Who Can We Trust with a Gun? Information Networks and Adverse Selection in Militia Recruitment

Abstract

How do the leaders of nonstate armed groups recruit new members? Most studies of recruitment of combatants focus on explaining the supply of fighters—who fights and what kinds of people volunteer to fight depending on the incentives offered. We know comparatively little about how the leaders of armed groups manage influxes of volunteers to ensure their quality. This article examines the questions of who recruits fighters, and how the capacities of recruiters affect the quality of the individuals who they recruit. The histories of three understudied civil militias in Sierra Leone are used to develop and refine a theory of screening in nonstate armed groups. Evidence from intensive fieldwork suggests that access to civilian information networks can allow the leaders of armed groups to successfully screen recruits and exclude low-quality types even when the pool of volunteers is flooded with opportunists.

Keywords

civil wars conflict social networks internal armed conflict

The hallmark of everyday life within civil warfare is not violence; it is uncertainty. Armed actors, whether insurgents or counterinsurgents, face similar challenges of gathering reliable information about friends and foes. This article explores the informational challenges that armed groups face when trying to recruit trustworthy fighters who, usually under limited supervision, will faithfully execute their duties. Studies of rebel recruitment (Weinstein 2005, 2007) suggest that material resource endowments are the primary determinant of the severity of adverse selection problems—that is, prospective fighters whose concealed opportunistic motives conflict with the goals of the organization. Weinstein has argued that armed organizations with access to material resources (e.g., diamonds) have the advantage of being able to offer incentives to entice new recruits. However, this ostensible advantage is ultimately a curse because such incentives tend to attract larger numbers of opportunists who crowd out more selfless or trustworthy individuals.

Inductively, the resource-curse theory is a poor fit for the cases of civil militias in Sierra Leone. More generally, armed groups may operate for many years with access to the same material resources, but undergo major changes in the quality of individuals who they recruit. As Hegghammer (2013) has noted with regard to terrorist recruitment, explaining differential recruitment in armed organizations requires consideration not only of the supply of willing fighters (Petersen 2001; Wood 2003; De Mesquita 2005; Humphreys and Weinstien 2008) but also of the strategies and tactics that recruiters employ in order to manage influxes of volunteers. I argue that recruiters can use social networks to ameliorate problems of adverse selection in armed groups by proactively gathering private information about the motivations and skills of prospective fighters.¹ Access to such critical information enables the screening and exclusion of undesirable types of would-be fighters even when such types are in relatively great supply. The successful vetting of militia recruits is of significant practical importance because it has the potential to preempt downstream agency problems, such as looting and abuse of civilian populations, that derive from individual opportunism under conditions of minimal supervision.

This article uses the cases of civil militias that operated in Sierra Leone (from 1992 to 2001) to develop and test a theory of how recruiters in armed groups can use information networks to screen recruits and exclude undesirable types. Given the importance of loot-able diamond wealth and greed-based motives in shaping the civil war in Sierra Leone (Collier and Hoeffler 2004; Lujala, Gleditsch, and Gilmore 2005), the conflict provides a challenging set of micro-level cases with which to evaluate theories that focus on social networks rather than on material resources and economic incentives.

As a political phenomenon, civil militias are an understudied category of nonstate armed groups. At the national level, civil militias are typically distinguished from rebel or insurgent groups by their relationship to the state. Civil militias, at their inception, are allies or extensions of official state militaries and defenders of political incumbents (Jentzsch, Kalyvas, and Schubiger 2015) However, this definition is of limited utility because the relationship between nonstate armed groups and the state is subject to change over the course of a conflict (Staniland 2015). This article emphasizes the extent to which civil militias and insurgent forces are similar at the micro-level because both kinds of groups face analogous managerial dilemmas (Cohen and Nordas 2015), irrespective of their relationship to the state. Counterinsurgent militias, like insurgent forces, tend to solve managerial problems in an ad hoc fashion, using local or foreign resources, while having inherently limited (or no) access to the significant infrastructural power of the state.

This article begins by clarifying the theoretical expectations of supply-side versus demand-side theories of recruitment. I briefly rehearse the resource-curse theory and state its testable implications in terms of the quality of fighters who are likely to join a given armed group. I then use the “least-likely” case (Eckstein 1975; Lijphart 1971) of a civil militia that recruited members from a diamond-rich area of Sierra Leone to construct a demand-side theory of how the leaders of nonstate armed groups can use information networks to screen volunteers during recruitment processes. Turning to the history of two other civil militias, I test the theory of information networks against supply-side explanations. I conclude by exploring the theory’s external validity and policy implications with reference to the case of recruitment in the Afghan Local Police (ALP).

Theorizing Adverse Selection Problems in Militia Recruitment

The leaders of armed groups face problems of adverse selection that are analogous to the dilemmas faced by employers in civil labor markets (Miller 1992; Greenwald 1986). The key characteristic of adverse selection problems in both civil and military contexts is asymmetrical information between principals and agents. In the case of militia recruitment, recruiters want armed agents who are martially skilled and trustworthy, but have limited information about the relevant characteristics of would-be fighters. The prospective members of armed groups have full knowledge of their quality or type, but cannot easily convey reliable information to recruiters.

If the benefits of membership in an armed group were only appealing to the kinds of individuals who leaders wanted, there would be no problems of adverse selection. But, the benefits of membership often have broad appeal. Material incentives, such as wages or other goods, may attract a disproportionately large number of undesirable types of volunteers and encourage them to lie about their motivations and skills (Weinstein 2005, 605). Also, membership in an armed organization presents unemployed young men and others on the lower rungs of society with unique opportunities for empowerment and advancement (Hoffman 2011). The seductive power of wielding a weapon tends to attract exactly the kind of opportunistic volunteers whom leaders of armed groups would prefer to exclude. Thus, recruiters will always face problems of adverse selection, but the intensity of those problems will vary according to the costs and benefits associated with membership.

A complete model of voluntaristic recruitment in armed groups requires an account of the supply of, and demand for, fighters. Recruitment processes begin when recruiters assess the resources available to them and choose the level of recruitment incentives to be offered (Weinstein 2005, 2007). Civilians then weigh the potential benefits of joining a given armed group (given the incentives offered) against the costs, and self-select into, or out of, the group. If costs are high and this becomes public knowledge, some individuals may decide to remain a civilian rather than submit to the costs of joining. Given the set of individuals who self-select into the pool of volunteers, recruiters can then decide whom to admit into the group. If recruiters decide to be selective, their abilities to do so are contingent on the availability of private information about the quality of individual volunteers. Once the volunteers have been screened, they undergo the rites of passage necessary to join the group. If these rites impose high costs—for example, physical and psychological trauma—there may be some attrition of volunteers during the induction process. The individuals who make it through induction then become full-fledged members of the armed group. The quality of any new cohort of fighters is ultimately conditioned by the three mechanisms described previously, namely, self-selection, screening, and attrition.

Self-selection: Resources and Incentives

The most prominent theory addressing the self-selection of volunteers into armed groups is Jeremy Weinstein’s theory of the rebel resource curse. His theory suggests that recruiters who have access to significant material resources will tend to offer recruitment incentives and will consequently attract a large proportion of low-quality volunteers whose primary motivations for joining are opportunistic—focused on self-enrichment, rather than furthering the goals of the armed organization (Weinstein 2005, 603). In contrast, recruiters without access to material resources rely on social ties and promises of future rewards to entice volunteers. The absence of material incentives discourages opportunists and attracts high-quality volunteers whose primary motivations for joining are intrinsic and rooted in solidarity with the underlying identity or goals that characterize the armed organization.

