When do ties bind? Foreign fighters,social embeddedness,and violence against civilians

Abstract

How do foreign fighters affect civilian victimization in the civil wars they join? Scholars of civil war have gone to great lengths to explain why states and insurgent groups victimize civilians, but they have not explicitly examined the impact of foreign combatants. Furthermore, while contemporary conventional wisdom attaches an overwhelmingly negative connotation to foreign fighters, history shows that the behavior of those who travel to fight in wars far from home varies significantly, especially when it comes to interacting with local populations. To address this variation, I demonstrate how differences in the embeddedness of foreign fighter populations combine with incentives that foreign fighters face to remain in the conflict zone over the long term to shape tendencies towards civilian victimization. My findings from an analysis of insurgent groups from 1990 to 2011 suggest that, overall, foreign fighters lead to escalations in violence against civilians. When comparing across groups that recruit foreign fighters, however, levels of violence differ depending on foreign fighter populations’ coethnicity to the rebel groups they join, and the distances they travel to reach a conflict zone. Specifically, the presence of coethnic foreign fighters leads to fewer escalations in violence, relative to the recruitment of non-coethnic individuals from non-neighboring states. The study provides empirical support to the claim that degrees of embeddedness across foreign fighter populations are important indicators of when and where their presence is likely to pose significant dangers to local populations.

Keywords

civil war civilian victimization escalation in violence foreign fighters political violence

How do foreign fighters affect the trajectories of the conflicts they join? Despite increasing concern in the academic and policy arenas regarding the threat posed by foreigners joining civil wars around the globe (Bakke, 2014; Braithwaite & Chu, 2018; Chu & Braithwaite, 2017), we still know relatively little about how these actors shape local conflict dynamics. In 2002, the United Nations highlighted the threat, declaring, ‘Whatever the nature of the conflict, mercenaries […] tend to escalate or at least perpetuate hostilities’ (UN, 2002). But history also shows that foreign fighters do not have uniform effects on the escalation of violence across civil wars. During the Spanish Civil War, the experiences of members of the transnational fighting force that fought on the Republican side stand out in sharp contrast to behavior implied in the UN’s declaration. Returning Americans explained how local civilians came to accept the foreign fighters who made up the International Brigades: ‘[locals] are tremendously interested in these comrades from other lands and they do everything possible to make them comfortable during their stay in the village’ (William Colfax Miller, as cited in Malet, 2013: 122). In the 1990s, foreigners who joined Croatian irregulars during Croatia’s war for independence characterized their experiences as follows: ‘we’re not in it for the killing. It’s about the camaraderie […] We are here because we want to be’ (Nir, 2012: 10). Such accounts suggest that populations of foreign fighters may engage in different behaviors across conflicts.

Figure 1.

Foreign fighters in civil wars, 1990–2011 (Malet, 2016)

The phenomenon of foreigners entering and exiting domestic conflicts dates back to the early 1800s and has increased over time (Malet, 2016). Foreign fighters have participated in civil wars across all regions, including in 96 civil conflicts from 1990 to 2011 (34% of the total) (Malet, 2016). The two panels in Figure 1 summarize these trends.

Despite the wealth of knowledge that studies on violence against civilians offer, they remain oblivious to how the addition of foreign combatants might change the ways rebel groups interact with civilian populations. While studies show that reliance on certain forms of external resources leads to escalations in civilian victimization (Salehyan, 2009; Salehyan, Siroki & Wood, 2014; Stewart & Liou, 2016), the effect of foreign human resources remains unclear. I combine and extend arguments of insurgent behavior to develop an explanation of how external human resources – foreign fighters – affect insurgent violence against civilians. In particular, I focus on variation in fighters’ embeddedness in the local social fabric of a conflict to explain civilian victimization. Different degrees of embeddedness can lead ‘roving bandits’ to become ‘stationary’ and change their behavior accordingly.

I use the conventional definition of foreign fighters: ‘non-citizens of conflict states who join insurgencies during civil conflict’ (Malet, 2013: 9) and who lack kinship ties to the militant groups that they join (Hegghammer, 2011). These individuals face particular hurdles when they join civil wars. Like their local fighter counterparts, they depend on civilians to survive during war, but their foreign origins preclude some of them from easily forming trust-based relationships needed to secure critical resources, intelligence, and compliance. Local populations may also be less likely to cooperate with foreign combatants as a result of linguistic or cultural barriers. Without cooperation, certain foreign fighters may turn to civilian abuse to force support.

This logic suggests that not all foreign fighters are created equal. Different national origins, language preferences, or cultural values may accentuate the foreignness of some, but allow others to behave more like local fighters. This leads to two observable implications. First, groups that recruit foreign fighters should be more likely to abuse civilian populations, relative to groups that recruit locally. Second, foreign fighters sharing traits, such as ethnicity, with local conflict actors should be less apt to commit violence against civilians, while those without shared traits should be more likely to victimize civilians.

My analysis of foreign fighters in insurgencies from 1989 to 2011 provides support for these hypotheses. I find that the presence of foreign fighters leads to higher average levels of civilian victimization at the armed group level. Disaggregating foreign fighters, though, also shows that those with shared traits are less likely to commit violence against civilians; groups that recruit coethnic foreign combatants from neighboring states commit less violence than those that recruit non-coethnic foreigners from non-contiguous states. The study thus demonstrates the importance of foreign fighters as a group, but also the value of disaggregating groups and beginning to unpack the mechanisms through which certain types of foreign combatants are likely to impact civil war dynamics.

Foreign fighters and civilian victimization during civil war

When armed groups recruit socially disembedded foreigners into their fighter base, the organization should become more likely to wield violence against local civilians. Conversely, armed groups relying on cadres of fighters with stronger ties to the local context should be less likely to abuse civilians. The notion that members of armed groups exhibit varying incentives to engage in abuse during war is well established in the conflict literature (Weinstein, 2007; Cohen, 2013; Manekin, 2013; Green, 2016). Arguments in this literature, however, cannot easily take account of foreign fighters. For example, in Weinstein’s (2007) theoretical framework of opportunist versus activist rebellions, foreigners could count as investors or consumers, depending on whether they join a civil war for ideological or pecuniary reasons.¹

What existing studies miss is that the rebel rank and file is unlikely to be monolithic, particularly in terms of the social resources with which individual rebels enter a conflict. In other words, individual rebels join with different social ties, cultural knowledge, linguistic abilities, religious or ideological leanings, or other values. While one rebel’s individual set of resources may strongly connect her to the local conflict context – for example if she has lived her entire life in the local area – other rebels may be non-local to the conflict zone. This is important, given that most individuals access materials, including those that might be necessary for war, through social ties (Lin, 2001: 43). Variation in rebels’ social resource bases is thus likely to affect the types of behavior that they adopt to accumulate the support and compliance they need during conflict.

Modes of fighter–civilian interaction

While some rebel groups coerce civilians to gain access to materials they need, many others encourage civilian contributions by promising security, dispute resolution, and other governance services (Mampilly, 2011). Rebel interactions with civilians span the continuum between coercion and mutual exchange, or threats and promises.

