tokyo was firebombed for the first time on March 9 (Werrell, 1996, p. 163). Yet the incendiary campaign contributed next to nothing to Japan’s surrender (Pape, 1996, pp. 118–120, 129–131). Even the effect of the atomic bombings of Hiroshima and Nagasaki on Tokyo’s capitulation is contested, with numerous scholars emphasizing the impact of the Soviet invasion of Manchuria (Hasegawa, 2005; Pape, 1996).

Although no coercive strategic bombing campaign since 1945 has equaled those of World War II in intensity, they have all achieved about the same result.23 Pape classified aspects of Operation Rolling Thunder (1965–1968) during the Vietnam War as punishment campaigns even though most targets were military, industrial, or transportation. Whatever it was, Rolling Thunder failed to compel North Vietnam to abandon its objective of conquering South Vietnam.24 The Iran–Iraq War (1980–1989) was characterized by periodic “wars of the cities,” during which both sides launched air attacks on economic and civilian targets in urban areas. Iraq gained the upper hand in these exchanges after its acquisition of longer-range missiles in 1988, a development that led some analysts to claim that “strategic bombing . . . and particularly the bombing of population centers, played the key role in ending the war” (Karsh, 1989, p. 217). No concrete evidence supports this conclusion, however, and it ignores the complete reversal of fortunes on the battlefield in 1988 as specially trained Iraqi divisions using combined arms operations pushed Iranian forces out of Iraqi territory (Talmadge, 2015). Gary Sick concluded that when Iran accepted the UN ceasefire resolution in July 1989, “the key factor was probably the changed conditions at the battlefront” (Sick, 1989, p. 242). Post–Cold War U.S. bombing campaigns—against Iraq in 1991 and Yugoslavia in 1999—have sought to affect civilian morale without inflicting civilian casualties, but these “punishment lite” campaigns have proved no more successful than their more brutal counterparts (Thomas, 2006). The failure of Russia’s blitzkrieg on Kyiv in February 2022 quickly prompted Moscow to shift to punishment bombing of Ukrainian civilians and civilian infrastructure (Downes, 2022), but this has proved just as unsuccessful as Russia’s ground offensives at winning the war.

Pape’s work, although it collects all cases of strategic bombing since its inception, relied on historical cases to trace the ineffectiveness of punishment and the efficacy (under certain conditions) of denial. Pape’s findings were subsequently confirmed in a quantitative study. Horowitz and Reiter (2001) found no correlation between civilian vulnerability and successful coercion in strategic bombing campaigns from 1917 to 1999. Military vulnerability, conversely, is strongly associated with compellence success.

Pape’s arguments for the futility of punishment bombing are concerned mainly with conventional bombing’s inability to inflict sufficient costs to overcome the goals target states aim to achieve in war. One of Pape’s mechanisms, however, concerns how bombing affects the attitudes of the bombed. Rather than turn the bombed population against its government, Pape argued, it generates hatred for the enemy instead, and may even strengthen civilian morale (Pape, 1996, pp. 24–27).

Evidence from World War II Germany offers only mixed support for this supposition. Bombing appears to have inspired many and diverse emotions among the bombed, including helplessness, fatalism, apathy, fear, guilt, anger, hatred, and desires for revenge. One interesting dynamic is the coexistence of anger among ordinary Germans against their own government and demands for severe reprisals against the enemy. Historians have found little evidence of general animus against the British. As one German woman put it in 1943, “Despite everything that we have suffered in the attacks, there’s not much hatred in Hamburg for the enemy” (quoted in Evans, 2010, p. 448). On the contrary, studies have found that bombing caused German civilians to direct their ire against Nazi officials and the Party itself. According to Evans (2010, p. 447),

Popular anger was not directed against the British for their “terror raids” . . . but against Göring and the German air force, which had patently failed to defend the homeland, and against the Nazi party, which had brought this destruction on Germany.

(See also Beck, 1986, p. 75)

Unfortunately, bombing increased the dependence of ordinary Germans on their government, and Nazi repression made an uprising futile. As Overy (2013, p. 309) put it, “What bombing did do was to increase the dependence of the population on both the state apparatus and the party organizations responsible for welfare, reducing even further the space for more serious dissent.” Indeed, the threat of government reprisals effectively stifled revolutionary sentiments “given the willingness of the regime to impose severe punishment on any open or dangerous forms of protest” (Overy, 2013, pp. 274–275). In the end, however, day-to-day concerns dominated thoughts of revolt: “Most Germans affected by the bombing were too busy trying to survive amidst the ruins, to reconstruct their shattered homes and disrupted lives, and to find ways to avoid getting killed to bother about things like revolution” (Evans, 2010, p. 463).

This evidence of lack of animus against the enemy sits uneasily alongside popular demands for revenge. According to Friedrich (2006, p. 427) among Germans, “Feelings of hatred rolled over the population. ‘There is almost unanimous agreement among the national comrades in their demand that the British be exterminated. Revenge against England cannot be strong enough.’” Similarly, one man from Essen said, “the hour of vengeance is now the only hope of the bomb victims” (quoted in Beck, 1986, p. 75). In Britain, there were similar sentiments. As one resident of Coventry put it in 1940, “It is now time for our deepest, most inspired anger. The whole of Coventry cries: ‘BOMB BACK, AND BOMB HARD’” (quoted in Holman, 2012, p. 394). The experience of being bombed, therefore, may cause civilians to loathe their own government and demand that it hit the enemy back harder at the same time.

Civilian Victimization as a Denial Strategy

Finally, as noted, coercive civilian victimization does not always follow the logic of punishment; in some cases, it aims to have coercive effects by weakening an opponent militarily—the logic of denial. The most common form that this takes in conventional wars is scorched earth campaigns. Such campaigns aim to devastate an area—destroying all livestock and foodstuffs, as well as dwellings, railroads, and any form of industry. Scorched earth can be used by both attackers and defenders. When wielded by attackers, scorched earth is used to destroy enemy supplies and means of waging war. Examples include the chevauchées of the Hundred Years’ War (1337–1453), Sherman’s March to the Sea and Sheridan’s devastation of the Shenandoah Valley during the American Civil War (1861–1865), and the U.S. Army’s destruction of Native American crops and animal herds in the numerous 19th-century wars against Native American tribes. Among the goals of the chevauchées of English King Edward III, for instance, was to compel an opponent who wished to avoid battle to fight and secondly to inflict “economic attrition” by destroying the adversary’s economic base (Rogers, 2002, pp. 56–57). At least one historian has argued that this violence was effective: “Noncombatants were made the victims of attack because attacking noncombatants worked, because it could facilitate the conquest of a fortified region, compel an enemy to do battle in the field, or hamstring him economically or politically” (Rogers, 2002, p. 62).