In the resource-curse theory, recruiters gain information about new recruits through a “signaling” mechanism, that is, individuals signal their type when they accept the offered incentives to join an armed organization. However, it is misleading to frame the resource-curse theory in terms of a signaling game, because recruiters only receive the signal (of acceptance or rejection of a given set of incentives) after an individual has already joined. Thus, leaders cannot use this information to make informed decisions about whom to exclude from membership in their armed organization. According to the resource-curse theory, strategies of proactively reducing informational asymmetries through “information gathering, vouching, and costly induction are more likely to be used by rebel organizations that rely on social endowments” (Weinstein 2005, 607). The corollary of this assumption is that groups with significant material resource endowments are less likely to engage in proactive strategies for reducing informational asymmetries. Thus, the theory suggests that recruiters will employ information-gathering strategies when they need them least and will eschew such strategies when they need them most. These implications are paradoxical to the point of being unrealistic. Setting aside these assumptions, the key insight of the resource-curse theory is that available resources may affect the incentives that recruiters offer, and incentives offered may, in turn, affect the quality of individuals who volunteer to join a given armed group. The theory supplies two sequential hypotheses with clear empirical implications:

Hypothesis 1: When the leaders of armed groups have access to significant material resources (e.g., diamonds or government sponsorship), they will tend to offer recruitment incentives.

Hypothesis 2: When the leaders of armed groups offer recruitment incentives, they will tend to attract a lower quality pool of volunteers.

If remunerative incentives attract more unwanted recruits, then increased hardships and costs of joining should discourage unwanted recruits. The next section explores this hypothesis.

Self-selection and Attrition: The Costs of Fighting

Civilians self-select into, or out of, the pool of volunteers on the basis of costs as well as benefits associated with joining an armed group. Costs may be imposed by group leaders—for example, enforced periods of hazing and training (Weinstein 2005, 607)—or may be contextual—for example, the risk of being killed in battle. If the costs associated with joining are public knowledge, then prospective militia members may self-select out of the recruitment process and never enter the pool of volunteers. If costs are not public knowledge, volunteers may choose to drop out once they are directly confronted with said costs.² Thus, the costs of membership may influence the quality of fighters in armed groups through two mechanisms, namely, self-selection and attrition.

For the costs of induction to have predictable effects on the quality of recruits, either the costs or the correlative benefits of joining must be discriminating in terms of the quality of prospective members. That is, high-quality individuals must be more willing, on average, to bear the costs of membership than low-quality individuals. The literature on “costly signaling” identifies a class of discriminating actions “that are too costly for a mimic to fake but affordable for the genuine article, given the benefit that each can expect in the situation” (Gambetta and Hamill 2005, 11). From the perspective of civilians, one of the potential benefits of joining an armed group is the ability to contribute to the long-term social or political goals of the group (Weinstein 2005, 2007). Presumably, such long-term, intangible benefits will only be attractive to individuals who are by definition intrinsically motivated and dedicated, hence high-quality recruits. The study of desertion and side-switching by Oppenheim et al. (2015) provides empirical support for this conjecture: as the physical and psychological costs of membership rise, intrinsically motivated combatants are less likely to drop out than their economically motivated peers. If we assume that attrition during recruitment functions in a similar fashion to desertion, then:

Hypothesis 3: Higher costs associated with joining an armed group will discourage low quality individuals, resulting in a higher quality pool of volunteers.

The following section moves beyond a focus on the choices that prospective fighters make about whether to join an armed organization. The most understudied stage of the recruitment process is that in which recruiters exercise the most control. Given a pool of volunteers, recruiters have varying capacities for actively investigating the quality of volunteers and filtering out undesirable types.

Screening: Access to Information Networks

This section offers a theory of screening mechanisms with deductive foundations in the economic literature on labor markets (Spence 1974; Greenwald 1986). Screening refers to a general set of strategies that employers can use to gather private information about job seekers, thus reducing informational asymmetries and ameliorating problems of adverse selection. Screening strategies often involve the use of information networks to collect recommendations or referrals that reveal private information about prospective employees (Montgomery 1991; Rees 1996). Empirical studies of employment markets suggest that employers use both formal and informal networks to gather information. Formal networks utilize institutions such as “state employment services, private fee-charging employment agencies…and school or college placement bureaus” (Rees 1996, 559). Informal networks typically involve referrals from existing employees, who vouch for the skills and motivations of new job applicants (Rees 1996, 562). Both kinds of networks provide employers with information that allows more selective recruitment of new employees. Mimicking firms in a civil labor market, state militaries can carry out their recruitment using formal and informal networks, whereas nonstate armed groups (by their very nature) will typically only have access to informal networks.

A Theory-building Case: The Donso Militia

The inductive basis for the theory of information networks and screening comes from the case of the Donso militia that formed and operated in Kono District, in eastern Sierra Leone. I use examples from Donso recruitment strategies to adapt theories from civil labor markets so that they can be applied to extra-governmental armed organizations. The history of Donso recruitment in Sierra Leone suggests that militias tended to form within well-established, local systems of reciprocity and informational exchange. These networks of reciprocity were essentially ubiquitous in Sierra Leone, and figures of provincial political authority derived their practical power by establishing themselves as central hubs of networked exchanges (Acemoglu, Reed, and Robinson 2014). The Donso militia started in 1993 when community leaders, called “chiefs,” responded to calls by members of the Sierra Leone Parliament to organize local defensive forces. The chiefs who took charge of Donso recruitment found that they could mobilize existing political patronage networks in order to gather information about, and thereby screen, would-be members of the nascent militia.

The Donso call-to-arms was disseminated through a hierarchy of provincial authorities, from high-ranking Paramount Chiefs to Section Chiefs, to Town Chiefs and local elders, until it reached individual communities (Hoffman 2011, 74). Town Chiefs used their privileged local knowledge to gather recommendations from the family members or guardians of would-be recruits to select people who they knew were likely to act as faithful servants of the hierarchy of chiefs. Interviews with former militia members suggested that there were, indeed, concerns about recruiting able-bodied individuals, but these martial characteristics were of secondary consideration to recruits’ loyalties.³

Keenly aware of the fault lines within their own communities, chiefs were legitimately concerned with delegating defense to people who might later turn against them. A chief who was in charge of Donso recruitment explained that he was specifically looking for individuals who were friendly to governmental authority and had positive reputations in their communities, because such individuals were the least likely “to betray [their] people” (see endnote 3). In particular, chiefs saw traditional hunters as ideal recruits because of their combination of martial skills and their status of being deeply embedded in chiefly patronage networks. Recommendations for recruits would then flow back up the same hierarchy until they reached the Paramount Chiefs who would grant final approval.

The success of this system of screening relied heavily on the ability of Paramount Chiefs to access extensive hierarchies of lesser chiefs who were embedded in networks of local knowledge, which included important information about the character of would-be fighters. The density of these networks was crucial because it allowed local knowledge to percolate upward and reach important regional decision makers. Working without formal institutions, and often without written records of any kind, Paramount Chiefs could not, on their own, have known enough about the thousands of individuals who would come before them as candidates to join regional militias. Yet with informal informational networks at their disposal, Paramount Chiefs were well positioned to make informed decisions about who could join civil militias.

