TY - JOUR
AU - Doctor, Austin, C
AB - Abstract Why do rebel organizations splinter into competing factions during civil war? To explain this outcome, I leverage variation in rebel leadership. I argue that rebel leaders draw on their pre-war experiences—i.e., their military and political experiences—to manage their organizations during conflict. These experiences bear unique patterns of rebel management and, thus, corresponding risks of fragmentation. Empirical evidence comes from a two-stage research design and original data featuring over 200 rebel leaders from 1989 to 2014. In the first stage, I estimate the probability of group fragmentation with a series of logistic regression models. In the second stage, I use Cox proportional-hazards models to estimate leadership effects on the rate of group fragmentation. Results indicate that variation in rebel leadership corresponds with unique risks of fragmentation. In particular, the results suggest that leaders with real military experience are best equipped to maintain group cohesion. This study offers insight into the processes by which rebel groups splinter into armed factions. In addition, it makes an important contribution to the broader discussion on the roles of structure and agency in shaping the dynamics of civil war. rebel leaders, rebel groups, rebel fragmentation, civil war Introduction Why do rebel organizations fragment into competing factions? Existing studies suggest that rebel fragmentation influences a number of important conflict processes, including conflict duration (Findley and Rudloff 2012), the intensity of conflict violence (Cunningham, Bakke, and Seymour 2012), and the durability of post-conflict peace (Nilsson 2010; Rudloff and Findley 2016). As such, understanding why rebel organizations splinter carries important implications for both theory and policy. The fragmentation literature emphasizes external pressures that compromise the bond between rebel leaders and their followers, such as peace negotiations, external sponsors, social networks, and relative military capabilities. The empirical record, however, suggests that experiences of rebel fragmentation vary widely within these contexts. To address this issue, I theorize that rebel fragmentation reflects a strongly held belief among group members—especially senior-level subcommanders—that the group leader is unlikely to deliver on their political, military, or material commitments. A motion of no confidence. This belief and, subsequently, a rebel group's risk of fragmentation, are especially informed by a leader's demonstrated capacity for organizational management. To substantiate this theory, I build on the premise that rebel leaders are not created equal; to the contrary, they vary considerably in their capacities for leadership. Rather than assume that rebel leaders acting within similar contexts will behave similarly, I describe the relationship between a leader's attributes—their pre-war military and political experiences—and the occurrence of fragmentation. In doing so, I make an important contribution to existing studies, which have generally been less attentive to the organization-level factors that prompt rebel subcommanders to form their own armed factions. More broadly, this study offers important perspective and empirical evidence to the long-standing discussion on the roles of structure and agency in armed conflict. Empirical evidence comes from an original dataset on over 200 rebel leaders and a two-stage research design in which I model the risk of rebel fragmentation as a function of variation in rebel leadership. In a series of binary logistic and Cox regression models, I find evidence that rebel groups’ susceptibilities to fragmentation are influenced by the nature of their leaders’ pre-war military experiences, where prior combat experience and prior military service are negatively associated with the probability and rate of fragmentation, respectively. Conversely, rebel leaders who previously held political office or a ruling position in a political party tend to struggle to maintain the loyalty of their subcommanders during war. More broadly, this study offers evidence that leaders wield independent agency in the anarchic environment of civil war. In particular, I demonstrate that rebel leaders draw on their unique repertoires of pre-war experiences to maintain organizational cohesion. In addition to offering insight into the process of rebel fragmentation, this study suggests that greater effort should be made to incorporate leaders into both theoretical and empirical approaches to the dynamics of armed intrastate conflict. What is Rebel Fragmentation? In this study, I use “fragmentation” to refer to an event in which a segment of a rebel organization formally and collectively exits that rebel organization and establishes a new, independent rebel organization (Lidow 2016; Tamm 2016).1 This same event is also described as “splintering” in various studies, and I use the terms interchangeably. In the existing literature, the concept of “fragmentation” is used to refer to various processes at different levels of analysis. In some cases, it is used to describe the nature of a rebel movement, the proliferation of conflict actors, or decreases in internal discipline. This conceptual stretching is likely to cause some confusion in future studies and, as a result, stands to cloud the advice scholars offer policymakers on this important issue. To clarify, I treat rebel fragmentation as an organization-level event, rather than a movement-level characteristic, that varies both within and between conflict spaces and across time. With this in mind, it is equally important to clarify which events are not considered cases of fragmentation in the context of this study. Episodes in which a rebel subcommander or fighter deserts, defects to another warring party, or stages an internal coup are not identical to events in which they create a competing armed faction. These processes reflect unique decisions, different solutions to similar circumstances, altogether. Indeed, the defection or desertion of group members may sometimes constitute a change of tack following an unsuccessful fragmentation attempt. While research on both the common and different circumstances that lead to these related outcomes would greatly benefit the literature, this issue falls outside the scope of the current analysis. As such, for the purposes of this study, I choose to conceptualize fragmentation narrowly, isolating it from other distinct, albeit related, outcomes of interest. The event of rebel fragmentation is ubiquitous in intrastate conflict. Indeed, according to Lidow (2016), at least one-third of rebel groups fragment at least once. An important body of research investigates the consequences of rebel fragmentation for conflict dynamics and outcomes. These studies indicate that the splintering of rebel organizations can have substantial influence on conflict intensity (Cunningham, Bakke, and Seymour 2012; Staniland 2014), conflict duration (Perkoski 2015; Mahoney 2017), and outcome (Kenny 2010; Driscoll 2012; Findley and Rudloff 2012).2 As such, it is important to ask: Why does fragmentation occur in the first place? Existing Explanations of Fragmentation The intra-organizational dynamics that influence group cohesion present a demanding puzzle (Kalyvas 2006; Blattman and Miguel 2010). Existing studies on the origins of rebel fragmentation have identified a number of factors associated with the occurrence of this event. First, Woldemariam (2016) shows that military performance is related to group fragmentation. Specifically, insurgent organizations are more prone to fragmentation when facing sustained military losses or loss of territorial control.3 Woldemariam interprets this finding as a function of breakdown in cooperation between rebel leaders and organizational elites, who reconsider the probability of success following repeated failures. Based on a case study of Jebha in the Ethiopian Civil War, Woldemariam (2016: 149) also finds that battlefield gains can prompt fragmentation, as territorial gains alter the incentives for continued cooperation. These results suggest the need for further investigation into the factors that may condition the effect of military performance on rebel cohesion. Second, a number of studies indicate that a rebel group's relationship with external sponsors shapes their susceptibility to fragmentation. Lidow (2016) argues that external patrons with interests that align with the rebel leader's own interests will make efforts to strengthen the leader's control in the group, thereby increasing the leader's ability to prevent fragmentation. Conversely, patrons with preferences that do not align with those of the rebel leader will keep the leader weak, contributing to a greater risk of fragmentation. Tamm (2016) similarly finds that the nature of a foreign sponsor's relationship with both the group leader and its senior leadership produces unique risks of fragmentation. Specifically, when sponsors work to maintain an imbalance of power in favor of the leader—through the distribution of resources—the group is less likely to fragment. When sponsors contribute to the balance of power between leaders and their subcommanders, however, effectively undermining leadership control, the rebel group is more likely to split into competing factions. These studies indicate that, while external relationships may be essential to the immediate sustainability of group operations, rebel leaders must weigh these relationships with their own capacities to maintain control over group operations. Third, Staniland (2014) expresses fragmentation as a function of the pre-war social bases from which rebel groups emerge. While leveraging a rather loose definition of fragmentation—to mean organizational fracturing in some instances