Scorched earth can also be used for defensive purposes. The French devastation of the Palatinate (1688–1689) during the War of the League of Augsburg was meant to block an invasion of France from the east. The logic of the approach was spelled out in a letter from Louis XIV’s military adviser Louvois to the French commander in December 1688: “His Majesty recommends to you to completely ruin . . . all the places that you leave along the lower and upper Neckar so that the enemy, finding no forage or food whatever, will not try to approach there” (quoted in Lynn, 2002, p. 83). Lynn argued that the strategy backfired in part because “French barbarity . . . helped to forge a European alliance against France . . . . [a]nd fueled a bitter animosity within the German people against the French that festered for centuries” (Lynn, 2002, p. 88).

A more recent example of scorched earth for defensive purposes is the U.S. strategic bombing campaign in North Korea following China’s intervention in November 1950. As the Chinese advance threatened to drive UN forces out of North Korea, General Douglas MacArthur, commander of UN forces, reportedly told the U.S. ambassador to South Korea that “he intended to turn the narrow stretch of territory between UN lines and the border into a ‘desert’ incapable of supporting Communist troops” (quoted in Downes, 2018, p. 214). “As UN units retreated,” Downes writes (Downes, 2018, p. 217), “US airpower implemented a kind of aerial scorched earth policy behind them.” The incendiary campaign eventually destroyed every city, town, and village of any size in North Korea. Although the broader intent of campaigns like these is to hamper an enemy’s military operations, their immediate effect is to inflict hardship and death on civilians, causing untold civilian suffering (see also Conway-Lanz, 2006, pp. 83–121). Clearly in this case coercive civilian victimization by denial failed to halt the Chinese advance.

Quantitative Evidence

There are only a handful of systematic assessments of the effectiveness of civilian victimization in conventional wars. Downes and Cochran (2010), for example, divided victimization into coercive and eliminationist types and assessed their effects on the likelihood of victory. Their statistical analysis found that coercive victimization has no relationship to war outcomes.25 The authors cautioned, however, that selection bias could be skewing the results. The literature on the causes of coercive victimization has found that it is implemented when states are in desperate straits in an attempt to reduce their own losses or compel the adversary to concede (Downes, 2006, 2008). It is thus possible that the positive effect of coercive victimization could be masked by the difficult circumstances in which it is used.

In a follow-up study, Downes and Cochran (2018) implemented the statistical techniques necessary to test whether selection bias reduced the effect of coercive victimization in their earlier study. Downes and Cochran used matching to locate highly similar cases in which belligerents faced desperate circumstances yet did not target civilians. The results of this analysis found that selection bias did not explain the null effect of coercive victimization on victory. This result constitutes important evidence for the ineffectiveness of coercive victimization in conventional interstate wars.

In sum, it is hard to draw firm conclusions from the evidence on the effectiveness of the different forms of coercive civilian victimization. What is known, however, suggests that coercion by punishment—with the possible exception of post-siege massacres—is ineffective at obtaining its ends. Although there is anecdotal evidence of success of coercive victimization by denial in the distant past, there is no systematic evidence on the effectiveness of scorched earth or devastation strategies.

Eliminationist Victimization

Eliminationist victimization is a strategy intended to alter the demographic balance in a given territory in favor of one’s own group by eliminating members of other groups. Echoing the prior discussion of irregular wars, eliminationism can be done more or less discriminately. In the Spanish Civil War, for example, Balcells (2017) showed that in the wake of territorial acquisitions by the Nationalists or Republicans, violence was used selectively to eliminate strong enemy supporters but the most killings were carried out in areas where violence could tip the local demographic balance one way or the other. In municipalities where the Nationalists (for example) were already demographically preponderant, violence was unnecessary; in municipalities where they were vastly outnumbered, however, a vast number of killings would have been required to change the balance. Therefore, killings were most prevalent where the Left–Right balance was relatively even and decreased as the balance moved away (in either direction) from the middle.

The less discriminate—and more recognizable—form of eliminationist victimization is cleansing. The aim of cleansing is to radically reduce the population of a particular group—defined in ethnic or ideological terms—in a given piece of territory so as to facilitate the perpetrator gaining secure, enduring control over the area, which has often just been seized from the enemy.26 Cleansing accomplishes this by using homicides, expulsions, destruction of property, and other forms of violence to induce mass flight. The goal is not typically to kill everyone in the targeted group, but to kill enough to cause the rest to flee and never return. Civilians who share the identity of the enemy state are commonly perceived to be of questionable loyalty—a “fifth column” poised to rebel if given the opportunity. As Downes (2008, p. 36) wrote,

Believing that there is little or no possibility of gaining the support of members of the opposing group, belligerents . . . attack the other group’s civilians to avoid the risk—if not always the actuality—of being attacked themselves.

Removing these people not only removes the threat, it strengthens one’s own claim to the territory. Although individual perpetrators may be motivated by some blend of nationalism, racism, hatred, or greed, cleansing is a purposeful strategy to establish lasting control over territory.27

In some cases, a government may enact cleansing in an interstate war inside its own borders against a population it views as an actual or potential fifth column. Examples of “domestic” cleansing include the Armenian deportations in the Ottoman Empire during World War I, the expulsion of most of western Anatolia’s Greek population during the Greco-Turkish War (1919–1922), and Stalin’s deportations of the Chechens and other ethnic groups from the Caucasus during World War II.

How should the effectiveness of eliminationist victimization be measured? One possibility is to look at the effect of such strategies on victory. In an analysis of eliminationist victimization in interstate wars, Downes and Cochran (2010) found that it increased the likelihood that the perpetrating actor would win the war. Given that eliminationist victimization is possible only when a belligerent has entered enemy territory—and thus when it is already “winning”—it is possible that eliminationist victimization could be endogenous to victory. To see if this was the case, Downes and Cochran (2018) implemented an instrumental variables strategy using wars fought over territory as an instrument for eliminationist victimization. The results showed that eliminationist victimization was no longer associated with victory after accounting for endogeneity.