Within the systems that chiefs established, individuals could only be nominated for membership in a militia if they had family members who could vouch for their character. The individuals who were providing character references on behalf of would-be fighters would have to appear before a local chief. The referee would answer the chief’s questions, which usually pertained to the individual’s family lineage, role in the community, and criminal background. If the chief was satisfied with the recommendation, he would create a document bearing his signature, and the prospective recruit would then carry that document to the regional Paramount Chief, who had the final say regarding his candidacy. Alternately, the local chief would create a list of worthy candidates that would then be sent by courier to the Paramount Chief. The recommendations that community members delivered to local chiefs were at least as reliable as recommendations provided in the context of a civil job market. In nearly all cases, the local chief who received recommendations would personally know the individuals providing those recommendations, and it was understood that recommenders would be held responsible for the actions of the individuals whom they recommended.

The Donso case suggests, more generally, that the ability of armed groups to solve problems of adverse selection primarily depends on the presence of densely networked, hierarchical linkages between community leaders with deep local knowledge and higher ranking officials with the authority to select recruits. Variations in the availability of social and patronage networks arise during the course of a conflict because such networks are inherently fragile. Violence can disrupt networks by causing forced migration, death, and mistrust among civilians. Because chiefs were major hubs within networks of reciprocity and clientelistic exchange, the loss of even a few chiefs from a given network (through death or migration) could fragment a network to the point that entire communities would no longer be connected with each other, and both the speed and quality of information flows would decrease (Barr, Ensminger, and Johnson 2009).

To derive clear and falsifiable predictions, I assume that whenever groups have access to intact information networks, they will use them. The theory of information networks and screening thus suggests the following hypothesis:

Hypothesis 4: Groups with access to intact information networks will be more effective at screening volunteers and will induct higher quality members.

Research Design

I turn to the history of two additional civil militias in Sierra Leone to evaluate the theory against alternative explanations. Those two militias were contemporaneous with, but operated autonomously from, the Donso and thus provide independent evidence with which to test the theory of information networks. The founders of these two militias gave the groups names that reflect their regional-ethnic origins: in the north, the Tamaboro, and in the south, the Kamajor. Like the Donso, the Tamaboro and Kamajor militias formed in response to encouragement from members of the Sierra Leone government. In each of these cases, government officials called upon regional and local chiefs to manage processes of militia formation. The centrality of chiefly leadership in the Donso, Tamaboro, and Kamajor militias leant a high level of uniformity to the initial managerial strategies in those incipient armed groups.

The empirical portion of this article presents the history of the Tamaboro and Kamajor civil militias as a longitudinal and cross-sectional process tracing analysis that can also be thought of as four distinct case studies. Each regional history is parsed into two time periods—early (up to 1996) and late (from 1996, onward)—yielding four cases. Both the Tamaboro and Kamajor militias were founded with access to significant material resources, and thus are cases in which resource-curse mechanisms were most likely to be observed, and screening and costly induction mechanisms were least likely to be activated. Despite theoretical expectations to the contrary, both armed groups managed to establish systems of screening through information networks, recruiting cohorts of high-quality volunteers. Yet war has a tendency to destabilize and reconfigure systems of social control, and the Tamaboro and Kamajor militias each experienced a series of crises that disrupted existing systems of screening. These violent setbacks present a useful set of historical conditions for analyzing the causes and consequences of changes over time in militia recruitment strategies. The fact that each militia recruited its members from the same regional pool of civilians before and after the crises helps to control for numerous unobservable factors associated with regional differences in prewar social and political conditions as well as regional geographies and demographics.

The four case studies, summarized in Table 1, were chosen in an attempt to maximize variation in the contextual factors that might have affected recruit quality. These cases exhibit variations in the availability of material resources, the costs associated with militia membership, and the availability of intact information networks, facilitating a parallel examination of the self-selection, screening, and attrition mechanisms that are likely to affect the quality of individuals who are recruited into militias. In terms of the measurement of relevant constructs, I evaluate competing theories based not only on the correlation of causal variables with the outcome variable (recruit quality) but also on the operative processes or mechanisms that connect key determinants with recruitment outcomes.

Table 1.

Summary of Case Study Variation.

	Determinants of recruit quality				Outcomes
	Hypothesis 1: Material resources	Hypothesis 2: Recruitment incentives	Hypothesis 3: Costs of membership	Hypothesis 4: Chiefly screening	Recruit quality
Tamaboro 1992 to 1994	Yes	Yes	Low	Yes	High
Northern CDF 1997 to 2001	No	No	Low	No	Low
Early Kamajor 1993 to 1996	Yes	No	High	Yes	High
Kamajor CDF 1997 to 2001	Yes	No	High	No	Low

Note: CDF = Civil Defense Force.

Measurement

This article is premised on a somewhat controversial claim—that it is possible to measure recruit quality. Jeremy Weinstein (2005, 607) has observed that the issues of asymmetric information that give rise to adverse selection problems during recruitment also give rise to problems of measurement during social-scientific analysis. The true quality of recruits is private information that is not necessarily related to observable traits of individuals.

The Donso case suggests that chiefs prioritized the loyalty of recruits above all. To be sure, chiefs wanted individuals who would be successful on the battlefield, but this appears to have been a secondary consideration. Chiefly preferences suggest two important indicators for the quality of recruits, namely, age and prior experience as hunters.

Young men constituted the most readily available and physically capable source of military labor, but they also had much to gain by overthrowing systems of chiefly authority that seemed only to perpetuate their status as the lowest of the low (Abdullah and Muana 1998; Fanthorpe 2001; Peters 2011). Men who were past their mid-twenties were far more likely to fit chiefs’ descriptions of the ideal militia member. Such men had probably sired children and become embedded in their communities—having been integrated into local patronage networks as well as secret networks of social accountability embodied in the regional, male initiation societies that are common throughout Sierra Leone (Leach 1994; Ferme 2001). At some point around forty-five to fifty years of age, increases in age will not be associated with further increases in loyalty, but may actually be associated with some losses in physical capacity as a fighter. Thus, chiefs probably considered men aged between twenty-five and forty-five to be ideal recruits in terms of both their potential loyalty and their fitness.

In addition to age, prior experience with a firearm was an even more reliable indicator of a recruit’s potential loyalty to chiefs. Most of the individuals who legally owned guns during peacetime in Sierra Leone were “traditional” hunters who owed their ability to hunt to chiefly authority. The Chiefdom Council Act from the 1960s established systems of informal character references in which “firearms permits were issued by police authorities on the recommendation of village headmen and the local chief” (Alie 2005, 74). Chiefs showed a strong preference for recruiting local hunters, not only because they could easily adapt their hunting skills to stalking human prey but also because they were already embedded in networks of chiefly patronage. Hunters derived their right to own guns from chiefly authority and acknowledged that authority any time they hunted by sharing a portion of their bounty with their chief.⁴

Age and prior experience with firearms are less-than-ideal measures of the underlying construct of the loyalty (i.e., quality) of prospective recruits. Chiefs who had ample information and time would have delved much deeper—inquiring about a candidate’s family background, criminal record, and community status. As a foreign researcher, I could not hope to gain access to the same levels of private information as chiefs. Thus, I employ age and prior experience with firearms as proxies for recruit quality because they are readily measurable traits that could have been strategically misrepresented during chiefly screening interviews, but were unlikely to be misrepresented during a casual interview with a presumably disinterested researcher. Under asymmetrical information, volunteers could have easily lied about their age (e.g., a sixteen-year-old could claim to be nineteen) and hunting experience, in the same way that they could lie about whether they had criminal records. Only locally held knowledge could unveil such lies during screening processes. To the extent that chiefs were able to access important private information about volunteers, this should be reflected in the quality of militia recruits as measured through age and hunting experience.