and a lack of discipline in others—Staniland (2014) describes how group structures can change throughout a conflict as a result of society's underlying network of working ties, with identifiable implications for group cohesion. “Insurgents,” he argues, “go to war with the networks they have, for better or worse” (Staniland 2014, 20). This burgeoning body of research makes an important contribution, provoking questions about the limitations of individual agency in rebel cohesion and the role that pre-war factors play in shaping the conflict environment. Finally, rebel groups’ organizational structures may affect their vulnerability to factionalization. Indeed, “[t]he management of intense [fragmentation] is one of the more important considerations driving the structural choices of organizational leaders” (Sinno 2008, 93). Existing studies, however, produce disparate conclusions on the nature of the relationship between structure—especially the degree of centralization in the chain-of-command—and fragmentation. On the one hand, Shapiro (2013) argues that, in response to sustained pressure from counterinsurgent forces, militant groups may strategically decentralize in order to survive. This has the unintended consequence of limiting leadership control over operation units and increases the probability of splinter factions emerging. In a related study, Asal, Brown, and Dalton (2012) find that the organizational structures of ethnopolitical organizations affect their respective proclivity to splinter. In particular, organizations with more centralized leadership structures are more robust to risks of fragmentation. On the other hand, Day (2017) finds the decentralized leadership structure—featuring a number of geographically distant, semiautonomous cells—of Joseph Kony's rebel group, the Lord's Resistance Army, helped the organization remain cohesive in the face of significant military and territorial losses.4 Additionally, Sinno (2008, 93) argues that “less centralized organizations do not require high levels of coordination and can therefore manage factionalism through regional or functional segregation until they acquire the resources to eliminate it.” The list of plausible theoretical mechanisms connecting organizational structure to fragmentation is large and merits additional study. Overall, the existing literature explains fragmentation as a function of weak organizational control, which is compounded by external pressures on group cohesion. Fragmentation studies suggest that, all else equal, the latent threat of splintering is kept at bay when leaders sufficiently demonstrate their own capacity to achieve stated group objectives. When intra-organizational rivals can no longer be controlled or satisfied, then an internal faction is increasingly likely to break away and form a new group. These studies also raise important points for further investigation. Leaders play more than an anecdotal role in the theoretical frameworks described by existing studies, though they are generally treated as a unitary set of actors. Does variation in rebel leadership offer needed leverage over the complex issue of fragmentation? Rebel Leaders in Civil War Intrastate war is forged by people and tempered by context. A country does not go to war with itself ex nihilo; rather, intrastate war reflects a strategic sequence of decisions and actions made by people. Some people, inevitably, shape this process more than others. Recent studies suggest that rebel leaders—the individuals who mobilize and manage insurgent forces—play a significant role in influencing the behavior and viability of armed rebel groups in conflict settings. Like other human organizations, rebel groups rely on capable managers, leaders responsible for translating group objectives into action. In the words of an Al-Qaeda operations manual (Al-Muqrin 2008, 109), “[l]eadership is what unifies, shapes, and executes. Leaders unify in the sense they integrate under one umbrella all the cadres, efforts, capabilities, and experience that the movement possesses.” A theoretical focus on these individuals offers unique and needed insight into patterns of rebel group management. In the spirit of Horowitz, Stam, and Ellis (2015), I argue that a rebel leader's attributes and pre-war experiences shape the choices they make in war.5Williams (2016, 9) puts it colorfully: “Like wars, recipes don't make themselves—there must be cooks, and it is important to know whether one is dealing with a novice or a master chef.” Building on this observation, I provide a framework that both policymakers and scholars can use to assess the aptitude of rebel leaders in conflict zones across time and space. A leader-based approach is well-suited to provide empirical leverage and theoretical clarity over rebel fragmentation. Generally, recent approaches to rebel leadership and the cohesion or behavior of their organizations share a common theme: leaders mostly