Studies hint at the importance of embeddedness in the local social fabric of a conflict when it comes to developing mutually beneficial systems of exchange. In order to gain voluntary cooperation from civilians, rebels secure loyalty by taking into consideration local demands (Mampilly, 2011), and civilians are more likely to view rebels as legitimate when they embody locally grounded values and norms (Kasfir, 2015; Förster, 2015). Even when rebels seek to make sweeping changes to a sociopolitical order, they often adopt local values and narratives that evoke beliefs shared with civilians. This helps to strengthen their claims to authority (Hoffmann, 2015). Social embeddedness thus eases access to critical information and blurs the distinction between rebel and civilian. When civilians see insurgent groups as extensions of their own communities, rebels can rely at least partially on promises and compromise to seek compliance and popular support.

The alternative is to seek cooperation through coercion, a strategy employed by states and insurgent groups alike (Mason & Krane, 1989; Valentino, 2004; Downes, 2006; Kalyvas, 2006; Weinstein, 2007). While violence yields more immediate returns and is typically cheaper than investments into systems of exchange, it is likely to backfire, especially when rebels cannot employ selective violence because they lack deep local knowledge (Kalyvas, 2006).

Theories of rebel violence highlight how civilian constituencies determine rebels’ organizational structures, and that violence from rank and file combatants is ultimately a by-product of group recruiting practices (Humphreys & Weinstein, 2006; Weinstein, 2007; Cohen, 2013, 2016). In particular, Weinstein (2007) shows how rebellions that attract high-commitment individuals, or investors, obtain resources from non-combatants by striking bargains based on their knowledge of established norms and social networks. Opportunistic rebellions, on the other hand, attract consumer rebels as a result of their reliance on economic endowments that decrease their need to rely on local populations and thus on local knowledge. While activist rebellions can use violence strategically thanks to insurgents’ social ties to the local conflict context, opportunistic rebellions lack the local ties necessary to control violence. Once conflict is underway, ties that bind armed groups to local populations during war – in particular shared ethnicity – continue to influence patterns of victimization (Ottman, 2017). When militants share a common ethnicity with local populations, they are less likely to victimize civilians as doing so would be tantamount to attacking one’s own community (Fjelde & Hultman, 2014; Stanton, 2015).

These findings are consistent with the idea that rebels’ embeddedness into the social fabric of a conflict shapes propensities towards or away from abusive behavior. Though the social capital with which rebels join a conflict is an important precursor to many of the dynamics that inform economistic logics of rebel behavior, its role remains relatively under-theorized. For rebels to exhibit the qualities of so-called ‘stationary’ bandits (Olson, 1993), they must begin with or develop deep access into local communities. Familiarity with local norms and values that inform civilians’ demands helps them strike bargains with local populations and bring violence under control (Weinstein, 2007; Mampilly, 2011; Kasfir, 2015). The absence of social ties, on the other hand, drives the predatory behavior of ‘roving’ bandits who rely overwhelmingly on social capital or opportunities that may be extra-local to a conflict context. To reap the benefits offered by access to natural resources or openings into criminal networks, for instance, rebels need not familiarize themselves with local norms and customs, and ties to local populations may be less important than ties to foreign actors. Thus, it is not necessarily the type of endowment that drives rebel behavior towards civilians, as Weinstein (2007) argues, but rather the extent to which original endowments are representative of an armed group’s embeddedness into the social fabric of a conflict.

Given this body of research, I argue that rebels’ social ties to a local conflict context are insufficiently explored as a determinant of rebel behavior. This is surprising, given that many theories of rebel violence and governance highlight the important role that local knowledge and ultimately social embeddedness play in shaping insurgents’ ability to seek support and compliance from civilians. Against this background, I reconsider the role of social embeddedness in informing rebel behavior and apply it to understanding the effects of foreign fighters on civilian victimization.

Recruits, local ties, and violence against civilians

In general, foreign fighters should be less embedded in local population networks. They are likely to experience deficits in social capital relative to local fighters because they lack social ties to local conflict actors. Additionally, foreign fighters may not speak the local language, and be oblivious to the norms and values of local populations. But some foreign fighters may be more foreign than others, and different qualities of foreignness likely affect trajectories of violence differently. For instance, foreigners who enter conflicts fluent in the local language or who share local customs have more resources with which to embed. They can communicate directly with local actors, and have a better understanding of the norms and values that inform the expectations of the ordinary population. Similarly, foreigners who travel from neighboring states may benefit from direct or indirect ties to members of local populations. Pre-existing social connections to the local conflict context, or the ability to embed via common language and shared understandings, should help some foreign rebels establish the social capital necessary to gather critical materials. These qualities may also enhance the extent to which foreign fighters see themselves as part of the civilian population, rather than as outsiders with a reduced stake in the conflict’s ultimate outcome. More socially embedded groups should thus seek to avoid potentially destabilizing levels of violence, relative to disembedded groups. Below, I expand on these logics to develop hypotheses linking foreign fighters to civilian victimization.

The individuals who make up an armed group’s rank and file – its recruits – are critical military resources (Gates, 2002). Recruits are both the mechanism for pulling local knowledge and resources into an organization and the face of the armed group in particular communities. Recruits who are deeply embedded into the local social fabric of a conflict due to multiple, diverse ties to local populations may be a particularly valuable asset to armed groups, insofar as they are well positioned to benefit from local knowledge (Kalyvas, 2006; Lyall, 2010; Parkinson, 2013). Local ties of everyday kinship, marriage, friendship, and community-based relationships are central to organizational re-emergence and survival during times of crisis (Parkinson, 2013), and also provide information, trust, and shared political meanings that serve as foundations of new armed groups (Staniland, 2012, 2014). Stanton (2015) also finds that militias that recruit from the same ethnic or religious group as rebels are less likely to use violence against civilians, because their members likely benefit from enhanced sources of information. Domestic forces are frequently better suited for combat because of their intimate knowledge of the population and terrain (Salehyan, 2010), and common language and familiarity with local customs make it easier for sides involved in a civil war to identify loyalties (Schutte, 2015).

Given the importance of prior relationships and shared traits to rebel behavior, the presence of foreign combatants should add a layer of complexity to rebel–civilian dynamics. For instance, foreign combatants may upset the balance of ethnic overlap between rebel groups and their civilian constituency, leading a group expected to carry out low levels of violence to turn to high levels of victimization (Fjelde & Hultman, 2014). Consider the reaction of an Ivorian civilian when asked about the recruitment of Liberian fighters by the Ivorian Popular Movement of the Great West (MPIGO) rebels: ‘In the beginning when the rebels arrived everyone was happy […] Later, we were not happy. We realized the Liberians were not nice’ (HRW, 2003: 30). Alternatively, divisions within rebels’ civilian constituencies might foster environments ripe for opportunistic violence when foreigners are present. Feuding civilians might manipulate foreign fighters into settling old scores, taking advantage of the latter’s detachment from the pre-conflict context to convince them to partake in acts of revenge.