As Downes and Cochran (2018) admitted, however, the effectiveness of eliminationist victimization is probably not best assessed by looking at its impact on the likelihood of victory and defeat. Rather, because the goal of this strategy is secure and lasting control over territory and/or political dominance in an area, the effectiveness of eliminationist victimization is best gauged by studying its aftermath. There are at least four possible metrics. The first and most immediate is whether the violence actually changed the demographic balance between the two groups in the targeted region. A second is whether enemy attacks increased or decreased in the aftermath. For example, is cleansing followed by partisan warfare launched by remnants of the targeted population, or does the adversary’s military strike back in an attempt to recapture the lost territory? Once the war is over, and assuming the side doing the eliminating wins, a third longer-term measure of effectiveness is whether eliminationist victimization solidifies the demographic dominance of the winning side. This could be measured by the onset of insurgency or terrorism in the cleansed area or by a new war launched by the state of the cleansed population. An example of the latter is Azerbaijan’s 2020 attack on Armenia, which aimed to take back territory occupied by Armenia in the previous war (1992–1994). In a democracy, demographic dominance after victimization of the kind described by Balcells (2017) could be investigated by whether wartime killings are associated with electoral victories by candidates from the winning side (for a related dynamic in civil wars, see Daly, 2022). Finally, the effectiveness of eliminationist victimization could be measured by observing whether it triggers third-party intervention against the perpetrator, as in Bosnia or Kosovo.

There is plentiful evidence that cleansing succeeds in reducing the targeted population (metric #1) and sometimes results in permanent changes in the demographic balances of the states involved (metric #3). During the First Balkan War (1912–1913), for example, Bulgaria, Serbia, and Greece massacred and expelled large numbers of Turks (International Commission, 1993; McCarthy, 1995, pp. 135–177). After documenting scores of massacres of Muslims and the destruction of hundreds of villages in the first war, McCarthy wrote,

The obvious intention of those who murdered Muslims and forced their exodus was to “de-Turkify” the Balkans . . . . Balkan Christians pursued policies that would insure that Muslim refugees would not return and that those who had not left would do so.

(McCarthy, 1995, p. 148)

In this endeavor, they were remarkably successful: comparing population estimates from before and after the wars, McCarthy found that the Muslim population of the areas conquered by Bulgaria, Greece, and Serbia declined by 62%, or more than 1.4 million people (McCarthy, 1995, pp. 162–164). Other cleansing campaigns have resulted in even more complete elimination of the targeted population. Between 75% and 80% of the Arab population living within what became the state of Israel fled during the Palestine War (1947–1949; Morris, 2004, pp. 588–589, 604).28 Describing the results of the Greco-Turkish War (1919–1922), which ended with the burning of Smyrna and the evacuation of 150,000–200,000 Greeks, Naimark wrote, “The nearly 3,000-year history of the Greek presence on the Aegean coast of Anatolia came to an abrupt conclusion” (Naimark, 2001, p. 52). The German presence in Central and Eastern Europe was similarly ended by the expulsion of 11 million–12 million Germans at the end of World War II, and the Drina Valley in eastern Bosnia was virtually emptied of Bosnian Muslims during the Bosnian War.

The examples in the preceding paragraph, however, are a biased sample. Researchers would need to compile a list of all cleansing attempts—not just successes—to arrive at a real estimate of relative rates of success and failure. Little research investigates the immediate effects of cleansing (metric #2) or how often cleansing draws intervention (metric #4).

Avenues of Future Research on Civilian Victimization During Conventional Wars

Given that the literature on the effectiveness of civilian victimization in conventional wars is even less developed than the underdeveloped literature on effectiveness in irregular wars, questions for further research are relatively basic. Possibly the only finding that enjoys anything close to a scholarly consensus is that punishment bombing doesn’t work. However, little information is available about how bombing affects the attitudes of the bombed and whether bombing has longer-term consequences for localities or societies.

With regard to the effectiveness of other forms of coercive victimization, the state of knowledge is very limited and thus basic research is needed to answer questions like how often do sieges and blockades succeed? Are there conditions under which they are more or less successful? When they succeed, is civilian deprivation the key cause? When attackers destroy cities that resist does that deter other cities from resisting? Just as denial is the more effective strategy of coercion, is civilian victimization that hinders an adversary’s military strategy more effective than strategies that inflict pain on civilians in order to sap their will to resist?

With regard to eliminationist victimization, scholars know essentially nothing other than that there are examples where states have successfully carried out ethnic cleansing. Yet there is no indication of how representative those cases are of the broader universe of cleansing attempts, whether cleansing succeeds at reducing violence in the short or long term, or how often it triggers third-party intervention intended to reverse it.

In attempting to answer all of these questions, scholars will need to be attuned to problems of selection bias. Studies should pay attention to the context in which strategies of victimization are implemented. Do prevailing theories, for example, predict that civilians will be targeted primarily when belligerents are losing, and hence those strategies may be effective and yet still fail? Or do states attempt ethnic cleansing only against geographically concentrated groups that reside close to an international border and when recipient states or interested parties are too weak or distracted to react? Factors that affect the conditions under which strategies of civilian victimization are used may also affect how successful they are.

Conclusion: The Ethics of Studying the Effectiveness of Civilian Victimization

A summary of important debates and key findings in the literature on the effectiveness of civilian victimization in both irregular and conventional war produces few definitive answers. The bumper sticker version of the findings—“violence is effective when selective” in irregular war and “punishment never works” in conventional war—obscures more than it reveals. Few studies have verified the effectiveness of selective violence, and multiple studies have now found that indiscriminate violence can be effective. Similarly, little is known about punishment outside of strategic bombing and next to nothing about whether eliminationist violence works. There is much yet to be discovered in this field.