serve as faceless conduits, channeling the dominant influence of environmental features.6 For instance, Weinstein (2007, 7) argues that “[d]ifferences in how rebel groups employ violence are a consequence of initial conditions that leaders confront.” By treating rebel leaders as interchangeable actors, structural theories fall short of leveraging them as sources of explanatory power. In leader-based explanations, Ahlquist and Levi (2011, 3) point out, “[t]he trick is to clarify the relationship between context and the particular attributes and tasks of leaders.” This study investigates the microfoundations of rebel fragmentation and offers a novel explanation of how rebel leaders draw on their personal experiences and characteristics to wield independent agency on group cohesion. A Motion of No Confidence In this section, I outline a leader-based, organizational theory of rebel fragmentation. The splintering of rebel groups in recent and ongoing insurgencies suggests that fragmentation reflects a belief among senior subcommanders that the group leader is incapable of delivering on their political, military, or material commitments. A motion of no confidence. To be clear, my theory does not attempt to explain a rebel leader's ability to win, overall, but to explain their capacity to maintain a unified campaign throughout the conflict. To account for the intra-group dynamics that lead to rebel fragmentation, I highlight the interaction between two sets of rebel group members. The first set is rebel leaders. Similar to Prorok (2016), I use the term “rebel leader” to refer to the individual that is most responsible for exercising power in a rebel organization, not just the operations commander or organizational figurehead. If rebel leaders are responsible for the effective management and cohesion of their organizations, organizational fragmentation must represent some perceived failure in their leadership performance. For example, Bacon and Arsenault (2017, 2) argue that Osama bin Laden's “greatest strength as the leader of Al Qaeda was the masterful way in which he cultivated and reinforced internal and external unity, successfully managing difficult personalities within his group and within allied groups.” Conceptualized thus, the origins of organizational fragmentation can be investigated according to a leader's characteristics and capacity for leadership. The second set of actors central to the process of fragmentation is rebel subcommanders. I use the term “subcommander” to refer to any person at an intermediate level of authority in the rebel organization, especially those at a level just under the rebel leader. In effect, these are the individuals to whom rebel leaders delegate, in part, the management of armed battalions, political and social programs, diplomatic missions, logistics, and intelligence operations (Shapiro 2013; Lidow 2016). Armed rebellion is a massive undertaking; by default, most rebel leaders are compelled to delegate responsibility to carefully selected lieutenants—e.g., National Union for the Total Independence of Angola (UNITA) general Geraldo Muengo Ukwachitembo “Kamorteiro” or Islamic State (ISIS) commander Abu Abdulrahman al-Bilawi. In practice, the vast majority of fragmentation episodes are instigated by one or multiple of these subcommanders. For example, in 1994, a number of senior commanders of the National Patriotic Front of Liberia (NPFL) rebel group, including Minister of Defense Tom Woewiyu, Minister of Justice Lavala Supuwood, and Minister of Internal Affairs Samuel Dokie, led a splinter faction called the Central Revolutionary Council (NPFL-CRC). These three commanders—all of whom came from positions in the NPFL's civilian administration body—also convinced military leader Nixon Gaye, who led the NPFL Marines Division, to join them (Lidow 2016, 123). The faction was short-lived; NPFL leader Charles Taylor captured and executed Gaye a year later. After recruitment, subcommanders continue to update their beliefs about the competency of their leaders. This belief—and, subsequently, a rebel group's risk of fragmentation—is especially informed by the strategic and tactical decisions made by group leaders (McLauchlin and Pearlman 2012; Rudloff and Findley 2016). It is ranking subcommanders who make the critical decision to leave, form a competing faction, and covertly recruit fighters to their new organization. In this study, I make an important contribution to this story. I argue that this choice to fragment reflects a lack of confidence in the leader's capacity to match strategies and tactics to group objectives and, ultimately, to achieve victory. By treating subcommanders as rational decisionmakers, we gain theoretical clarity over the strategic decision to leave and form an alternative rebellion. The choice to split from an active rebel group is clouded by high levels of uncertainty and risk. Accordingly, it must represent a strongly