Groups that employ socially disembedded outsiders may face unique hurdles relative to those that rely on cadres of fighters who are networked into local communities. It seems improbable, for instance, that foreign recruits benefit from the same dense networks of local quotidian ties that local rebels draw upon. Foreign members of a rank and file might lack the practical skills necessary to interact with local populations, such as language. This complicates critical tasks like gathering food, securing shelter, and collecting information and intelligence. Moreover, civilians may less readily comply with foreign combatants who lack a shared understanding of local customs. The difficulties that cognitive dissonance poses to warring factions are an established reality of war: ‘The absence of cognitive access to local customs can provide incentives for brutal and indiscriminate behavior’ (Schutte, 2015: 1109).

Several contemporary armed groups provide examples of some of these dynamics. The Abkhaz conflict in Georgia in the early 1990s featured the participation of a number of outsiders, individuals who did not reside in Georgia prior to war breaking out. Many looted the homes of fleeing Georgians, and Georgian prisoners of war reported that foreign fighters committed some of the worst pillage and atrocities of the war (HRW, 1995). In Chechnya, the arrival of foreign insurgents created divisions within the original nationalist movement, pitting indigenous elements against transnational Islamists and introducing Chechen-on-Chechen violence. The arrival of foreign combatants led to more violence aimed specifically at civilians, and the introduction of suicide bombing tactics that many local Chechens actively resisted (Bakke, 2014). In sum, this logic leads to the hypothesis that the recruitment of foreign fighters into civil wars is likely to induce heightened violence against civilians:

Hypothesis 1: Armed groups that recruit foreign fighters will cause more civilian casualties than groups with a predominantly local membership base.

However, not all foreign fighters exhibit equal levels of foreignness. Some may share language, ethnicity, religious commitment, ideology, or customs that ease their interactions. Any of these factors might shape the extent of their foreignness relative to locals. For example, members of the Armenian diaspora who joined Nagorno-Karabakh rebels in their struggle against Azerbaijan were likely perceived as less foreign than British Muslims who fought with Hamas in Israel.

A particularly important feature is language, a facet of ethnicity (Horowitz, 1985).² When a rebel group’s rank and file includes individuals who are unable to communicate directly with local civilians, it will likely face higher hurdles when it comes to establishing systems of mutual exchange with local populations. For civilians, the addition of linguistically distinct fighters introduces another layer of uncertainty to an already unstable environment, and throws into question the legitimacy of the rebel group. Such impasses may aggravate levels of cognitive dissonance between the armed group and civilians, reducing the extent to which the group can employ selective violence against locals or engage in service provision. Language often presents itself as ‘an inescapable medium of public discourse’, (Brubaker, 2013: 5–6) and even minor linguistic distinctions are easy to notice and loaded with symbolic connotations (Gellner, 1983: 58–62).

An observable implication of this logic is that linguistic – or ethnic – differences between foreign fighters and local conflict actors influence armed group behavior in particular ways. The linguistic barriers that some foreign fighter populations face during war reduce the ability of rebels to embed, and to develop ties that enable them to accumulate popular support via systems of mutual exchange. When foreign fighters share a common ethnicity with local rebels, foreigners’ abilities to embed into local population networks can work through local fighters. Shared language between foreign and domestic rebels can help the latter communicate the embeddedness of the former to local civilians, alleviating some of the cognitive dissonance issues I highlight above. In these situations, local rebels have the capacity to act as intermediaries for their foreign counterparts, and close gaps between populations of foreign fighters and local civilians.

I also expect that the distance foreign combatants travel to reach a conflict zone impacts the nature of their interactions with civilians. First, foreign origins may grant fighters from other countries anonymity within the conflict zone, which should lower barriers to violence. Consider, for instance, this statement from a foreign member of the Liberians United for Reconciliation and Democracy (LURD): ‘I even killed some who resisted […] I felt like I could do some of those things in Liberia because no one knew me there – they weren’t my people’ (HRW, 2005: 34). Given that violence applied over longer distances tends to become more intense (Schutte, 2017), I would expect propensities for violence to increase in function with the physical distance a foreign fighter has traveled to reach a conflict zone.

Second, immunity from revenge may lower barriers to violence for some foreign fighters. Kalyvas (1999) demonstrates the salience of revenge as a driver for killing in the Algerian context. Foreign fighters who come from distant geographical locations may have less to fear when it comes to retaliatory attacks, particularly those that target their kin. The farther removed a rebel is from her home state, the more secure she feels that acts of violence committed on the battlefield will not lead to reprisals against her family.

Finally, accounts of foreign fighters traveling long distances to reach civil conflicts stress their lack of local, contextual knowledge. In Côte d’Ivoire, most of the foreign combatants who arrived in 2002–03 did not know the name of the Ivoirian rebel groups actively fighting the government, and only a few were familiar with the name Laurent Gbagbo, the Ivoirian president (HRW, 2005). This unfamiliarity and outright disregard for local conflict dynamics may be a sign of disinterest in the ultimate outcome of a struggle, sentiments that might obstruct groups’ investments into systems of rebel governance. This is likely particularly true if foreigners hail from non-neighboring countries and areas beyond reach of a conflict’s potential contagious effects.

This logic speaks to the value of looking at two indicators in particular: foreign fighter ethnicities and the distance that they travel to reach a conflict. This does not mean that other factors related to foreign fighters are unimportant. For instance, the proportion of foreign to local rebels likely conditions some of the dynamics outlined above, and other measures of foreignness, such as ideology or rivalry or active conflict between the host and foreign states might also affect patterns of civilian victimization. Given the limits of current data, however, I focus on the two aforementioned indicators as an initial investigation of my mechanisms. I add two hypotheses:

Hypothesis 2: When armed groups recruit foreign fighters, violence against civilians increases when foreign fighters do not share a common ethnicity with the rebel group.

Hypothesis 3: When armed groups recruit foreign fighters, violence against civilians increases when foreign fighters are from non-contiguous states.

Research design and data

In order to assess whether foreign fighters affect the propensity of armed groups to abuse civilians, I design a series of tests on a dataset with the conflict dyad as the unit of analysis. I use the Non-State Actors (NSA) dataset (Cunningham, Gleditsch & Salehyan, 2013) as a base, resulting in 997 dyad-year observations spanning 1989–2011.³

My dependent variable is the annual number of civilian fatalities caused by insurgent groups from the Uppsala Conflict Data Program (UCDP) one-sided fatalities dataset (Eck & Hultman, 2007; Allansson, Melander & Themnér, 2017). Because the structure of the dependent variable is a non-normally distributed cross-section of annual counts of insurgent-inflicted civilian deaths, I use a negative binomial model in my main analysis and include a number of additional tests to assess the robustness of my primary findings. I cluster the standard errors on the dyad in all models to account for within-conflict dependencies.