But what about the ethics of this kind of work? For academics, the ethics and morality of studying the effectiveness of violence is at least as old as the Milgram and Zimbardo experiments (Milgram, 1975; Zimbardo, 2007). How can scholars conduct such research and be confident that their findings are not put to use in real conflicts?29 The easy way out is if the research findings comport with our moral beliefs, as when studies find that indiscriminate bombing in counterinsurgency is counterproductive, causing people to shift their support to the rebels (Dell & Querubin, 2018; Kocher et al., 2011). Scholars can then say they are simply demonstrating that acting immorally is also ineffective. But when the two clash, things are not so simple. The flip side of the finding that indiscriminate violence does not work, after all, is that selective violence does work (Kalyvas, 2006). Even worse, other studies have found that indiscriminate violence is effective under certain conditions (Downes, 2007; Downes et al., n.d.; Lyall, 2009; Zhukov, 2015, 2023).

However, there are at least three major reasons why ignoring the effectiveness of violence strategies may be harmful. First, assuming rather than demonstrating that unethical violence is ineffective may lead policymakers to draw the wrong conclusions from past conflicts when attempting to identify more ethical strategies. For instance, historical research demonstrates that a selective reading of past counterinsurgency cases that exaggerated the effectiveness of “hearts and minds” methods while downplaying frequent use of indiscriminate violence contributed to naive expectations about the success of contemporary counterinsurgency campaigns (Birtle, 1998; Hack, 2021). If significant levels of civilian victimization played a role in past successes, policymakers must either identify new, more ethical, means of solving the same strategic problems or more seriously weigh the value of intervention.

Second, many of the strategies of civilian victimization discussed here have been widely and extensively used, with many highly publicized cases already offering plenty for armed actors to study. In addition, in many cases, armed actors have proved quite capable of learning about the effectiveness of different strategies of civilian victimization over time without any help from academics.30 This should diminish concerns that armed actors benefit from scholarship investigating whether violence works. Instead, it is argued that studying the effectiveness of civilian victimization is far more important for updating scholars and policymakers seeking to mitigate violence than it is for armed actors. Understanding whether something works is crucial for developing the best response to prevent, counteract, or defeat it. For instance, given that armed actors often use civilian victimization strategies when desperate (Downes, 2008), the development of strong norms limiting civilian victimization becomes even more important when unethical strategies prove effective. Such efforts will likely prove more useful than simply hoping that such strategies will fail.

Finally, identifying the conditions under which types of violence “work” is important not only for understanding why armed actors may continue to use certain types of violence, but also for its implications for other areas of conflict research. For example, the question of whether state indiscriminate violence is ever effective goes a long way toward explaining whether mobilization or resiliency is more important for explaining insurgent survival. Does indiscriminate violence fuel an insurgency by accelerating mobilization or do certain insurgent groups develop means to mitigate the damage and survive? Similarly, if Gilbert (2022) was correct that selective violence comes with long-term reputational costs, then that opens new questions about the trade-off between violence to ensure civilian collaboration and the provision of goods and services. These are only a few examples of how addressing debates on the effectiveness of civilian victimization will help advance other areas of conflict research.