held belief that the expected utility of creating an alternative rebel organization outweighs the expected utility of staying.7 A thorough explanation of fragmentation, then, will need to specify both patterns of group leadership and how these patterns are likely to affect the beliefs of senior commanders in their organizations (Nagel and Doctor 2019). This study argues that this calculation is informed by subcommanders’ perceptions about the rebel leader's ability to deliver on their promises and achieve group objectives.8 A number of studies (Lidow 2016; Fjelde and Nilsson 2018; Woldemariam 2018) and recent conflict cases—e.g., the Sudan Liberation Movement/Army-Minni Minnawi (SLM/A-MM) in South Sudan, the Oromo Liberation Front (OLF) in Ethiopia, or the Karen National Union (KNU) in Myanmar—suggest the existence of some common flash points of fragmentation, such as the onset of peace talks or significant shifts in battlefield outcomes. These also can be understood within my theoretical framework, where subcommanders take advantage of the information generated by these events to update their beliefs about their leader's ability to select effective and desirable battlefield and negotiation strategies. In this study, I focus on the broader relationship between leader type and fragmentation, generalizing about patterns of leadership across the ebb and flow of conflict events. Rebel Leadership and Fragmentation To substantiate this theory, I build on the premise that rebel leaders are not created equal. Leaders vary considerably in their capacities for leadership. And some experiences prepare individuals to more effectively lead a rebel organization. Accordingly, rather than treat these individuals as a monolith, I investigate leaders’ patterns of management according to their pre-war experiences. I argue that different leader types, operating under similar constraints and conditions, manage their organizations differently. Corresponding with these unique patterns of leadership, rebel groups face unique risks of fragmentation. Why do leaders’ past experiences matter? I draw on the work by Horowitz, Stam, and Ellis (2015, 10) on state leaders, which argues that “[b]ackground experiences matter in part because they form a mental Rolodex that both citizens and leaders turn to when making strategic decisions…Background experiences represent a pool of lessons learned, shaping a leader's judgement about which strategies are more or less likely to succeed or fail.” This “mental Rolodex” shapes how leaders form objectives, consider risk, and evaluate the strength of particular strategies. I make the case that we can generalize across such characteristics to form expectations of how rebel leaders will lead during times of intrastate war.9 I leverage two rebel leader attributes—prior military experience and prior political experience—to explain why some groups split into competing factions while others remain cohesive. Military Experience While often presented in news media as military figures, brandishing Kalashnikovs and fatigues, many rebel leaders do not have meaningful military backgrounds. Building on existing studies on military performance and fragmentation, I propose at least two reasons for my belief that a leader's military experience shapes rebel organizations’ proclivities to fragment. First, rebel groups with leaders who have prior military experience should be better able to convince their followers of their respective ability to achieve military objectives—and, indeed, will emphasize military means to achieve group objectives (Willingham 2017). Sustained experiences of military defeat cause members of a rebel campaign to reconsider their immediate and future prospects of achieving their objectives (Christia 2012). Woldemariam (2016, 3–4) puts it succinctly: In settings where a rebel organization is losing territory, often through a set of major shocks, the incentives to cooperate are reduced, as battlefield losses suggest that the collective enterprise that is organized rebellion no longer guarantees the survival of the organization's constituent units. Put differently, losing territory prompts an organization's constituent units to question the cooperative bargain that is at the heart of the rebel organization. All things equal, fragmentation is more likely in such contexts. Building on this premise, a leader's personal military experience should enable them to better execute a critical responsibility of insurgent management: the war campaign. Accordingly, all else equal, these leaders will enjoy higher levels of confidence in their abilities to lead, muting subcommanders’ desires to create a competing splinter faction. Second, leaders with a history of military activity draw on these experiences to build organizations with higher levels of discipline and centralization of command, both of which tend to be associated with greater group cohesion (Asal, Brown, and