Explanatory variables

Following recent studies on the effect of foreign fighters on conflict outcomes (Chu & Braithwaite, 2017), my main independent variables come from existing data on foreign fighters in civil conflicts collected by Malet (2016).⁴ My primary independent variable accounts for foreign fighters joining a non-state group during civil war. Foreign fighters is a binary variable coded 1 if foreign recruits were present in the ranks of an insurgency, and 0 otherwise. To assess the impact of degrees of foreignness, I use two variables. Coethnic is a binary variable coded 1 if the foreign recruits who joined the armed group shared the same ethnicity as the majority of the rebel group. Beyond neighboring is a dichotomous variable coded 1 if the foreign fighters traveled from a non-neighboring country to fight in a civil war.⁵

There are notable limitations to the foreign fighter data. First, information is missing on foreign recruits’ overall numbers relative to local fighters. While some groups comprise predominantly foreigners, foreign recruits are only a minority in others.⁶ Despite this variation, I expect that the cases in Malet’s (2016) dataset reflect a certain threshold in terms of size that allows for reasonable comparison.

Second, the data do not account for the timing of foreign combatants’ arrival into conflicts. This information is important, since foreigners may have arrived after escalations in civilian victimization. However, most anecdotal evidence suggests that foreign combatants joined prior to groups turning to civilian victimization, and were often influential in the formation of rebel groups (HRW, 1997; HRW, 2005; Bauters, 2012). I therefore do not expect the exclusion of timing information to severely bias the results of the current analysis.

Finally, there is little way of precisely knowing the extent to which the ethnicity of foreign recruits overlaps with that of civilian populations across conflicts. While the Coethnic variable accounts for ethnic similarities between foreign fighters and members of the rebel group, it is agnostic as to whether ethnic overlap also existed with civilian populations. However, the fact that nearly 75% of rebel groups recruit from ethnic groups suggests that overlap between foreign and local fighters’ ethnicities likely corresponds to overlap between foreign fighters’ ethnicities and the ethnicities of populations from which the insurgency recruits (Wucherpfennig et al., 2012). This indicates that the Coethnic variable is likely a valid construct of the mechanisms elaborated above.

Controls

I include a set of control variables across each model to account for alternative explanations of civilian victimization, and for factors that might increase the propensity of foreign fighters to participate in an insurgency.

First, I include a set of insurgency-level controls. I assume that civilian casualties may be a function of rebel capabilities relative to their opponent (Asal & Rethemeyer, 2008; Wood, 2014). The variable Rebel strength is a composite measure of rebel strength relative to the government ranging from 1 to 5 and comes from the NSA data. The presence of lootable resources might also increase the propensity of groups to abuse civilians and attract opportunistic foreigners to a conflict (Weinstein, 2007; Metelits, 2009). Lootable resources is a binary indicator for the presence of gems or drugs within the conflict area, and comes from Lujala (2009), Gilmore and colleagues (2005), and Buhaug & Lujala (2005). Studies also suggest that insurgent aims and ideology affect patterns of civilian victimization (Kaufmann, 1996; Toft, 2003; Gutiérrez-Sanín & Wood, 2014). Furthermore, groups active in recent Islamist conflicts have pursued transnational agendas (Walter, 2017) and experienced high participation from foreign fighters. I include a binary indicator that accounts for whether an insurgent group was involved in a Territorial conflict from the UCDP Dyadic Dataset (Harbom, Melander & Wallensteen, 2008), and dummy variables for Islamist conflicts and whether a group adhered to a Leftist ideology. The former is from the NSA data, while the latter comes from Wood & Thomas (2017). I also include a control for conflict Duration, which is the count of years since the start of the conflict.

Next, I include a set of country-level controls and variables that account for the strategic interaction between sides during civil war. A state’s Population density accounts for the possibility that larger population concentrations provide increased opportunities for victimization. I also include Income growth (annual change in GDP per capita) to account for the effect of economic growth. These variables are from the World Bank Development Indicators (WDI) (World Bank, 2017). Studies also find a positive relationship between conflict intensity and civilian victimization (Hultman, 2012; Wood, 2010). The variable Intensity is coded 1 for years in which the conflict reached or exceeded 1,000 battle-deaths (0 otherwise), and comes from the UCDP Dyadic Dataset (Harbom, Melander & Wallensteen, 2008). I also add a variable for Government violence to account for government-perpetrated civilian victimization, since insurgent violence against civilians may be a response to government targeting practices (Costalli & Moro, 2012; Schneider, Bussman & Ruhe, 2012). It is also possible that the presence of foreign fighters drives government violence against civilians, as punishment for supporting externally resourced rebel groups. This variable comes

Figure 2.

Distribution of foreign fighter types across insurgent groups, 1989–2011

from the UCDP one-sided violence dataset. To address potential issues related to temporal dependence, I include a one-year lag of the dependent variable.

Analysis and results

Of the 241 unique insurgent groups in the NSA data from 1989 to 2011, 77 (about 32%) included foreign fighters in their ranks. Figure 2 provides a breakdown of foreign fighters in insurgencies by type. Foreign fighter populations fall into one of four categories: (1) non-coethnic fighters from non-contiguous states; (2) coethnic fighters from neighboring states; (3) coethnic fighters from non-neighboring states; and (4) non-coethnic fighters from neighboring states.

An independent-samples t-test with civilian targeting grouped into the episode format of the original NSA data shows a significant difference in the number of civilians killed by groups with no foreign fighters (mean = 118) and those with foreign fighters (mean = 741; p = 0.02). Among groups with foreign fighters, there is a difference in the number of civilians killed by groups with Type 1 foreign fighters (mean = 812) and Type 2 foreign fighters (mean = 328), which represent the most and least disembedded foreign fighter populations, respectively.⁷ This provides strong initial support for Hypotheses 1–3.

Results

To test my hypotheses linking foreign fighters to one-sided violence, I first estimate a series of four negative binomial models. Table I presents the results. Models 1a–4a test the effect of foreign fighters on the full set of observations, and Models 1b–4b test the effect of foreign fighter characteristics on the subset of dyad-years that include foreign fighters. Across all specifications, the coefficient for Foreign fighters is large, positive, and statistically significant, consistent with my first hypothesis. In all but one of the models on the subset of observations that include foreign fighters, the coefficient for Coethnic foreign fighters is large, negative, and statistically significant, consistent with Hypothesis 2.

Results from Model 1a show that groups with foreign fighters have an incident rate of killing civilians that is 5.5 times higher than that of groups with no foreign fighters. In Model 1b, groups that recruit coethnic foreign fighters are less likely to abuse civilians, relative to groups that recruit non-coethnic foreigners. However, the coefficient for foreign fighters from non-contiguous states is also negatively related to civilian victimization and statistically significant, running counter to the expectations of Hypothesis 3.

When insurgency-level controls are incorporated (Model 2a), the Foreign fighters variable retains its sign and strong statistical significance. Among insurgencies that included foreign fighters (Model 2b), however, foreigners’ characteristics do not predict victimization. Instead, rebel aims predict abuse: Territorial conflicts are associated with significantly fewer civilian deaths.⁸ The measure for Foreign fighters in Model 3a, which includes state-level controls and variables accounting for dynamic interaction between opposing sides, again retains its statistical significance and sign. Model 3b also shows that groups that use Coethnic foreign fighters are significantly less likely to engage in civilian victimization, relative to those that recruit non-coethnic foreigners, but the coefficient for Beyond neighboring foreign fighters is not statistically significant.