  • Abrahms, M. (2006). Why terrorism does not work. International Security, 31(2), 42–78.
  • Abrahms, M. (2012). The political effectiveness of terrorism revisited. Comparative Political Studies, 45(3), 366–393.
  • Abrahms, M., & Potter, P. B. K. (2015). Explaining terrorism: Leadership deficits and militant group tactics. International Organization, 69(2), 311–342.
  • Almosawa, S. (2022, March 16). As U.S. focuses on Ukraine, Yemen starves. The Intercept.
  • Balcells, L. (2012). The consequences of victimization on political identities: Evidence from Spain. Politics & Society, 40(3), 309–345.
  • Balcells, L. (2017). Rivalry and revenge: The politics of violence during civil war. Cambridge University Press.
  • Balcells, L., & Stanton, J. (2021). Violence against civilians during armed conflict: Moving beyond the macro- and micro-level divide. Annual Review of Political Science, 24, 45–69.
  • Balcells, L., & Sullivan, C. (2018). New findings from conflict archives: An introduction and methodological framework. Journal of Peace Research, 55(2), 137–146.
  • Baldick, R. (1964). The siege of Paris. History Book Club.
  • Barma, N. H., & Goldgeier, J. (2022). How not to bridge the gap in international relations. International Affairs, 98(5), 1763–1781.
  • Beck, E. R. (1986). Under the bombs: The German home front, 1942–1945. University Press of Kentucky.
  • Beehner, L., Berti, B., & Jackson, M. T. (2017). The strategic logic of sieges in counterinsurgencies. Parameters, 47(2), 77–86.
  • Bell, A. M. (2022). Combatant socialization and norms of restraint: Examining officer training at the U.S. Military Academy and Army ROTC. Journal of Peace Research, 59(2), 180–196.
  • Biddle, S. (2021). Nonstate warfare: The military methods of guerrillas, warlords, and militias. Princeton University Press.
  • Birtle, A. J. (1998). U.S. army counterinsurgency and contingency operations doctrine, 1860–1941. Center of Military History.
  • Blair, C. W. (2022). Restitution or retribution? Detainee payments and insurgent violence. Journal of Conflict Resolution, 66(7–8), 1356–1392.
  • Bloom, M. (2005). Dying to kill: The allure of suicide terrorism. Columbia University Press.
  • Bradbury, J. (1992). The medieval siege. Boydell Press.
  • Bulutgil, Z. H. (2016). The roots of ethnic cleansing in Europe. Cambridge University Press.
  • Cohen, D. K. (2016). Rape during civil war. Cornell University Press.
  • Condra, L., & Shapiro, J. N. (2012). Who takes the blame? The strategic effects of collateral damage. American Journal of Political Science, 56(1), 167–187.
  • Conway-Lanz, S. (2006). Collateral damage: Americans, noncombatant immunity, and atrocity after World War II. Routledge.
  • Costalli, S., & Ruggeri, A. (2015). Indignation, ideologies, and armed mobilization: Civil war in Italy, 1943–45. International Security, 40(2), 119–157.
  • Cunningham, K., Bakke, K., & Seymour, L. (2012). A plague of initials: Fragmentation, cohesion, and infighting in civil wars. Perspectives on Politics, 10(2), 265–283.
  • Daly, S. Z. (2022). Violent victors: Why bloodstained parties win postwar elections. Princeton University Press.
  • Davis, R. G. (1995). German rail yards and cities: U.S. bombing policy, 1944–1945. Air Power History, 42(2), 46–63.
  • De la Cueva. (1998). Religious persecution, anticlerical tradition, and revolution: On atrocities against the clergy during the Spanish Civil War. Journal of Contemporary History, 33(3), 355–369.
  • Dell, M., & Querubin, P. (2018). Nation building through foreign intervention: Evidence from discontinuities in military strategies. The Quarterly Journal of Economics, 133(2), 701–764.
  • Dixon, R. (2020, November 11). Azerbaijan’s drones owned the battlefield in Nagorno-Karabakh—And showed future of warfare. Washington Post.
  • Douhet, G. (1942). The command of the air (D. Ferrari, Trans.). Coward-McCann. (Original work published 1921)
  • Downes, A. B. (2006). Desperate times, desperate measures: The causes of civilian victimization in war. International Security, 30(4), 152–196.
  • Downes, A. B. (2007). Draining the sea by filling the graves: Investigating the effectiveness of indiscriminate violence as a counterinsurgency strategy. Civil Wars, 9(4), 420–444.
  • Downes, A. B. (2008). Targeting civilians in war. Cornell University Press.
  • Downes, A. B. (2012). Military culture and civilian victimization: The allied strategic bombing of Germany in World War II. In D. Rothbart, K. V. Korostelina, & M. Cherkaoui (Eds.), Civilians and modern war: Armed conflict and the ideology of violence (pp. 72–95). Routledge.
  • Downes, A. B. (2018). Creating a cordon sanitaire: U.S. strategic bombing and civilian victimization in the Korean War. In A. Barros & M. Thomas (Eds.), The civilianization of war: The changing civil-military divide, 1914–2014 (pp. 196–220). Cambridge University Press.
  • Downes, A. B. (2022, March 26). Putin’s war against civilians is not new—Nor will it work. The Hill.
  • Downes, A. B., & Cochran, K. M. (2010). Targeting civilians to win? Assessing the military effectiveness of civilian victimization in interstate war. In A. Lawrence & E. Chenoweth (Eds.), Rethinking violence: States and non-state actors in conflict (pp. 23–56). MIT Press.
  • Downes, A. B., & Cochran, K. M. (2018). It’s a crime, but is it a blunder? Investigating the military effectiveness of civilian victimization. In N. Foote & N. Williams (Eds.), Civilians and warfare in world history (pp. 288–312). Routledge.
  • Downes, A. B., & Greenhill, K. G., & Rangazas, R. (n.d.). Controlling hearts and minds: Population concentration in counterinsurgency operations. George Washington University.
  • Dube, O., & Vargas, J. F. (2013). Commodity price shocks and civil conflict: Evidence from Colombia. The Review of Economic Studies, 80(4[285]), 1384–1421.
  • Epstein, R. A., & Kaplan, O. (n.d.). Responsible policy engagement: A new imperative for scholars. University of Denver.
  • Evans, R. J. (2010). The Third Reich at war, 1939–1945. Penguin Books.
  • Fazal, T. M., & Greene, B. C. (2015). A particular difference: European identity and civilian targeting. British Journal of Political Science, 45(4), 829–851.
  • Feuer, A. (2023). Environmental warfare tactics in irregular conflicts. Perspectives on Politics, 21(2), 533–549.
  • Fortna, V. (2015). Do terrorists win? Rebels’ use of terrorism and civil war outcomes. International Organization, 69(3), 519–556.
  • Frank, R. B. (1999). Downfall: The end of the Imperial Japanese Empire. Random House.
  • Friedrich, J. (2006). The fire: The bombing of Germany, 1940-1945. Columbia University Press.
  • García-Ponce, O., Young, L. E., & Zeitzoff, T. (2023). Anger and support for retribution in Mexico’s drug war. Journal of Peace Research, 60(2), 274–290.