Dalton 2012; Shapiro 2013). Leaders with military experience are inherently exposed to and socialized into the strict norms of military command. Accordingly, as these leaders mobilize, structure, and manage their organizations, they will aim to replicate these norms in their own rebel campaigns. In a study on armed jihadist organizations in Africa, for example, Hazen (2013) argues that many militant groups were able to maintain higher levels of cohesion during seasons of sustained counterinsurgent activity, in part because of their organizational structure. Indeed, these organizations—led by individuals with years of military experience under their belts—often adopt more “traditional military-like” structures (Hazen 2013, 589). “The quality of the command hierarchy increases in its importance,” Hazen finds, “as it plays a greater role in the organizations’ cohesion” (Hazen 2013, 601). Based on these mechanisms, I propose the following testable hypothesis. H 1: All else equal, rebel groups led by leaders with prior military experience will be less likely to fragment than those led by leaders without prior military experience. Political Experience While there is a sizable literature on how rebel groups transform into and perform as political parties following civil war, we know far less about the fortunes of rebel groups whose leaders emerge from positions of formal political authority or political organizations. Speaking to this issue, I explain how rebel leaders with previous experience in national politics face unique risks of fragmentation. Insurgent campaigns, as a necessity, include some mix of political and military operations. While leaders with prior political experience may enjoy higher levels of popular support or generally be more effective at building robust political structures, these leaders should be more vulnerable to the threat of fragmentation. Although political officials turned insurgent leaders are likely to be skilled in organizational management, these qualities are insufficient for waging a viable militant campaign. In addition to mobilizing fighters and attracting outside supporters, rebel leaders are also responsible for acquiring arms, training recruits, and, ultimately, winning battles. “Leaders need to design organizational structures that educate and train soldiers, procure and maintain equipment…fight on the battlefield, and conduct numerous other tasks that militaries need to accomplish” (Jones 2017, 89). Rogue politicians, in particular, may struggle to carry out the tasks required of the commander-in-chief and, playing to their own strengths, will tend to emphasize political operations over military means. Given what existing studies indicate about the importance of military efficacy to group cohesion, when leaders fail to acquire and allocate sufficient resources for military operations and subsequently fail to succeed on the battlefield, former politicians are especially likely to lose the confidence of their subcommanders. For example, in 1989, Liberian rebel leader Charles Taylor, a former finance minister in President Samuel Doe's government, launched an insurgency at the helm of the rebel group known as the NPFL. In a public British Broadcasting Corporation (BBC) broadcast, Taylor stated that the NPFL's primary objective was “to get that boy Doe off the backs of the Liberian people” (Ellis 1999, 75).10 Relative to its lofty ambitions, the NPFL had humble beginnings; the group's first military assault was carried out with only a handful of shotguns (Waugh 2011). Though an overall success, this initial offensive experienced a number of setbacks and confidence in Taylor's leadership waned.11 Within the first two months of Taylor's leadership, Prince Yormie Johnson—an NPFL subcommander and former lieutenant in the Liberian National Guard—broke away from the NPFL to lead a competing faction. Citing frustration with the absence of infrastructure to carry out military operations and lack of trust in Taylor's ability to make good on his extravagant promises, Johnson left with thirty-five soldiers and formed a new group, the Independent National Patriotic Front of Liberia (INPFL). The case of the NPFL and INPFL suggests that, when rebel leaders come from a political background, they may struggle to signal credibly their abilities to manage military-forward insurgent campaigns. Based on this underlying mechanism, I propose the following testable hypothesis: H 2: All else equal, rebel groups led by leaders with prior political experience will be more likely to fragment than those led by leaders without prior political experience. Data and Research Design To test my proposed hypotheses, I use an original collection of data, the Rebel Leaders in Civil War Dataset (RLCW).12 RLCW identifies and characterizes 206 rebel leaders in 158 rebel groups across sixty-five conflicts from 1989 to 2014. Observations are recorded at the rebel-year unit of analysis. The