Finally, Models 4a and 4b, which include the full set of control variables, provide support for Hypotheses 1 and 2. Specifically, the inclusion of foreign fighters continues to predict civilian victimization. Among rebel organizations that use foreign fighters, the variable for Coethnic foreign fighters is significant at the p < .05 level, and negatively correlated to violence. However, the

Table I.

Negative binomial regression results

	Model 1a	Model 1b	Model 2a	Model 2b	Model 3a	Model 3b	Model 4a	Model 4b
Foreign fighters (FF)	1.719**		1.513**		0.997**		0.667*
	(0.585)		(0.446)		(0.294)		(0.335)
Beyond neighboring FF		–1.600*		0.555		0.111		0.887
		(0.827)		(1.164)		(0.644)		(1.112)
Coethnic FF		–1.665*		–0.700		–1.796**		–1.215*
		(0.794)		(0.668)		(0.613)		(0.520)
Rebel strength			1.027**	0.598			0.830**	0.311
			(0.377)	(0.556)			(0.293)	(0.330)
Lootable resources			0.729*	0.989			0.598*	0.150
			(0.310)	(0.799)			(0.300)	(0.761)
Territorial conflict			–1.230**	–2.889**			–0.872*	–2.311**
			(0.494)	(1.077)			(0.404)	(0.527)
Islamist conflict			0.321	–0.546			–1.529^†	1.672
			(0.781)	(1.300)			(0.988)	(0.879)
Ideology – Leftist			–0.501				–0.794
			(0.480)				(0.507)
Conflict duration			0.070*	0.044			0.068*	0.043
			(0.030)	(0.061)			(0.032)	(0.038)
Population density (ln)					0.233*	0.141	0.374**	0.328*
					(0.100)	(0.148)	(0.148)	(0.154)
Income growth					–0.022^†	–0.010	0.015	–0.003
					(0.012)	(0.012)	(0.015)	(0.019)
Gov’t violence (ln)					0.132*	0.319**	0.080	0.184
					(0.052)	(0.066)	(0.065)	(0.116)
Rebel violence (t–1)					0.413**	0.428**	0.340**	0.366**
					(0.042)	(0.089)	(0.042)	(0.097)
Conflict intensity					2.105**	2.408**	1.968**	2.255**
					(0.406)	(0.532)	(0.422)	(0.515)
Constant	3.553**	6.233**	1.342	3.900**	0.758	1.805**	–1.320	0.542
	(0.270)	(0.788)	(0.938)	(1.423)	(0.506)	(0.720)	(0.971)	(1.028)
Base category for FF variables	No FF	Neighboring FF; Non-coethnic FF	No FF	Neighboring FF; Non-coethnic FF	No FF	Neighboring FF; Non-coethnic FF	No FF	Neighboring FF; Non-coethnic FF
Observations	997	361	806	277	806	277	806	277

† p ≤ 0.10, * p ≤ 0.05, ** p ≤ 0.01. Standard errors clustered on dyadid in parentheses. Models 1b–4b are on observations that include foreign fighters.

Figure 3.

Panels (a) and (b) show coefficient plots for the variables Foreign fighters and Coethnic in Table I, respectively

physical distance that foreign combatants travel to reach a conflict zone still does not influence rebels’ abuse of civilians. There is therefore no support for Hypothesis 3. Figure 3 presents the Foreign fighters and Coethnic FF coefficient estimates from Table I.

Substantively, Model 4a shows that groups with foreign fighters have an incident rate of killing civilians that is twice as high as that of groups with no foreign combatants. Comparing across groups that recruit foreign fighters, Model 4b shows that groups that recruit coethnic foreign fighters have an incident rate of killing civilians that is 0.3 times that of groups that recruit non-coethnic foreign fighters. Figure 4 plots the predicted numbers of civilian deaths from Models 4a and 4b. With all control variables set to their means, groups with foreign fighters are over twice as likely to victimize civilians, relative to groups with no foreign combatants. Among groups that recruit foreign fighters, those that recruit coethnic combatants kill 3.5 times fewer civilians than groups that recruit non-coethnic individuals. The combined results confirm that the predictive accuracy of the models and the Foreign fighters and Coethnic variables are both very high.

Alternative specifications

I perform a number of additional tests to assess the robustness of my primary findings. On the whole, these checks continue to provide strong support for Hypotheses 1 and 2. Full regression results for all robustness tests appear in the Online appendix.

To address the possibility that structurally different processes are responsible for a number of the zeroes present in the dependent variable, I use a zero-inflated negative binomial (ZINB) regression to reproduce the main models’ results. The results of Model 5a (see Online appendix) are largely the same as those from Model 4a: the Foreign fighters variable is positively related to violence and statistically significant at the p < .001 level. Statistically significant coefficients in the logit inflation reveal that territorial conflicts, those that take place in sparsely populated areas, groups’ lack of access to lootable resources, low conflict intensities, and previous rebel violence predict the absence of civilian victimization. The robustness test therefore confirms the original findings that groups with foreign fighters are more likely to abuse civilians, relative to groups with predominantly local fighter bases.

The ZINB replication of Model 4b also lends further support to the expectation that, among groups that recruit foreign fighters, those that recruit coethnic foreign combatants are less likely to abuse civilians relative to groups that recruit non-coethnic foreigners. In the negative binomial component of Model 5b, the coefficient for Coethnic

Figure 4.

Predicted number of civilian deaths based on the recruitment of foreign fighters (Panel a) and coethnic foreign fighters (Panel b)

foreign fighters is statistically significant at the p < .001 level and negatively correlated to civilian victimization. However, the results also show that groups that recruit foreign fighters from non-contiguous states are also significantly less likely to abuse civilians, relative to groups with other types of foreign fighters. Other significant predictors of violence include large population densities, government violence, and high conflict intensities. Territorial conflicts are negatively associated with civilian victimization, and results from the logit component of the model are the same as in Model 5a.

To further probe the result regarding the behavior of foreign fighters from non-neighboring states, I employ an alternative specification of the foreign fighter variables that better reflects the four categories depicted in Figure 2. As currently constructed, the Coethnic and Beyond neighboring variables do not differentiate between groups that use coethnic foreigners from non-contiguous states and those that use coethnics from contiguous states. The main models may thus only partially capture the differential impacts of various types of foreign fighter populations.

I construct a categorical variable that reflects the distribution of foreign fighter populations articulated above. In Models 6 and 7 (Tables A3 and A4 of Online appendix), I test the effect of Foreign fighter type on rebel violence using a negative binomial and ZINB regression, respectively. Both tests are performed on the subset of observations that include foreign fighters. In Model 6, I find that groups that recruit coethnic foreign fighters from neighboring states (Type 2 foreign fighters) are significantly less likely (p = .001) to engage in civilian victimization relative to groups that recruit non-coethnic foreigners from non-contiguous states (Type 1 foreign fighters).⁹ These results hold in the ZINB model (Model 7): rebels that recruit coethnic foreign fighters from neighboring states have an incident rate of violence that is 0.5 times that of groups that recruit non-coethnic foreigners from non-neighboring states (p = 0.04). Figure 5 presents the predicted effects of Foreign fighter type on civilian victimization. In Model 6, foreign fighter populations embedded into local conflict contexts via both common ethnicity and proximate geographic origins, or just the latter (Type 2 and Type 4 foreign fighters), are significantly less likely to kill civilians, relative to disembedded populations that fall into the Type 1 category.¹⁰

Figure 5.