I operationalize material resources as sponsorship from a national-level political patron, or access to loot-able resources (e.g., alluvial diamond deposits that could be mined using minimal infrastructure). When they are present and accessible, material resources only affect recruitment outcomes when commanders actually convert those resources into recruitment incentives that are offered to attract prospective members. Finally, recruitment incentives affect recruit quality through the mechanism of self-selection of prospective recruits into the pool of volunteers. When recruitment incentives are broadly attractive (to both high- and low-quality types), we expect to see large pools of eager volunteers. This is in contrast to a situation in which militia membership is more narrowly attractive to intrinsically motivated volunteers who will show up in comparatively small numbers to join.

The costs of membership can take the form of enforced periods of training and hazing or the payment of a membership fee. The risks associated with joining a militia are more subjective. I assume that civilians attach a lower level of risk to militia membership when militias have the upper hand over their military rivals. To the degree that costs affect recruit quality through the mechanism of self-selection, we should observe a larger pool of volunteers in response to lower costs (and vice versa). Costs can also affect recruit quality through attrition, which can be measured in terms of the number of prospective members dropping out during initiation processes.

I operationalize information networks as hierarchical patronage flows among chiefs and between chiefs and the members of their communities. When assessing the effects of information networks, I look for hierarchies of chiefs transmitting local knowledge about recruits. Information networks are considered disrupted when a number of high-ranking (hence well networked) chiefs are killed or permanently displaced from their communities. When networks are intact, they affect recruit quality through the mechanism of screening, which involves individuals who are intimately familiar with recruits vouching for their character. The killing or displacement of chiefs provides evidence of the disruption of networks. I use oral histories to determine whether or not chiefs used intact networks to screen recruits.

The following sections analyze the parallel histories of the Tamaboro and Kamajor militias. I begin with an overview of the national-level historical context in which both militias formed and operated, highlighting major shifts in the risks associated with becoming a militia member versus remaining a civilian. I then present the regional-level history of each militia as a set of paired case studies, parsed at the midpoint of the war. I conclude the empirical analysis with a discussion of the correspondence between the process tracing evidence and major hypotheses.

National-level Context

The National Provisional Ruling Council (NPRC) coup in 1992 marked the beginning of a concentrated period of state-orchestrated militia formation. Representatives of a newly installed military government traveled to the northern and (war affected) southeastern regions of Sierra Leone and contacted local chiefs, asking them to assist with the creation of local defensive militias. Enlisting the support of chiefs was a logical move, given the historical relationships between chiefs and armed mobilizations in the countryside. During the precolonial and colonial eras, chiefs in Sierra Leone had maintained personal militias of “war boys” who could be rapidly mobilized for offensive or defensive purposes against neighboring chiefdoms, or most famously, against the colonial administration during the Hut Tax War of 1898 (Abraham 1976, 65). In the postcolonial context, chiefs retained significant social influence and organizational resources that allowed them to manage militia recruitment.

During the early years of militia recruitment, the potential efficacy of forming local defensive militias was completely untested, but civilians generally greeted the idea with optimism. Positive expectations about what militias might be capable of accomplishing were bolstered by the fact that militia formation was being underwritten by members of a popular military government. As a result, chiefs did not struggle to find volunteers to serve in early militias. Civilians were willing to join militias not only out of a sense of patriotic duty but also because of the prestige and excitement associated with these early, popular mobilizations. Early militia members hoped that their service would help bring the war to a quick close, allowing them rapidly to return to civilian life. Yet initial successes of these defensive mobilizations proved hollow.

The case studies that follow are parsed at the midpoint of the war in 1996, because that year corresponds to a major national-level shift in the collective fortunes of civil militias throughout the country. Sierra Leone is a small country, and the Tamaboro and Kamajor militias were both subject to the same major shifts in the balance of power with their enemies. From 1995 to early 1997, the Tamaboro and Kamajor militias benefited militarily from an alliance with state-hired counterinsurgency experts—a private military company known as Executive Outcomes (EO)—and then both militias were seriously shaken by the Armed Forces Revolutionary Council (AFRC) military coup in May 1997, which followed shortly after the departure of EO. EO operators significantly increased the military effectiveness of civil militias by injecting them with force-multiplying factors, including communications technology, air support, and high levels of strategic and tactical expertise (Johnston 2008, 133). The AFRC coup rapidly reversed the military gains that counterinsurgent forces had made during the period of EO operations. The coup was staged by a sizable contingent of the Sierra Leone Army that seized control of the capital joined forces with the Revolutionary United Front (RUF) rebels against whom militias had been fighting. The rebel–junta alliance that resulted from the coup represented a major shift in the military balance of power away from civil militias and toward their newly combined enemies.

As a result of the 1997 AFRC coup, it became far more dangerous for civilians to join a militia. The combined AFRC-RUF forces were heavily armed with captured weaponry, and on the offensive. Throughout the country, militias lost significant ground before they were able to regroup and launch a coherent counteroffensive. Significant numbers of civilians still volunteered to join militias, but they were now keenly aware of the risks involved. The initial excitement and optimism that had surrounded early militia mobilizations were replaced with the somber realization that the war would probably last for several more years and that the chances of victory were deeply uncertain.

The period leading into the 1997 coup also witnessed a national-level organizational change in the civil militias in Sierra Leone. In 1996, a newly elected president, Ahmad Tejan Kabbah, tried to consolidate his control over disparate regional militias by bringing them under a single, national-umbrella organization, dubbed the Civil Defense Forces (CDF). The creation of the CDF may have led to marginal increases in the coordination among militia commanders in the field, but it did not fundamentally alter the nature of recruitment processes.

Beneath the national-level contours of the conflict, there were significant regional variations among militias. The following sections address the histories of the Tamaboro and Kamajor militias individually, highlighting unique local contexts and tracing processes that explain variations in recruit quality over the course of the war.

The Early Tamaboros in the North

The Tamaboro militia was unique in terms of the amount of support it received from the government of Sierra Leone. A senior officer in the NPRC junta provided the impetus for the formation of the Tamaboro in 1992, making it the first prostate civil militia in Sierra Leone (Gberie 2005). At the encouragement of the NPRC, chiefs issued a call to arms and promised that the government would provide volunteers with a bag of rice as “traveling allowance.”⁵ This provision of material recruitment incentives predisposed Tamaboro recruiters to face greater numbers of opportunistic joiners than Kamajor recruiters. The offer of rice attracted approximately 6,000 initial volunteers. This flood of potentially opportunistic joiners was entirely consistent with the predictions of theories that emphasize how the provision of material recruitment incentives cause large numbers of undesirable volunteers to self-select into armed organizations.

Northern chiefs could have accepted every volunteer into their militias, but instead they decided to screen new members. At least some community leaders suspected that a significant number of volunteers were attracted by the offer of rice and were not sufficiently skilled or trustworthy to make reliable fighters. Militia recruiters imagined that these opportunistic joiners might prove cowardly and difficult to control, and might bring the North a “bad image” (see endnote 5). Responding to these doubts and fears, community leaders started an elaborate process of screening volunteers, and gradually whittled the pool of several thousands down to a much smaller number—estimates vary between approximately 350 to 1,000 fighters—who were mobilized and sent to assist the Sierra Leone Army in fighting the rebels (Hoffman 2011; Fithen 1999). This time-consuming screening took place despite the fact that the RUF rebels had already captured the eastern city of Koidu Town, giving them access to the main highway, from which they could launch strikes on Freetown and Kabala.