sample descriptive statistics are presented in Table 1. Table 1. Descriptive statistics Variable . N . Mean . St. dev. . Min. . Max. . Fragmentation 1,411 0.045 0.208 0 1 Military training 1,411 0.528 0.499 0 1 Military combat 1,411 0.592 0.492 0 1 Military service 1,411 0.198 0.399 0 1 Political office 1,411 0.122 0.327 0 1 Political party 1,411 0.104 0.306 0 1 Tenure 1,411 5.989 5.091 1 26 Founder 1,411 0.556 0.497 0 1 Splinter group 1,411 0.196 0.397 0 1 State sponsor 1,411 0.468 0.499 0 1 Rebel one-sided violence 1,411 124.851 1084.055 0 30,110 Foreign military intervention 1,411 0.332 0.471 0 1 Conflict intensity 1,411 568.274 1672.729 0 32,937 Multiparty conflict 1,411 0.870 0.336 0 1 Democracy 1,411 0.518 0.500 0 1 Variable . N . Mean . St. dev. . Min. . Max. . Fragmentation 1,411 0.045 0.208 0 1 Military training 1,411 0.528 0.499 0 1 Military combat 1,411 0.592 0.492 0 1 Military service 1,411 0.198 0.399 0 1 Political office 1,411 0.122 0.327 0 1 Political party 1,411 0.104 0.306 0 1 Tenure 1,411 5.989 5.091 1 26 Founder 1,411 0.556 0.497 0 1 Splinter group 1,411 0.196 0.397 0 1 State sponsor 1,411 0.468 0.499 0 1 Rebel one-sided violence 1,411 124.851 1084.055 0 30,110 Foreign military intervention 1,411 0.332 0.471 0 1 Conflict intensity 1,411 568.274 1672.729 0 32,937 Multiparty conflict 1,411 0.870 0.336 0 1 Democracy 1,411 0.518 0.500 0 1 Open in new tab Table 1. Descriptive statistics Variable . N . Mean . St. dev. . Min. . Max. . Fragmentation 1,411 0.045 0.208 0 1 Military training 1,411 0.528 0.499 0 1 Military combat 1,411 0.592 0.492 0 1 Military service 1,411 0.198 0.399 0 1 Political office 1,411 0.122 0.327 0 1 Political party 1,411 0.104 0.306 0 1 Tenure 1,411 5.989 5.091 1 26 Founder 1,411 0.556 0.497 0 1 Splinter group 1,411 0.196 0.397 0 1 State sponsor 1,411 0.468 0.499 0 1 Rebel one-sided violence 1,411 124.851 1084.055 0 30,110 Foreign military intervention 1,411 0.332 0.471 0 1 Conflict intensity 1,411 568.274 1672.729 0 32,937 Multiparty conflict 1,411 0.870 0.336 0 1 Democracy 1,411 0.518 0.500 0 1 Variable . N . Mean . St. dev. . Min. . Max. . Fragmentation 1,411 0.045 0.208 0 1 Military training 1,411 0.528 0.499 0 1 Military combat 1,411 0.592 0.492 0 1 Military service 1,411 0.198 0.399 0 1 Political office 1,411 0.122 0.327 0 1 Political party 1,411 0.104 0.306 0 1 Tenure 1,411 5.989 5.091 1 26 Founder 1,411 0.556 0.497 0 1 Splinter group 1,411 0.196 0.397 0 1 State sponsor 1,411 0.468 0.499 0 1 Rebel one-sided violence 1,411 124.851 1084.055 0 30,110 Foreign military intervention 1,411 0.332 0.471 0 1 Conflict intensity 1,411 568.274 1672.729 0 32,937 Multiparty conflict 1,411 0.870 0.336 0 1 Democracy 1,411 0.518 0.500 0 1 Open in new tab Dependent Variable The dependent variable, rebel fragmentation, captures instances in which a segment of a rebel organization formally and collectively exits that rebel organization and establishes a new, independent rebel organization. In RLCW, fragmentation is issued a binary measure, which is coded “1” if a rebel group splits into competing factions during a given rebel year and “0” otherwise. Fragmentation is a relatively rare event, given the unit of analysis, and is recorded as occurring in only sixty-four of the 1,411 rebel-year observations. Viewed from another perspective, 33 percent of the sampled rebel organizations fragment at least once—observed in thirty-five of the sixty-five conflicts featured in RLCW. The distribution of fragmentation in RLCW generally matches those found in similar datasets. Rudloff and Findley (2016) find that fragmentation has occurred in at least 45 percent of civil wars active since 1989. Likewise, of the 171 sub-Saharan African rebel groups sampled in his data, Michael Woldemariam (2018) finds that 32 percent split. Given the binary structure of the dependent variable, I test the proposed hypotheses in a series of logistic regression models. In all models, I include fixed effects for each country and year to correct for any omitted factors at these levels that may influence the process by which leaders directly affect group cohesion. Independent Variables The main contribution of the RLCW is its longitudinal measures of rebel leadership. In building the data, similar to Prorok (2016), RLCW coders identified the individual that is most responsible for exercising power in a rebel organization, not just the operations commander or organizational figurehead. In some rebel groups, operations are distributed between specialized military and political wings. In practice, however, the operations of each is typically overseen by a common leader, though deputies may be delegated to manage each branch. Accordingly, in such cases, to identify and code the group leader, we simply looked for evidence of the highest-ranking decisionmaker in the group. To code the set of leader-level variables, RLCW coders referenced a broad collection of source materials using a set