Predicted number of civilian deaths by foreign fighter type, based on Models 6 and 7

Finally, to account for temporal variation concerns that arise from the dyad-year unit of analysis, I include checks where I group civilian victimization by cross-section with NSA dyads as the unit of analysis.¹¹ I use the total number of civilians killed by each group as the new dependent variable. Time-variant state-level covariates retain their value during the year the conflict began, and a dichotomous Government violence variable accounts for whether state forces engaged in civilian victimization during the conflict. The results of negative binomial Models 8a–8d confirm the findings of the time-variant analysis: groups that recruit foreign fighters have an incident rate of killing civilians that is 2.5 times that of groups that do not use foreign combatants (p = .03 in Model 8d).¹²

Discussion

The findings from the statistical analyses provide strong evidence in support of the hypothesis that rebels that recruit foreign combatants are more likely to abuse civilians, compared to groups that rely predominantly on a local fighter base. The presence of foreign fighters is consistently positively correlated to rebel violence against civilians, with models showing that foreign fighter-heavy rebels kill as many as two to three times more civilians than rebels that do not use foreigners. This finding is robust across time-variant and cross-sectional analyses of civilian victimization.

Furthermore, the time-variant models provide consistent support to Hypothesis 2, that groups with coethnic foreign fighters are less likely to abuse civilians, relative to groups with non-coethnic individuals. According to Model 4b, rebels that recruit Coethnic foreign fighters kill an average of 15 civilians annually. In comparison, rebels that recruit non-coethnic foreigners cause nearly 3.5 times as many civilian deaths annually. Models 6 and 7 also show that groups that recruit foreigners from neighboring states who share a common ethnicity with local conflict actors are significantly less likely to abuse civilians, relative to groups that use non-coethnic foreigners from non-contiguous states. While the latter kill between 49 and 56 civilians annually, doubly embedded foreign fighter populations (i.e. coethnic foreigners from neighboring countries) lead groups to kill between 6 and 17 civilians annually, depending on model specifications. These differences are significant at the p < 0.05 level. The combination of coethnicity and proximate geographic origins thus best predicts which foreign fighter populations are least likely to influence rebel groups towards civilian victimization. This adds nuance to the original argument that links rebel social embeddedness to civilian victimization, and suggests that different forms of embeddedness are mutually reinforcing when it comes to dampening propensities towards violent behavior.

The empirical analyses also provide substantial support for the role of other conflict- and state-level factors in influencing rebel strategies towards civilians. Large population densities provide opportunities for rebels to kill civilians, consistent with findings from the literature. At the conflict level, territorial conflicts are significantly less likely to experience civilian victimization. High numbers of battle-deaths and previous levels of rebel violence are also significant predictors of rebel violence. There is, however, less evidence that group-level factors consistently explain why rebels abuse local populations. While the presence of lootable resources and measures of rebel strength are significant and positively related to violence in some models, the finding is not consistent across specifications.

Most importantly, the measure indicating the presence of foreign fighters within rebel groups remains significantly correlated to high levels of civilian victimization even when controlling for other significant state- and conflict-level predictors. This increases confidence that the observed relationships are in fact the result of the addition of disembedded outsiders into rebel groups’ ranks, rather than the product of state- or conflict-wide patterns of violence. The time-variant models on subsets of observations that include foreign fighters help to identify possible mechanisms through which foreign fighters influence rebel propensities for violence. Specifically, the recruitment of strongly disembedded fighters – those that come from non-neighboring countries and do not share common traits with local conflict actors – leads groups to kill more civilians than when they recruit doubly embedded foreign fighters. Foreign fighter populations that are coethnic to the rebel groups they join and also come from neighboring states are associated with significantly less violence than their disembedded counterparts.

The study constitutes an important launching point for the further study of foreign fighters specifically, and the effect of ethnicity on civilian victimization more broadly. Notably, the non-findings in the cross-sectional variations of the analysis highlight the need for more disaggregated data on foreign fighter populations that better measure the extent to which fighters might be structurally or culturally embedded into local conflict contexts. Additionally, because the identity of civilian victims is unknown in current one-sided violence data, it is difficult to tell whether and to what extent groups actually victimize non-coethnics. The findings therefore validate the importance of ongoing efforts in both areas (see e.g. Moore, 2018; Fjelde et al., 2016).

Conclusion

Foreign fighters have featured prominently in rhetoric and public discussion as representative of the most ‘fanatical and dangerous type of enemy’ (Li, 2010: 363). While evidence from recent or ongoing conflicts in places such as Côte d’Ivoire or Syria confirms the image of foreign fighters as figures of threat, global data on their presence in civil war suggest that systematic linkages between foreign militants and hyper-violence may rely on flawed understandings of this specific actor category. In this article, I test hypotheses about foreign fighter abuse of civilians generally, as well as about the tendency for specific types of foreign fighters to abuse civilians during war. I find that when groups draw on foreign fighters, conflicts are more likely to see high levels of civilian victimization. However, I also find that groups that recruit specific types of foreign fighters – coethnic combatants from neighboring states – are less likely to abuse civilians, compared to groups that recruit non-coethnic combatants from non-contiguous states. Degrees of embeddedness across foreign fighter populations are therefore important indicators of when and where their presence is likely to pose significant danger to local populations.

Admittedly, the specific mechanisms through which foreign fighters influence insurgencies towards or away from violence cannot be confirmed through statistical methods alone. In particular, the study does not answer the following questions: how do different markers of foreignness impact foreign fighters’ embeddedness into local conflict contexts, and how might groups’ organizational structures condition the influence that foreign combatants have on insurgencies’ interactions with civilians? Are groups that recruit non-coethnic foreign fighters more likely to engage in ethnic targeting of civilians, relative to those that recruit coethnics? Future qualitative research might shed greater light on specific processes and pathways.

The study has important implications for future research and policy. In particular, the social and other resources with which individual rebels enter a conflict likely have an important influence on group behavior during conflict. Though I adopt a strictly state-centric conceptualization of foreignness, being judged a foreigner is often subjective and frequently extends beyond simplistic definitions of citizenship. Indeed, the struggle over who is a local and who is a stranger is often itself the driver of conflict (Jackson, 2006). Rebels’ social embeddedness may also change according to territorial gains and losses, or as areas of operations shift. As such, the study urges further evaluation of dynamics linking social embeddedness to violence when armed groups include ‘local’ cadres of fighters, especially within conflicts across time and space. Finally, investigating the specific conflict- and group-level conditions that make the participation of ultra-violent foreign fighters more likely will be an important component of the growing effort to better understand the effect of foreign fighters on conflict outcomes.