Descriptions of the most desirable traits for Tamaboro recruits suggest that the character and community status of would-be Tamaboros was of paramount importance. Paul Kortenhoven was an American missionary who lived in Koinadugu and knew many of the early Tamaboro fighters. He recalls Tamaboro recruitment and the emphasis placed on recruits having positive reputations within their communities: “You couldn’t be a Tamaboro without […] some presence, some respect, you know by, um, your compatriots.”⁶ Kortenhoven suggested that chiefs looked for individuals who were established and contributing members of their communities and who thus had a vested interest in the local, political status quo.

During screening, parents and guardians of new recruits would have to appear before the chief to vouch for the character of the would-be militia member.⁷ These in-person recommendations served a dual purpose. They revealed otherwise hidden information about the character and background of the individuals proposing to join the militia, and they also came with the added understanding that the recommender would, to some extent, be culpable for the actions of the person whom they recommended.

As a result of careful screening, the Tamaboro militia was a relatively small group of high-quality fighters. Multiple sources suggest that the original Tamaboro consisted mostly of “big men”—meaning men who were considered mature adults and accomplished members of their communities, and most of whom were also experienced hunters.⁸ By far the smallest and most elite civil militia ever created in Sierra Leone, the Tamaboro, once selected, deployed along with a contingent of the Sierra Leone Army and managed to drive the RUF out of most of the southeastern regions of the country.

The Late Northern CDF: Dead and Displaced Chiefs

Northern militias (successors to the Tamaboro) halted recruitment in 1994 and did not resume until late 1996 or early 1997. These later recruitment drives were strongly affected by the fact that numerous civilian leaders and militia leaders had been killed during the interim. By late 1993, the Tamaboro militia, working alongside government troops, had successfully cleared the rebels out of Kono District as well as parts of Kailahun District (further south and east). At the time, the war seemed all but won, and Tamaboros promptly resumed their civilian lives. In November 1994, a resurgent RUF launched a vengeful offensive on the home of the Tamaboro in Koinadugu District (Keen 2005, 126). Rebels and defectors from the military reached as far as the major northern city of Kabala, which had served as the recruitment center for the Tamaboro. Along the way, rebels killed hundreds of civilians (including chiefs), abducted several foreign aid workers, and assassinated some of the recently demobilized leaders of the Tamaboro militia, including the legendary female fighter Marie Keita and the powerful Tamaboro leader Daembaso Samura (Gberie 2005, 83).

Recruitment in the Northern CDF was a major departure from earlier Tamaboro recruitment. Renewed recruitment drives did not involve the offer of bags of rice (or any other material incentives) to attract recruits. The formation of the Northern CDF was encouraged by the government but received minimal sponsorship. Given low material resource endowments and the nonoffer of recruitment incentives, the pool of individuals who volunteered for membership in the Northern CDF was probably of higher quality, on average, than the pool of individuals who volunteered for membership in the Tamaboro (when bags of rice were offered as recruitment incentives).

A former coordinator of the Northern CDF explained that control of recruitment was left in his hands and the hands of other local Coordinators.⁹ Civilian chiefs were generally not involved in the screening of new recruits in the Northern CDF, because many of the chiefs had either died or fled the area by that time.¹⁰ A rebel incursion during late 1994 resulted in the assassination of several civilian chiefs who were probably involved in Tamaboro recruitment, as well as some of the recently demobilized leaders of the Tamaboro militia. In 1998, another wave of rebel invasions into the North drove many of the remaining Paramount Chiefs and lower-level Section Chiefs to flee Koinadugu District for the safety of the capital city or for Guinea.¹¹ From 1998 to 2001, rebels controlled large stretches of the roads leading in and out of Koinadugu District (to the point that humanitarian aid and Economic Community of West African States Monitoring Group military supplies had to be flown into Kabala Town by helicopter) making it impossible for chiefs who had fled to return to their homes.

The recruitment of new members in the Northern CDF reflected the serious disruption of earlier recruitment networks. Given the death and flight of local civilian leaders, the task of gathering private information about would-be militia members fell almost exclusively to militia commanders. A police officer who was based in Kabala during the war suggested that the recruitment of the Northern CDF was much more haphazard and cursory than that of the Tamaboro and that very few (if any) civilian chiefs or elders were involved in the vetting of recruits for the Northern CDF.¹² Interviews with individuals recruited from 1997 to 1999 provide independent confirmation of the fact that many new recruits were interviewed, but only by men who were inside the CDF, and few (if any) of whom were chiefs.

Coordinators in the Northern CDF attempted to screen fighters, but found that they could do little with the limited information at their disposal. Northern screening processes may have still excluded some undesirable individuals, but qualitative evidence suggests that few, if any, of the individuals who tried to join were actually rejected.

Despite the fact that the Northern CDF did not face a large pool of opportunistic volunteers, recruits who joined the Northern CDF were of decidedly lower quality than their Tamaboro predecessors. Fighters who joined the new Northern CDF were generally much younger, and very few of them claimed any hunting experience. A former member of both the original Tamaboro and (later) the Northern CDF lamented the influx of volatile youths into the Northern CDF:

The only problem was that youths are hard to control. Later when they came from the battlefront and there was no food for example, they would get annoyed and […] threaten to kill everyone present.¹³

The Early Kamajors in the South

The Kamajor militia formed, circa 1993 to 1994, in southern regions that contained deposits of easily mined, alluvial diamonds. Although Kamajor factions did not offer recruitment incentives to attract volunteers, the Kamajor initiation process provided recruits with the inestimable benefit of being made magically bulletproof. On its own, this benefit could have attracted a significant number of individuals who volunteered to join the Kamajors simply to undergo the protective rituals that initiators offered. However, this unique benefit of Kamajor membership was, to some extent, counterbalanced by the fact that the associated rites of passage were known to be physically and psychologically traumatic (Hoffman 2011). At least in the early stages of the war, the paired costs and benefits of initiation appear to have cancelled each other out and preempted adverse selection problems that would have plagued the Kamajors had induction into the militia been completely costless to volunteers. It was uncommon for civilians to join the Kamajors exclusively to gain access to the protective benefits of initiation ceremonies. Yet chiefs still had reason for concern.