  • Gilbert, D. (2022). The logic of kidnapping in civil war: Evidence from Colombia. American Political Science Review, 116(4), 1226–1241.
  • Grebler, L., & Winkler, W. (1940). The cost of the World War to Germany and to Austria-Hungary. Yale University Press.
  • Grousset, R. (1970). The empire of the steppes: A history of Central Asia. Rutgers University Press.
  • Hack, K. (2021). The Malayan emergency: Revolution and counterinsurgency at the end of empire. Cambridge University Press.
  • Harff, B. (2003). No lessons learned from the Holocaust? Assessing risks of genocide and political mass murder since 1955. American Political Science Review, 97(1), 57–73.
  • Hasegawa, T. (2005). Racing the enemy: Stalin, Truman, and the surrender of Japan. Harvard University Press.
  • Hassner, R. (2022). Anatomy of torture. Cornell University Press.
  • Hoffman, B. (2006). Inside terrorism (2nd ed.). Columbia University Press.
  • Holman, B. (2012). “Bomb back, and bomb back hard”: Debating reprisals during the Blitz. Australian Journal of Politics & History, 58(3), 394–407.
  • Hoover-Green, A. (2018). The commander’s dilemma: Violence and restraint in wartime. Cornell University Press.
  • Horne, A. (1965). The fall of Paris: The siege and the commune, 1870–71. St. Martin’s Press.
  • Horowitz, M., & Reiter, D. (2001). When does aerial bombing work? Quantitative empirical tests, 1917–1999. Journal of Conflict Resolution, 45(2), 147–173.
  • Hultman, L. (2007). Battle losses and rebel violence: Raising the costs of fighting. Terrorism and Political Violence, 19, 205–222.
  • International Commission to Inquire into the Causes and Conduct of the Balkan Wars. (1993). The other Balkan wars. Carnegie Endowment for International Peace.
  • Jackson, M., Beehner, L., & Berti, B. (2016, December 7). Modern siege warfare: How it is changing counterinsurgency. Foreign Affairs.
  • Jervis, R. (1989). The meaning of the nuclear revolution: Statecraft and the prospect of Armageddon. Cornell University Press.
  • Jo, H. (2015). Compliant rebels: Rebel groups and international law in world politics. Cambridge University Press.
  • Kalyvas, S. (2006). The logic of violence in civil war. Cambridge University Press.
  • Kalyvas, S., & Balcells, L. (2010). International system and technologies of rebellion: How the end of the Cold War shaped internal conflict. American Political Science Review, 104(3), 415–429.
  • Karsh, E. (1989). Lessons of the Iran–Iraq war. Orbis, 33(2), 209–223.
  • Kocher, M., & Kalyvas, S. (2007). How “free” is free riding in civil wars? Violence, insurgency, and the collective action problem. World Politics, 59(2), 177–216.
  • Kocher, M., Pepinsky, T., & Kalyvas, S. (2011). Aerial bombing and counterinsurgency in the Vietnam War. American Journal of Political Science, 55(2), 201–218.
  • Kramer, A. (2022, December 18). Overnight drone attacks target Kyiv and two other regions of Ukraine. New York Times.
  • Kranzberg, M. (1950). The siege of Paris, 1870–1871: A political and social history. Cornell University Press.
  • Kydd, A. H., & Walter, B. F. (2006). The strategies of terrorism. International Security, 31(1), 49–80.
  • Lichtenheld, A. G. (2020). Explaining population displacement strategies in civil wars: A cross-national analysis. International Organization, 74, 253–294.
  • Lyall, J. (2009). Does indiscriminate violence incite insurgent attacks? Evidence from Chechnya. Journal of Conflict Resolution, 53(3), 331–362.
  • Lyall, J. (2019). Civilian casualties, humanitarian aid, and insurgent violence in civil wars. International Organization, 73(4), 901–926.
  • Lyall, J., Blair, G., & Imai, K. (2013). Explaining support for combatants during wartime: A survey experiment in Afghanistan. American Political Science Review, 107(4), 679–705.
  • Lynn, J. A. (2002). A brutal necessity? The devastation of the Palatinate, 1688–1689. In M. Grimsley & C. J. Rogers (Eds.), Civilians in the path of war (pp. 79–110). University of Nebraska Press.
  • Maeda, T. (2009). Strategic bombing of Chongqing by Imperial Japanese army and naval forces. In Y. Tanaka & M. B. Young (Eds.), Bombing civilians: A twentieth-century history (pp. 135–153). New Press.
  • Magaloni, B., Franco-Vivanco, E., & Melo, V. (2020). Killing in the slums: Social order, criminal governance, and police violence in Rio de Janeiro. American Political Science Review, 114(2), 552–572.
  • Magaloni, B., Robles, G., Matanock, A. M., Diaz-Cayeros, A., & Romero, V. (2020). Living in fear: The dynamics of extortion in Mexico’s drug war. Comparative Political Studies, 53(7), 1124–1174.
  • Manekin, D. (2020). Regular soldiers, irregular war: Violence and restraint in the Second Intifada. Cornell University Press.
  • Mason, T. D., & Krane, D. (1989). The political economy of death squads: Toward a theory of the impact of state-sanctioned terror. International Studies Quarterly, 33(2), 175–198.
  • McCarthy, J. (1995). Death and exile: The ethnic cleansing of Ottoman Muslims, 1821–1922. Darwin Press.
  • Melzer, N. (2009). Interpretive guidance on the notion of direct participation in hostilities under international humanitarian law. International Committee of the Red Cross.
  • Midlarsky, M. I. (2005). The killing trap: Genocide in the twentieth century. Cambridge University Press.
  • Milgram, S. (1975). Obedience to authority: An experimental view. Harper & Row.
  • Moghadam, A. (2008). The globalization of martyrdom: Al Qaeda, Salafi Jihad, and the diffusion of suicide attacks. Johns Hopkins University Press.
  • Morris, B. (2004). The birth of the Palestinian refugee problem revisited. Cambridge University Press.
  • Morrow, J. D. (2014). Order within anarchy: The laws of war as an international institution. Cambridge University Press.
  • Moyer, L. (1995). Victory must be ours: Germany in the Great War 1914–1918. Leo Cooper.
  • Naimark, N. (2001). Fires of hatred: Ethnic cleansing in twentieth-century Europe. Harvard University Press.
  • Newton, C., & Tucker, C. (2022). The efficacy of airpower in counterinsurgency. Security Studies, 31(2), 218–250.
  • Offer, A. (1989). The First World War: An agrarian interpretation. Clarendon Press.
  • Overy, R. J. (2013). The bombers and the bombed: Allied air war over Europe, 1940–1945. Viking Penguin.
  • Pape, R. A. (1996). Bombing to win: Air power and coercion in war. Cornell University Press.
  • Pape, R. A. (2003). The strategic logic of suicide terrorism. American Political Science Review, 97(3), 343–361.
  • Pape, R. A. (2005). Dying to win: The strategic logic of suicide terrorism. Random House.
  • Pechenkina, A. O., Bausch, A. W., & Skinner, K. K. (2019). How do civilians attribute blame for state indiscriminate violence? Journal of Peace Research, 56(4), 545–558.
  • Petersen, R. D. (2002). Understanding ethnic violence: Fear, hatred, and resentment in twentieth-century Eastern Europe. Cambridge University Press.