For policymakers, patterns of rebel recruitment that extend beyond regional boundaries should be a warning sign of high levels of violence against civilians. My findings stress the importance of limiting the movement of outsiders into active conflict zones, and of the loss of local legitimacy that groups are likely to incur when they recruit socially disembedded foreign fighters. Given the findings of recent studies linking rebel governance to postwar democratization (Huang, 2016: 10), the presence of transient foreign fighters uncommitted to developing systems of mutual exchange with local populations should also be particularly concerning. Not only might limiting the flow of foreign fighters into intrastate conflicts alleviate propensities for civilian abuse, it may also have important long-term impacts, particularly as this concerns post-conflict stability and governance.

Supplemental Material

Supplemental Material, JPR804594_appendix - When do ties bind? Foreign fighters, social embeddedness, and violence against civilians

Supplemental Material, JPR804594_appendix for When do ties bind? Foreign fighters, social embeddedness, and violence against civilians by Pauline Moore in Journal of Peace Research

Footnotes

Replication data

The Online appendix, dataset, and do-files for the empirical analysis in this article can be found at .

Acknowledgements

The author thanks Deborah Avant, Erica Chenoweth, Anna Pechenkina, Jonathan Pinckney, Jeremy Speight, three anonymous reviewers, and the editorial team for helpful comments and suggestions.

Notes

References

Allansson

Marie

Melander

Erik

Themnér

Lotta

(2017) Organized violence, 1989–2016. Journal of Peace Research 54(4): 574–587.

Asal

Victor

Rethemeyer

R Karl

(2008) The nature of the beast: Organizational structures and the lethality of terrorist attacks. Journal of Politics 70(2): 437–449.

Bakke

Kristin M

(2014) Help wanted? The mixed record of foreign fighters in domestic insurgencies. International Security 38(4): 150–187.

Bauters

Jasper

(2012) A taxonomy of non-state armed actors in the Central African Republic. International Peace Information Service (IPIS) (http://ipisresearch.be/publication/taxonomy-non-state-armed-actors-central-african-republic/).

Bormann

Nils-Christian

Cederman

Lars-Erik

Vogt

Manuel

(2017) Language, religion, and ethnic civil war. Journal of Conflict Resolution 61(4): 744–771.

Braithwaite

Alex

Chu

Tiffany

(2018) Civil conflicts abroad, foreign fighters, and terrorism at home. Journal of Conflict Resolution 62(8): 1636–1660.

Brubaker

Rogers

(2013) Language, religion and the politics of difference. Nations and Nationalism 19(1): 1–20.

Buhaug

Halvard

Lujala

Päivi

(2005) Accounting for scale: Measuring geography in quantitative studies of civil war. Political Geography 24(4): 399–418.

Chu

Tiffany S

Braithwaite

Alex

(2017) The impact of foreign fighters on civil conflict outcomes. Research & Politics 4(3).

10.

Cohen

Dara Kay

(2013) Explaining rape during civil war: Cross-national evidence (1980–2009). American Political Science Review 107(3): 461–477.

11.

Cohen

Dara Kay

(2016) Rape During Civil War. Ithaca, NY: Cornell University Press.

12.

Costalli

Stefano

Moro

Francesco Niccolo

(2012) Ethnicity and strategy in the Bosnian civil war: Explanations for the severity of violence in Bosnian municipalities. Journal of Peace Research 49(6): 801–815.

13.

Cunningham

David E

Gleditsch

Kristian Skrede

Salehyan

Idean

(2013) Non-state actors in civil wars: A new dataset. Conflict Management and Peace Science 30(5): 516–531.

14.

Downes

Alexander B

(2006) Desperate times, desperate measures: The causes of civilian victimization in war. International Security 30(4): 152–195.

15.

Eck

Kristine

Hultman

Lisa

(2007) Violence against civilians in war. Journal of Peace Research 44(2): 233–246.

16.

Fjelde

Hanne

Hultman

Lisa

(2014) Weakening the enemy: A disaggregated study of violence against civilians in Africa. Journal of Conflict Resolution 58(7): 1230–1257.

17.

Fjelde

Hanne

Schubiger

Livia

Hultman

Lisa

Cederman

Lars-Erik

Hug

Simon

(2016) Ethnicity and wartime state violence against civilians. Working paper (https://snis.ch/wp-content/uploads/2017/11/8590_final_wp_fjelde_et_al_2016_apsa_0.pdf).

18.

Forster

Till

(2015) Dialogue direct: Rebel governance and civil order in Northern Côte d’Ivoire. In: Arjona

Ana

Kasfir

Nelson

Mampilly

Zachariah

(eds) Rebel Governance in Civil War. New York: Cambridge University Press, 203–225.

19.

Gates

Scott

(2002) Recruitment and allegiance: The microfoundations of rebellion. Journal of Conflict Resolution 46(1): 111–130.

20.

Gellner

Ernest

(1983) Nations and Nationalism. Ithaca, NY: Cornell University Press.

21.

Gilmore

Elisabeth

Gleditsch

Nils Petter

Lujala

Päivi

Rød

Jan Ketil

(2005) Conflict diamonds: A new dataset. Conflict Management and Peace Science 22(3): 257–275.

22.

Gutiérrez-Sanín

Francisco

Wood

Elisabeth J

(2014) Ideology in civil war: Instrumental adoption and beyond. Journal of Peace Research 51(2): 213–226.

23.

Harbom

Lotta

Melander

Erik

Wallensteen

Peter

(2008) Dyadic dimensions of armed conflict, 1946–2007. Journal of Peace Research 45(5): 697–710.

24.

Hegghammer

Thomas

(2011) The rise of Muslim foreign fighters: Islam and the globalization of jihad. International Security 35(3): 53–94.

25.

Hoffmann

Kasper

(2015) Myths set in motion: The moral economy of Mai Mai governance. In: Arjona

Ana

Kasfir

Nelson

Mampilly

Zachariah

(eds) Rebel Governance in Civil War. New York: Cambridge University Press, 158–179.

26.

Green

Amelia Hoover

(2016) The commander’s dilemma: Creating and controlling armed group violence. Journal of Peace Research 53(5): 619–632.

27.

Horowitz

Donald L

(1985) Ethnic Groups in Conflict. Berkeley, CA: University of California Press.

28.

Huang

Reyko

(2016) The Wartime Origins of Democratization: Civil War, Rebel Governance, and Political Regimes. New York: Cambridge University Press.

29.

Hultman

Lisa

(2012) Military offensives in Afghanistan: A double-edged sword. International Area Studies Review 15(3): 230–248.

30.

Human Rights Watch (HRW) (1995) Georgia/Abkhazia: Violations of the laws of war and Russia’s role in the conflict. 7(7) (https://www.hrw.org/report/1995/03/01/georgia/abkhazia-violations-laws-war-and-russias-role-conflict).

31.

Human Rights Watch (HRW) (1997) What Kabila is hiding: Civilian killings and impunity in Congo. 9(5) (https://www.hrw.org/report/1997/10/01/what-kabila-hiding/civilian-killings-and-impunity-congo).

32.