A former Kamajor administrator provided one of the most direct statements of the adverse selection problem in incipient militias in the South. He accented the difficulties involved in controlling groups of young community members who were not only armed, but also magically empowered through bullet-proofing initiation rites:

All the paramount chiefs and the people said: “Now, we have a lot of people who are bad fellows in our chiefdoms. There are a lot of young men who are wayward. If you initiate them into this [Kamajor] society—you protect them—they will turn their guns against their own people.”¹⁴

The secret rites necessary to join the Kamajor society could only be carried out by a small number of “initiators,” known for their special powers to make people invulnerable. Even with the added step of initiation, it is clear that the screening process prior to initiation involved the same kind of familial recommendation as in the Tamaboro group and that the Paramount Chiefs were still the sole arbiters of membership in the Kamajor militia:

When you want to go join […] Your mother and father and sister and brother take you to the Paramount Chief. Then the Paramount Chief recommends you to the initiators.¹⁵

Early initiates into the Kamajors recalled that new recruits were sent to initiators in groups that the Paramount Chiefs had selected and that carried the authorization of the chief’s signature. It was unthinkable for someone to go on his own and without the endorsement of the Paramount Chief. In both the Tamaboro and Kamajor militias, these screening systems were founded on the ability of Paramount Chiefs to collect information through hierarchies of lesser chiefs who garnered and transmitted local knowledge through existing patronage networks.¹⁶

Classes of Kamajor initiates from 1993 through 1996 were typically of very high quality. My interviews with individuals who joined the early Kamajor movement suggest that they were almost all experienced hunters. Elizabeth Lavalie, whose husband (Alpha Lavalie) had helped to start and run a southeastern chapter of the Kamajors, had engaged with the Kamajors and their leadership throughout the course of the conflict. She further confirmed that early Kamajor recruits were almost exclusively hunters and mature men who “had their principles, and you will not be recruited as a [Kamajor] if you had any tainted character.”¹⁷

The Late Kamajors: Rogue Initiators

In contrast to northern militias, the Kamajors never demobilized or halted their recruitment in the middle of the war. As the favored ethnic militia of President Kabbah, the Kamajors were allowed to continue their recruitment and operations even as the vice president traveled the country encouraging other militias to disarm. As a result, militia leadership and recruitment practices had a high degree of continuity from the early 1990s into the middle years of the war.

By the beginning of 1997, Kamajor initiators had begun what has been described as “the commercialization of the initiation process” ( The Prosecutor v. Moinina Fofana and Allieu Kondewa , SCSL-04-14-T-790, 22001). Entrepreneurial initiators circumvented the control of civilian leaders and established a steadily growing stream of new initiates who were allowed to join as unscreened individuals, provided that they were able to pay increasing initiation fees.

Already under threat by the commercialization of Kamajor initiation, systems of chiefly screening broke down completely as a result of the AFRC coup in May 1997, which forced loyal militia members to flee into the bush. Even as Kamajors began to recapture territory from rebel occupation, systems of chiefly screening were not restored. By late 1997, the military and administrative wing of the Kamajor CDF had completely lost control over some of their initiators. Most notoriously, the Kamajor initiator, “High Priest” Allieu Kondewa, had thousands of cult-like followers whose status as militia members was not subject to chiefly approval, and who were convinced that they were superior not only to civil authorities but also to other Kamajor fighters (Arthy 2004). Perhaps learning from Kondewa’s example, a number of other initiators also began to sell bulletproofing ceremonies to individuals who were not approved by the chiefs.

By 1998, most of the chiefly networks in the South had been restored, but there was no screening of recruits in regions where initiators had strong influence. Initiators had hijacked the process of creating Kamajors and had no particular interest in the quality of the individuals whom they were initiating. As a result, recruitment of Kamajors from mid-1997 onward was effectively indiscriminate in areas controlled by initiators who inducted anyone who could afford to pay. Those new, unscreened initiates would attach themselves to an existing militia unit or simply start their own.

Similar to the early years of Kamajor recruitment, volunteers weighed the benefits of becoming bulletproof against the physical and psychological trauma of ritual hazing, which was widely known to be a part of the Kamajor initiation process (Hoffman 2011; Wlodarczyk 2009). These aspects of Kamajor initiation do not seem to have changed significantly over time.

The main alteration of the cost–benefit calculus involved the added requirement, as a result of the commercialization of initiation, that recruits pay a significant amount of money in order to join. This additional cost of joining may have discouraged some opportunistic individuals whose only reason for joining was to become bulletproof. Presumably prospective members who were strongly motivated to join the Kamajors in order to defend their communities would be more willing to pay the fee necessary to be initiated and would also have been more likely to find local patrons within their communities who would be willing to help them pay the fee.

Ultimately, recruitment drives following the AFRC coup in 1997 resulted in the induction of large numbers of low quality recruits. Kamajor recruits after 1997 were much younger than their predecessors, and very few had any experience as hunters. Elizabeth Lavalie, cited earlier, drew a stark contrast between the experienced men who joined before 1997 and the wayward young people who joined afterward. She also cites the commercialization of initiation as having been the primary cause of the precipitous decrease in recruit quality over time:

In later years […] you had these […] priests coming in [claiming]: “I can make you bulletproof.” […] And it was then you had so many young people coming in with various shady characters.¹⁸

Summary and Discussion of Case Study Evidence

The Tamaboro and Kamajor militias both started with access to material resource endowments. The Tamaboro had significant government sponsorship and the Kamajor had access to loot-able diamond deposits. If we look only at the initial distribution of resources, Hypothesis 1 would lead us to predict that both militias would have relied on incentives to attract new recruits and would have neglected to build strong ties to local communities. Contrary to these theoretical expectations, only the leaders of the Tamaboro militia made use of the available resources to offer recruitment incentives, and both militias were deeply embedded in local communities.

If we focus instead on recruitment incentives, Hypothesis 2 leads us to expect that the early Tamaboro militia would have been flooded with opportunistic recruits. Hypothesis 3 reinforces this prediction, because there were no countervailing costs associated with joining the Tamaboro: recruits were neither trained nor hazed, nor had any way of assessing the dangers that might confront them on the battlefield. Civilians self-selected into the pool of Tamaboro volunteers primarily because they knew that they would receive bags of rice. Some proportion of those volunteers were also motivated by a sense of patriotic duty, but these motivations were private information. Given the offer of incentives (Hypothesis 2), and the absence of major costs (Hypothesis 3), the supply-side predictions are clear, that is, large numbers of opportunists should have self-selected into the pool of volunteers. Thousands did, in fact, turn up in Kabala to join the Tamaboro, suggesting that the benefits associated with joining were broadly attractive, and the perceived costs were few. Yet very few opportunists actually became members of the Tamaboro.

The screening procedures that chiefs implemented whittled thousands of volunteers down to a few hundred high-quality individuals. Given the substantial nondiscriminating benefits and low costs associated with joining, self-selection and attrition mechanisms cannot explain the high levels of recruit quality in the early Tamaboro. The selectivity of chiefly screening processes is the only clear explanation.

Unlike the Tamaboros, early Kamajor volunteers faced a fairly narrow balance between the costs and benefits of membership. Most importantly, recruiters had access to significant resources in the form of easily mined diamonds, but did not attempt to liquidate those resources, and did not offer recruitment incentives. The only immediate benefit of joining the early Kamajors was the opportunity to the magical bulletproofing ceremony that was mostly untested at that point in the war. The countervailing cost associated with joining was the physical and psychological stress of the bulletproofing ceremony and related initiation rituals. Although great secrecy surrounded Kamajor initiation, it became widely known that initiation was an arduous process. These costs probably led some civilians to self-select out of the pool of volunteers, but there is no evidence that they led to attrition once individuals had entered the pool of volunteers. Given the relative balance between the immediate costs and benefits of joining, civilians self-selected into the pool of volunteers almost exclusively on the basis of nonmaterial and discriminating benefits—namely, a chance to defend their community and serve their country. Such benefits were only attractive to individuals who were, by definition, trustworthy recruits. In sum, supply-side determinants suggest that the pool of Kamajor volunteers should have been of significantly higher quality than the pool of individuals who volunteered to join the early Tamaboro.

The ranks of the early Kamajors were, in fact, filled with high-quality recruits. These individuals had been subject to chiefly screening, but it is not clear that chiefs’ efforts were necessary. The outcome in the case of the early Kamajors was overdetermined by the combined effects of self-selection and screening mechanisms.