  • Prawdin, M. (1940). The Mongol empire, its rise and legacy. G. Allen & Unwin.
  • Richardson, J. (1982). Paris under Siege: A Journal of the Events of 1870-1871 kept by Contemporaries and Translated and Presented by Joanna Richardson. Folio Society.
  • Rigterink, A. S. (2020). Diamonds, rebel’s and farmer’s best friend: Impact of variation in the price of a lootable, labor-intensive natural resource on the intensity of violent conflict. Journal of Conflict Resolution, 64(1), 90–126.
  • Rogers, C. J. (2002). By fire and sword: Bellum hostile and “civilians” in the Hundred Years’ War. In M. Grimsley & C. J. Rogers (Eds.), Civilians in the path of war (pp. 33–78). University of Nebraska Press.
  • Rozenas, A., Schutte, S., & Zhukov, Y. (2017). The political legacy of violence: The long-term impact of Stalin’s repression in Ukraine. Journal of Politics, 79(4), 1147–1161.
  • Ruiz, J. (2007). Defending the republic: The Garcia Atadell brigade in Madrid, 1936. Journal of Contemporary History, 42(1), 97–115.
  • Salehyan, I., Siroky, D., & Wood, R. M. (2014). External rebel sponsorship and civilian abuse: A principal-agent analysis of wartime atrocities. International Organization, 68(3), 633–661.
  • Sanín, F. G., & Wood, E. J. (2014). Ideology in civil war: Instrumental adoption and beyond. Journal of Peace Research, 51(2), 213–226.
  • Schelling, T. (1966). Arms and influence. Yale University Press.
  • Schirmer, J. G. (1998). The Guatemalan military project: A violence called democracy. University of Pennsylvania Press.
  • Schubiger, L. (2021). State violence and wartime civilian agency: Evidence from Peru. Journal of Politics, 83(4), 1383–1398.
  • Schutte, S. (2017). Violence and civilian loyalties: Evidence from Afghanistan. Journal of Conflict Resolution, 61(8), 1595–1625.
  • Schwartz, R. A., & Straus, S. (2018). What drives violence against civilians in civil war? Evidence from Guatemala’s conflict archives. Journal of Peace Research, 55(2), 222–235.
  • Shandroff, G. (1972). The evolution of area bombing in American doctrine and practice [Doctoral dissertation, New York University].
  • Shapiro, J. (2013). The terrorist’s dilemma: Managing violent covert organizations. Princeton University Press.
  • Shaver, A., & Shapiro, J. N. (2021). The effect of civilian casualties on wartime informing: Evidence from the Iraq War. Journal of Conflict Resolution, 65(7–8), 1337–1377.
  • Sheline, A. (2023). Ending counterproductive U.S. involvement in Yemen. Quincy Institute.
  • Sick, G. (1989). Trial by error: Reflections on the Iran-Iraq War. Middle East Journal, 43(2), 230–245.
  • Silverman, D. (2020). Too late to apologize? Collateral damage, post-harm compensation, and insurgent violence in Iraq. International Organization, 74(4), 853–871.
  • Souleimanov, E. A., & Siroky, D. S. (2016). Random or retributive? Indiscriminate violence in the Chechen Wars. World Politics, 68(4), 677–712.
  • Souleimanov, E. A., Siroky, D. S., & Krause, P. (2022). Kin killing: Why governments target family members in insurgency, and when it works. Security Studies, 31(2), 183–217.
  • Stanton, J. A. (2016). Violence and restraint in civil war: Civilian targeting in the shadow of international law. Cambridge University Press.
  • Steele, A. (2017). Democracy and displacement in Colombia’s civil war. Cornell University Press.
  • Straus, S. (2006). The order of genocide: Race, power, and war in Rwanda. Cornell University Press.
  • Straus, S. (2012). Retreating from the brink: Theorizing mass violence and the dynamics of restraint. Perspectives on Politics, 10(2), 343–362.
  • Straus, S. (2015). Making and unmaking nations: War, leadership, and genocide in modern Africa. Cornell University Press.
  • Sullivan, C. M. (2012). Blood in the village: A local-level investigation of state massacres. Conflict Management and Peace Science, 29(4), 373–396.
  • Talmadge, C. (2015). The dictator’s army: Battlefield effectiveness in authoritarian regimes. Cornell University Press.
  • Tarrant, V. E. (1989). The U-boat offensive 1914–1945. Arms & Armor.
  • Terraine, J. (1989). The U-boat wars 1916–1945. G.P. Putnam’s Sons.
  • Thomas, H. (1961). The Spanish civil war. Harper & Brothers.
  • Thomas, J. (2014). Rewarding bad behavior: How governments respond to terrorism in civil war. American Journal of Political Science, 58(4), 804–818.
  • Thomas, W. (2006). Victory by duress: Civilian infrastructure as a target in air campaigns. Security Studies, 15(1), 1–33.
  • Valentino, B. A. (2004). Final solutions: Mass killing and genocide in the 20th century. Cornell University Press.
  • Valentino, B. A., Huth, P., & Croco, S. (2006). Covenants without the sword: International law and the protection of civilians in times of war. World Politics, 58, 339–377.
  • Vincent, C. P. (1985). The politics of hunger: The allied blockade of Germany, 1915–1919. Ohio University Press.
  • Walzer, M. (1977). Just and unjust wars: A moral argument with historical illustrations. Basic Books.
  • Weinstein, J. (2007). Inside rebellion: The politics of insurgent violence. Cambridge University Press.
  • Werrell, K. (1996). Blankets of fire: U.S. bombers over Japan during World War II. Smithsonian Institution.
  • Wiseberg, L. S. (1975). The statistics jungle: Measuring war, plague, famine, and fire. Society, 13(2), 53–60.
  • Wood, E. (2003). Insurgent collective action and civil war in El Salvador. Cambridge University Press.
  • Wood, R. M. (2010). Rebel capability and strategic violence against civilians. Journal of Peace Research, 47(5), 601–614.
  • Wood, R. M. (2014). Opportunities to kill or incentives for restraint? Rebel capabilities, the origins of support, and civilian victimization in civil war. Conflict Management and Peace Science, 31(5), 461–480.
  • Wood, R. M. (2016). Intrastate conflict and civilian victimization. In W. R. Thompson (Ed.), Oxford Research Encyclopedia of Politics. Oxford University Press.
  • Wood, R. M., & Kathman, J. (2014). Too much of a bad thing? Civilian victimization and bargaining in civil war. British Journal of Political Science, 44(3), 685–706.
  • Wood, R. M., & Kathman, J. (2015). Competing for the crown: Inter-rebel competition and civilian targeting in civil war. Political Research Quarterly, 68(1), 167–179.
  • Zhukov, Y. M. (2015). Population resettlement in war: Theory and evidence from Soviet archives. Journal of Conflict Resolution, 59(7), 1155–1185.
  • Zhukov, Y. M. (2017). External resources and indiscriminate violence: Evidence from German-occupied Belarus. World Politics, 69(1), 54–97.
  • Zhukov, Y. M. (2023). Repression works (just not in moderation). Comparative Political Studies, 56(11), 1663–1694.
  • Zimbardo, P. G. (2007). The Lucifer effect: Understanding how good people turn evil. Random House.