Human Rights Watch (HRW) (2003) Trapped between two wars: Violence against civilians in Western Côte d’Ivoire. 15(14) (https://www.hrw.org/reports/2003/cotedivoire0803/).

33.

Human Rights Watch (HRW) (2005) Youth, poverty, and blood: The lethal legacy of West Africa’s regional warriors. 17(5) (https://www.hrw.org/report/2005/04/13/youth-poverty-and-blood/lethal-legacy-west-africas-regional-warriors).

34.

Humphreys

Macartan

Weinstein

Jeremy M

(2006) Handling and manhandling civilians in civil war. American Political Science Review 100(3): 429–447.

35.

Jackson

Stephen

(2006) Sons of which soil? The language and politics of autochtony in Eastern DR Congo. African Studies Review 49(2): 95–123.

36.

Kalyvas

Stathis N

(1999) Wanton and senseless? The logic of massacres in Algeria. Rationality and Society 11(3): 243–285.

37.

Kalyvas

Stathis N

(2006) The Logic of Violence in Civil War. Cambridge: Cambridge University Press.

38.

Kasfir

Nelson

(2015) Rebel governance – Constructing a field of inquiry: Definitions, scope, patterns, order, causes. In: Arjona

Ana

Kasfir

Nelson

Mampilly

Zachariah

(eds) Rebel Governance in Civil War. New York: Cambridge University Press, 21–46.

39.

Kaufmann

Chaim

(1996) Possible and impossible solutions to ethnic civil wars. International Security 20(4): 136–175.

40.

Darryl

(2010) A universal enemy? ‘Foreign fighters’ and legal regimes of exclusion and exemption under the ‘Global War on Terror’. Columbia Human Rights Law Review 41(2): 355–427.

41.

Lin

Nan

(2001) Social Capital: A Theory of Social Structure and Action. New York: Cambridge University Press.

42.

Lujala

Päivi

(2009) Deadly combat over natural resources: Gems, petroleum, drugs, and the severity of armed civil conflict. Journal of Conflict Resolution 53(1): 50–71.

43.

Lyall

Jason

(2010) Are coethnics more effective counterinsurgents? Evidence from the second Chechen war. Americal Political Science Review 104(1): 1–20.

44.

Malet

David

(2013) Foreign Fighters: Transnational Identity in Civil Conflicts. Oxford: Oxford University Press.

45.

Malet

David

(2016) Foreign Fighter Data: 2016 Version (http://davidmalet.com/The_Foreign_Fighter_Project.php).

46.

Mampilly

Zachariah

(2011) Rebel Rulers: Insurgent Governance and Civilian Life During War. Ithaca, NY: Cornell University Press.

47.

Manekin

Devorah

(2013) Violence against civilians in the Second Intifada: The moderating effect of armed group structure on opportunistic violence. Comparative Political Studies 46(10): 1273–1300.

48.

Mason

T David

Krane

Dale A

(1989) The political economy of death squads: Toward a theory of the impact of state-sanctioned terror. International Studies Quarterly 33(2): 175–198.

49.

Metelits

Claire M

(2009) The consequences of rivalry: Explaining insurgent violence using fuzzy sets. Political Research Quarterly 62(4): 673–684.

50.

Moore

Pauline

(2018) It takes two: Combatant–civilian embeddedness and the emergence of systems of exchange during civil war. Working paper (http://www.paulinemoore.org).

51.

Nir

Arielle

(2012) In search of meaning: Foreign volunteers in the Croatian armed forces, 1991–1995. Contemporary European History 21(1): 1–17.

52.

Olson

Mancur

(1993) Dictatorship, democracy, and development. American Political Science Review 87(3): 567–576.

53.

Ottman

Martin

(2017) Rebel constituencies and rebel violence against civilians in civil conflicts. Conflict Management and Peace Science 34(1): 27–51.

54.

Parkinson

Sarah E

(2013) Organizing rebellion: Rethinking high-risk mobilization and social networks in war. American Political Science Review 107(3): 418–432.

55.

Salehyan

Idean

(2009) Rebels without Borders: Transnational Insurgencies in World Politics. Ithaca, NY: Cornell University Press.

56.

Salehyan

Idean

(2010) The delegation of war to rebel organizations. Journal of Conflict Resolution 54(3): 493–515.

57.

Salehyan

Idean

Siroky

David

Wood

Reed M

(2014) External rebel sponsorship and civilian abuse: A principal–agent analysis of wartime atrocities. International Organization 68(3): 633–661.

58.

Schneider

Gerald

Bussman

Margit

Ruhe

Constantin

(2012) The dynamics of mass killings: Testing time-series models of one-sided violence in the Bosnian civil war. International Interactions 38(4): 443–461.

59.

Schutte

Sebastian

(2015) Geography, outcome and casualties: A unified model of insurgency. Journal of Conflict Resolution 59(6): 1101–1128.

60.

Schutte

Sebastian

(2017) Geographic determinants of indiscriminate violence in civil wars. Conflict Management and Peace Science 34(4): 380–405.

61.

Staniland

Paul

(2012) Organizing insurgency: Networks, resources, and rebellion in South Asia. International Security 37(1): 142–177.

62.

Staniland

Paul

(2014) Networks of Rebellion: Explaining Insurgent Cohesion and Collapse. Ithaca, NY: Cornell University Press.

63.

Stanton

Jessica A

(2015) Regulating militias: Governments, militias, and civilian targeting in civil war. Journal of Conflict Resolution 59(5): 899–923.

64.

Stewart

Megan A

Liou

Yu-Ming

(2016) Do good borders make good rebels? Territorial control and civilian casualties. Journal of Politics 79(1): 284–301.

65.

Toft

Monica Duffy

(2003) The Geography of Ethnic Violence: Identity, Interests, and the Indivisibility of Territory. Princeton, NJ: Princeton University Press.

66.

United Nations (2002) The impact of mercenary activities on the right of peoples to self-determination. Human Rights Fact Sheet no. 28. Geneva: Office of the UN High Commissioner for Human Rights.

67.

Valentino

Benjamin

(2004) Final Solutions: Mass Killing and Genocide in the 20th Century. Ithaca, NY: Cornell University Press.

68.

Walter

Barbara F

(2017) The new new civil wars. Annual Review of Political Science 20: 469–486.

69.

Weinstein

Jeremy

(2007) Inside Rebellion: The Politics of Insurgent Violence. Cambridge: Cambridge University Press.

70.

Wood

Reed M

(2010) Rebel capability and strategic violence against civilians. Journal of Peace Research 47(5): 601–614.

71.

Wood

Reed 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.

72.

Wood

Reed

Thomas

Jakana L

(2017) Women on the frontline: Rebel group ideology and women’s participation in violent rebellion. Journal of Peace Research 54(1): 31–46.

73.

World Bank (2017) World Development Indicators (http://data.worldbank.org/data-catalog/world-development-indicators).

74.

Wucherpfennig

Julian

Metternich

Nils W

Cederman

Lars-Erik

Gleditsch

Kristian Skrede

(2012) Ethnicity, the state, and the duration of civil war. World Politics 64(1): 79–115.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.22 MB