During the second half of the war, there were greater similarities between northern and southern militias in terms of the costs and benefits that they presented to would-be militia members. Neither militia offered material incentives during recruitment, while the correlative risks associated with membership were far greater given the militarily disadvantaged position of militias after the AFRC coup.

For late militia volunteers, the cost–benefit calculus was muddled, making it difficult to derive clear predictions. In both militias, the costs of membership were clearly higher than they had ever been during the early years of the war. To the extent that opportunistic and untrustworthy types were discouraged by hardship and only interested in material benefits, they should have self-selected out of the pool of militia recruits later in the war. Yet militia membership also offered protective benefits that would have been appealing to all kinds of individuals. As the AFRC-RUF forces launched punitive raids against civilians, increasing numbers of civilians volunteered to join militias because of the safety they offered.

Especially for the Kamajors, it is not clear whether the quality of the pool of volunteers would have changed significantly from pre- to post-1996. Given rising initiation costs and rising risks associated with membership, it is highly likely that the quality of Kamajor volunteers increased marginally, but it is also possible that the pool of volunteers remained roughly the same over time.

In the north, the supply-side predictions are clearer. Given the withdrawal of material incentives, the pool of volunteers for the Northern CDF should have been of much higher quality than that of the early Tamaboros and should have been roughly equivalent to that of the late Kamajors.

Ultimately, supply-side determinants predict an increase in the quality of volunteers during the later years of the war. Yet there was a drastic decrease in the quality of militia recruits during that period. In both the North and South, my informants explained reductions in recruit quality as resulting from the demise of chiefly screening. In the North, the death and forced migration of chiefs impaired information networks and made screening ineffective; in the South, chiefs were displaced by rogue initiators who completely abandoned screening. Although the underlying historical causes were different, it is clear that recruiters’ capacities and decisions were the critical determinants of recruitment outcomes.

Conclusion

The leaders of incipient armed groups do not face a strict either/or choice between using material or social resources during recruitment. Furthermore, the mere availability of a resource (such as diamonds) is not enough to corrupt an organization or even to guarantee that it will be utilized at all (Staniland 2012). When solving novel problems, people tend to avail themselves of the tools to which they are accustomed, which in much of sub-Saharan Africa are social networks. When material incentives are offered, they do condition the pool of volunteers through the mechanism of self-selection, but the leaders of armed groups can use screening mechanisms to filter the pool of volunteers and exclude undesirable types who are attracted by material incentives.

Theories of recruitment in armed groups would benefit from a better understanding of both the recruiter’s and the recruit’s decision making. This article has argued that recruiters’ information-gathering capacities are critically important, but it is not clear why some recruiters utilize available information networks, while others do not. Demand-side theories of recruitment need to clearly distinguish capacity from will and then account for both. An important alternative demand-side explanation involves the recruiter’s assessment of the quantity (rather than the quality) of recruits needed. Why do some recruiters prefer small units of elite fighters, while others opt for massive battalions of cannon fodder? These choices are still poorly understood.

A more detailed consideration of recruits’ decision making reveals the critical importance of the nonmaterial benefits of group membership in addition to the material incentives, which are far better understood. Nonmaterial benefits can be general or discriminating (in terms of recruit quality), can change significantly over the course of a conflict, and are frequently the tie-breaking vote as civilians reckon the opportunity costs of joining an armed organization versus remaining a civilian.

To maximize the internal validity of the theory being developed and tested, I have limited the empirical cases considered to a single national context and a single type of nonstate armed group. The structure of patronage networks and their utility as conduits of information will vary with different cultural settings, social structures, and dominant technologies for transmitting information, but the fundamental problems of mobilizing and militarizing civilians—whether for the purpose of insurgency or counterinsurgency—are the same irrespective of the context. The theory of information networks is formulated and operationalized in such a way that it could be applied in any society that has established, identifiable patronage networks that have similar informational properties. The crucial aspects of information networks are their density, which facilitates the flow of information, and their capacity to allow detailed local knowledge to percolate upward through hierarchical relationships to reach individuals who control recruitment processes.

As evidence for the external validity of the theory, I briefly highlight the remarkable parallels between recruitment practices of militias in Sierra Leone and the practices established by US military forces for recruiting local militia-like police forces in Afghanistan. As a part of US counterinsurgency operations in Afghanistan (from circa August 2010), US soldiers encouraged communities to organize defensive militias known as the ALP (Hulslander and Spivey 2012). The Village Stability Operations (VSO) handbook, used by the US military in Afghanistan, reflects lessons learned from the early stages of US operations in Iraq and Afghanistan. The section of the handbook focusing on “Bottom-Up Mobilization” of the ALP places remarkable emphasis on the “screening and validation” of ALP members (Meffert and Bolduc 2011, 39-41). Councils of village elders, called Shuras, are at the center of the recruitment process. The first step in ALP recruitment is for local US forces to make certain that the village has a functioning Shura and an elected malik (tribal leader) who is recognized by the district-level government.¹⁹ The second step involves the empowering of local Shuras to recruit and screen new members (Meffert and Bolduc 2011, 41).

For US forces, the clear advantage of recruiting through local elders and Shuras was the fact that Shura members were embedded in local patronage networks and had intimate knowledge of their communities. Similar to the chiefs in Sierra Leone, the elders who were members of Shuras used their access to local knowledge to ensure that only appropriately motivated recruits would become members of the ALP.

The VSO handbook explicitly admonishes practitioners to “not shortcut the ALP screening and validation process in the effort to increase ALP numbers” (Meffert and Bolduc 2011, 46). Poorly vetted recruits can easily become a security threat. The handbook also places an explicit emphasis on recruiting individuals who had “extensive family ties to the area” and who would be likely to be loyal to local leaders (Meffert and Bolduc 2011, 49).

Given the gradually increasing US military presence in sub-Saharan Africa, it is important for military practitioners to understand that the mechanisms that led to the creation of successful counterinsurgent militias in Afghanistan can also be applied in African contexts. The key is for US forces to have a sufficiently deep understanding of their operational environments, including the contours of local patronage networks and the vested interests of indigenous power brokers.

However, policy makers also need to be cautious about facilitating militia formation toward the end of reinforcing state power. Successful recruitment strategies are inherently local in nature, and recruiters ultimately admit new members into militias on the basis of their likely obedience to local leaders, not national-level politicians. The distinctly local criteria and mechanisms that govern the induction of trustworthy recruits will tend to lead to the creation of civil militias that do a good job of serving local populations, but that will ultimately prove difficult for national-level politicians to control (Staniland 2015). National-level politicians run the distinct risk of creating local militias and empowering local strongmen who may turn against them.

Footnotes

Acknowledgments

Mohamed Tarawalley, Nyamakoro Sesay, Sahr Buffa, S. R. Fillie-Faboe, and Mama Munda Fortune, for sharing their insight. Bakarr Kororma, David Yarjah, Mohamed Gbondo, and Henry Macauley for assisting in my field research. David Waldner, Joe Miller, Robert Fatton, Todd Sechser, Denise Walsh, Arthur Abraham, Vera Mironova, two anonymous reviewers, and especially Deborah McDowell, Lawrie Balfour, and the fellows at the Carter G. Woodson Institute for African-American and African Studies at the University of Virginia for their invaluable feedback on this article.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Abdullah

Ibrahim

Muana

Patrick K.

. 1998. “The Revolutionary United Front: A Revolt of the Lumpenproletariat.” In African Guerrillas, edited by Clapham

Christopher S.