  • 1. Recent research on civilian victimization by criminal organizations outside of the context of war is omitted from this article. For recent examples of micro-level studies focusing on causally identifying the effects of criminal violence, see García-Ponce et al. (2023); Magaloni, Franco-Vivanco, et al. (2020); and Magaloni, Robles, et al. (2020).

  • 2. The study of nuclear strategy and deterrence was implicitly the study of mass violence against civilians but was rarely framed as such.

  • 3. Genocide studies, long dominated by work on the Holocaust and mainly the preserve of historians, psychologists, and sociologists, also received new impetus from political scientists working in both subfields (e.g., Harff, 2003; Midlarsky, 2005; Straus, 2006, 2015; Valentino, 2004).

  • 4. A third category, symmetric non-conventional, is left aside because it comprises a small fraction of conflicts and has not been well-studied.

  • 5. A third type of victimization is revenge or retribution, which simply repays violence with violence. Because revenge is driven by emotions, however, its effectiveness is hard to gauge because inflicting harm on the object of revenge is an end in itself. Therefore, the scope of this review is limited to the effectiveness of coercive and eliminationist victimization.

  • 6. For a critique of the conventional-irregular dichotomy, which argues that the two have much more in common than is commonly appreciated, see Biddle (2021).

  • 7. Since Kalyvas wrote, other scholars have introduced an intermediate category of violence. Collective targeting is selective violence at the level of the group but indiscriminate at the level of the individual (Steele, 2017; Straus, 2015). In other words, belligerents target a specific ethnic, religious, or ideological group owing to its presumed sympathy or loyalty to one of the belligerents. Within the group, however, who is targeted for violence is random and is not based on individual actions.

  • 8. More precisely, denial lowers the likelihood that the target will obtain whatever benefits it seeks, which lowers the overall benefit.

  • 9. Steele (2017) offered an example of an eliminationist logic where groups use collective targeting to expel members of another group from communities to consolidate their territorial control. However, the effectiveness of such strategies is yet to be examined.

  • 10. Indiscriminate violence may even undermine support for perpetrators long after the conflict ends (Rozenas et al., 2017). Balcells (2012) similarly found that experiencing “severe” victimization at the hands of one side during the Spanish Civil War increased support for the other side decades later—but severe is not necessarily indiscriminate.

  • 11. A recent study by Newton and Tucker (2022) found some support for both perspectives on indiscriminate violence, showing that air strikes lower insurgent strength in the long term by degrading insurgent capabilities, but civilian casualties may lead to greater insurgent attacks in the short term.

  • 12. For works that take opposing views, see Frank (1999) and Hasegawa (2005).

  • 13. As noted in the section “Introduction,” measuring the effectiveness of retributive victimization is problematic. Moreover, it is rare for this type of violence to rise to the level of policy. One possible exception identified by Balcells (2017) occurs after territory changes hands in conventional wars. In addition to eliminationist targeting of enemy supporters, Balcells found that municipalities where the adversary committed a greater number of killings in a previous period will suffer more killings if the other side takes the area back. She attributed this to desires to avenge the previous killings.

  • 14. The same war can contain more than one of these types. In the Yugoslav wars in the 1990s, the Serbs besieged multiple cities in Croatia (Vukovar, Dubrovnik) and Bosnia (most notably Sarajevo) while also expelling Croats and Bosnian Muslims from conquered territory. One form of victimization that does not easily fit this typology consists of raids, operations in which attackers enter enemy territory, carry out attacks, and then escape back to friendly territory. A guess is that they are typically coercive, using punishment to obtain concessions from the target government.

  • 15. On the difference between coercion and brute force, see Schelling (1966).

  • 16. If forced to take the target by storm, attackers historically often carried out large-scale violence against soldiers and civilians inside. Although this violence technically falls in the category of retributive victimization, conquerors frequently used it purposefully for coercive purposes against nearby cities. Thus, such massacres are discussed under coercive victimization.

  • 17. It is also unclear whether Aleppo and Grozny are cases of successful coercion or sieges ended by assault.

  • 18. The looting and rape that occurred after successful assaults were also considered a form of payment for soldiers in recognition of their suffering during the siege (and sometimes also actual payment for soldiers who had not been paid for long periods).

  • 19. Indeed, such massacres continued—albeit with decreasing frequency—into the 19th century.

  • 20. It is important to note, however, that these deaths cannot be attributed solely to the blockade. Poor German policies in many cases amplified rather than neutralized the shortages produced by the blockade, thereby enhancing its deadliness. Still, Offer (1989, p. 69) concluded, “It is wrong . . . to blame the German food crises on mismanagement alone. Blockade made them almost inevitable.”

  • 21. For statements by German leaders, see Offer (1989, p. 76). For the hypothesis that coercive victimization “works” only when military defeat is inevitable, see Downes and Cochran (2018, p. 305).

  • 22. Although British Bomber Command was the main culprit, the U.S. Army Air Forces conducted nearly 70 urban area raids and dropped about half of its total tonnage by radar through clouds, the accuracy of which was equivalent to British night bombing. See Shandroff (1972); Davis (1995); and Downes (2012).

  • 23. The exception in intensity was the bombing of North Korea during the Korean War, which may have killed hundreds of thousands of people, but the most deadly part of the bombing for civilians was not a punishment campaign. Rather, it followed a denial logic. See following.

  • 24. Pape coded each of the bombing campaigns of 1972—Freedom Train and Linebacker I and II—as denial campaigns.

  • 25. Their findings for eliminationist victimization are discussed later.

  • 26. Cleansing is thus selective at the group level but indiscriminate at the individual level (Steele, 2017).

  • 27. A slightly different form of eliminationist victimization takes place when war breaks out among intermingled populations, such that members of the opposing sides are “trapped” behind enemy lines. Concern with consolidating control over their territory and eliminating fifth columns can lead locally dominant groups to massacre or expel members of “enemy” groups. On such violence in the early stages of the Spanish Civil War, see Ruiz (2007), Thomas (1961, pp. 165, 176), and De la Cueva (1998).

  • 28. These percentages were obtained by dividing the number of Palestinians remaining (about 150,000) by low (600,000) and high (760,000) estimates of the total Palestinian population.

  • 29. On the broader question of how scholars can engage ethically with policymakers, see Barma and Goldgeier (2022) and Epstein and Kaplan (n.d.).

  • 30. Prominent examples include the evolution of civilian victimization during the Boer War (Downes, 2007), Malayan Emergency (Hack, 2021), or Guatemalan Civil War (Schirmer, 1998). Belligerents have also proved adept at learning from the experiences of others: The British and Americans soon replicated the “Weyler method”—named after Spanish general Valeriano Weyler, who pioneered “reconcentration” in the Cuban War of Independence (1895–1898)—in South Africa and the Philippines, respectively, herding hundreds of thousands of civilians into squalid camps, where large numbers died.