, 172–94. Oxford, UK: James Currey.

Abraham

Arthur

. 1976. Topics in Sierra Leone History: A Counter-colonial Interpretation. Freetown, Sierra Leone: Leone Publishers.

Acemoglu

Daron

Reed

Tristan

Robinson

James A.

. 2014. “Chiefs: Economic Development and Elite Control of Civil Society in Sierra Leone.” Journal of Political Economy 122 (2).

Alie

Joe

. 2005. “The Kamajor Militia in Sierra Leone: Liberators or Nihilists?” In Civil Militia: Africa’s Intractable Security Menace? Edited by Francis

David J.

, 51–70. Burlington, VT: Ashgate.

Arthy

Simon

. 2004. “Third Party Contribution to Conflict Transformation: Reflections from Sierra Leone.” Unpublished Dissertation for Masters in Anthropology of Conflict, Violence and Conciliation, University of Sussex, Brighton, UK.

Barr

Abigain

Ensminger

Jean

Johnson

Jeffrey

. 2009. “Social Networks and Trust in Cross-cultural Economic Experiments.” In Whom Can We Trust? How Groups, Networks and Institutions Make Trust Possible, edited by Cook

Karen S.

Levi

Margaret

Hardin

Russell

, 65–90. New York: Russell Sage Foundation.

Cohen

Dara K.

Nordas

Ragnhild

. 2015. “Do States Delegate Shameful Violence to Militias? Patterns of Sexual Violence in Recent Armed Conflicts.” Journal of Conflict Resolution 59 (5): 877–898.

Collier

Paul

Hoeffler

Anke

. 2004. “Greed and Grievance in Civil War.” Oxford Economic Papers, New Series 56:563–95.

De Mesquita

Bueno Ethan

. 2005. “The Quality of Terror.” American Journal of Political Science 49 (3): 515–30.

10.

Eckstein

Harry

. 1975. “Case Studies and Theory in Political Science.” In Handbook of Political Science, Vol. 7, edited by Greenstein

Fred

Polsby

Nelson W.

, 79–138. Reading, MA: Addison Wesley.

11.

Fanthorpe

2001. “Neither Citizen Nor Subject? ‘Lumpen’ Agency and the Legacy of Native Administration in Sierra Leone.” African Affairs 100:363–86.

12.

Ferme

Mariane C.

2001. The Underneath of Things: Violence, History, and the Everyday in Sierra Leone. Berkeley: University of California Press.

13.

Fithen

David Caspar

. 1999. “Diamonds and War in Sierra Leone: Cultural Strategies for Commercial Adaptation to Endemic Low-intensity Conflict.” PhD thesis, Department of Anthropology, University College London, London, UK.

14.

Gambetta

Diego

Hamill

Heather

. 2005. Streetwise: How Taxi Drivers Establish Customer’s Trustworthiness. New York: Russell Sage Foundation.

15.

Gberie

Lansana

. 2005. A Dirty War in West Africa: The RUF and the Destruction of Sierra Leone. Bloomington: Indiana University Press.

16.

Greenwald

Bruce C.

1986. “Adverse Selection in the Labour Market.” The Review of Economic Studies 53:325–47.

17.

Hegghammer

Thomas

. 2013. “The Recruiter’s Dilemma: Signalling and Rebel Recruitment Tactics.” Journal of Peace Research 50 (1): 3–16.

18.

Hoffman

Danny

. 2011. The War Machines: Young Men and Violence in Sierra Leone and Liberia. Durham, NC: Duke University Press.

19.

Hulslander

Robert

Spivey

Jake

. 2012. “Village Stability Operations and Afghan Local Police.” Prism 3 (3): 125–38.

20.

Humphreys

Macartan

Weinstein

Jeremy M.

. 2008. “Who Fights? The Determinants of Participation in Civil War.” American Journal of Political Science 52:436–55.

21.

Jentzsch

Corinna

Kalyvas

Stathis

Schubiger

Livia Isabella

. 2015. “Militias in Civil Wars.” Journal of Conflict Resolution 59 (5): 755–769.

22.

Johnston

Patrick

. 2008. “The Geography of Insurgent Organization and its Consequences for Civil Wars: Evidence from Liberia and Sierra Leone.” Security Studies 17:107–37.

23.

Keen

David

. 2005. Conflict and Collusion in Sierra Leone. New York: Palgrave.

24.

Leach

Melissa

. 1994. Rainforest Relations: Gender and Resource Use among the Mende of Gola, Sierra Leone. Edinburgh, UK: Edinburgh University Press for the International African Institute.

25.

Lijphart

Arend

. 1971. “Comparative Politics and the Comparative Method.” The American Political Science Review 65 (3): 682–93.

26.

Lujala

Paivi

Gleditsch

Nils Petter

Gilmore

Elisabeth

. 2005. “A Diamond Curse?: Civil Waranda Lootable Resource.” Journal of Conflict Resolution 49:538–62.

27.

Meffert

Richard P.

Bolduc

Donald C.

. 2011. Village Stability Operations and Afghan Local Police: Bottom-up Counterinsurgency. Bagram Airbase, Afghanistan: Headquarters, Combined Joint Special Operations Task Force.

28.

Miller

Gary J.

1992. Managerial Dilemmas: The Political Economy of Hierarchy. New York: Cambridge University Press.

29.

Montgomery

James D.

1991. “Social Networks and Labor-market Outcomes: Toward an Economic Analysis.” The American Economic Review 81 (5): 1408–18.

30.

Oppenheim

Ben

Steele

Abbey

Vargas

Juan F.

Weintraub

Michael

. 2015. “True Believers, Deserters, and Traitors: Who Leaves Insurgent Groups and Why.” Journal of Conflict Resolution 59 (5): 794–823.

31.

Peters

Krijn

. 2011. War and the Crisis of Youth in Sierra Leone. New York: Cambridge University Press.

32.

Petersen

Roger D.

2001. Resistance and Rebellion: Lessons from Eastern Europe. New York: Cambridge University Press.

33.

Rees

Albert

. 1996. “Information Networks in Labor Markets.” The American Economic Review 56 (1/2): 559–66.

34.

Spence

Michael

. 1974. Market Signaling: Informational Transfer in Hiring and Related Screening Processes. Cambridge, MA: Harvard University Press.

35.

Staniland

Paul

. 2012. “Organizing Insurgency: Networks, Resources, and Rebellion in South Asia.” International Security 37:142–77.

36.

Staniland

Paul

. 2014. Networks of Rebellion: Explaining Insurgent Cohesion and Collapse. New York: Cornell University Press.

37.

Staniland

Paul

. 2015. “Militias, Ideology, and the State.” Journal of Conflict Resolution 59 (5): 770–793.

38.

The Prosecutor v. Moinina Fofana and Allieu Kondewa . SCSL-04-14-T-790, Annex A, Request for leave to supplement the Fofana Sentencing Brief, 21997-22005.

39.

Weinstein

Jeremy M.

2005. “Resources and the Information Problemin Rebel Recruitment.” Journal of Conflict Resolution 49:598–624.

40.

Weinstein

Jeremy M.

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

41.

Wlodarczyk

Nathalie

. 2009. Magic and Warfare: Appearance and Reality in Contemporary African Conflict and Beyond. New York: Palgrave Macmillan.

42.

Wood

Elisabeth Jean

. 2003. Insurgent Collective Action and Civil War in El Salvador. Cambridge, UK: Cambridge University Press.