of systematic coding guidelines, included in the supplementary online appendix. Data on rebel leaders are sourced from hundreds of government reports, news publications, academic case studies, historical accounts, and conflict expert reports. Data taken from secondary sources were corroborated by rebel group publications, rebel leader interviews, and other primary sources, when available. The main independent variables in this study—prior military experience and prior political experience—capture important sources of variation in rebel leadership. I evaluate the effects of each of these characteristic-types through multiple indicators of each type. I test three forms of prior military experience: formal military training, direct combat experience, and prior service in a state military. For example, prior to the onset of the Angolan Civil War (1975–2002), UNITA rebel leader Jonas Savimbi received several months of military training at the Nanking Military Academy in China and gained combat experience during the Angolan War for Independence (1961–1974). Prior to the civil war, Savimbi did not serve in a state military. These three indicators are not mutually exclusive but serve to capture various forms of relevant military activity. Plausibly, one may speak more loudly than another in shaping patterns of fragmentation. As such, rather than mask these sources of information in a single binary aggregate measure, I test each directly for its effects on fragmentation. Altogether, of the 206 leaders in the sample, ninety-four are coded with prior military training, 115 are coded with sustained combat experience, and forty-one are coded as previously being active in a formal state military.13 The influence of a leader's prior political experience is tested with two binary indicator variables: experience as head of a national political party and holding national political office. For example, Liberian rebel leader George Boley is coded as holding prior national political office. Before founding the Liberian Peace Council rebel group—a gargantuan misnomer—and entering the First Liberian Civil War (1989–1997), Boley served in appointments as the national Minister of Presidential Affairs and, later, Minister of Education under Liberian President Samuel Doe. Boley was never the head of a national political party. RLCW records thirty-nine rebel leaders with some form of prior political experience. Of these, twenty-two were formerly the head of a political party and twenty-eight are coded as having previously served in national political office. Interestingly, while over half of rebel leaders have a background in military operations, less than one-fourth emerge from high-ranking political positions. Figure 1 shows the distribution of all five leadership characteristics included in the analysis. Figure 1. Open in new tabDownload slide Distribution of leader characteristics Figure 1. Open in new tabDownload slide Distribution of leader characteristics Importantly, the military and political dimensions of a leader's pre-war experience are not mutually exclusive. That is, a rebel leader can have both military and political experience, or neither. In our sample of 206 rebel leaders, 14 percent of those with military experience also have some form of political experience. Relatedly, 39 percent of the rebel leaders with political experience also have military experience. Control Variables In addition to the main independent variables, I include a set of control variables that may shape subcommanders’ assessments of the costs and benefits of creating a splinter faction. At the leader level, I include two controls. First, to control for the likely case of time dependency in the model, I include both a linear and squared polynomial time term, which accounts for the length of a leader's tenure at time t. That is, leaders may experience fragmentation in a general temporal pattern. This control also serves to address potential confounding effects in which leaders with political or military experience may be more likely to sustain longer campaigns and, thus, have more opportunities to fragment. Second, it is quite plausible that individuals who emerge into positions of leadership from the ranks of a preexisting rebel group will be selected by other group members according to certain qualities, and those qualities may also be associated with unique risks of fragmentation. That is, it is unlikely that leaders who emerge in medias res are randomly distributed. To address this possible source of endogeneity, I include a control for leaders who form their respective rebel groups as founders, as recorded in RLCW, in contrast to those who receive leadership appointments from a preexisting group during an active conflict. At the group-year level, I control for whether or not a rebel organization is itself a splinter group, as recent studies suggest that these may enjoy higher levels of cohesion (Perkoski 2015; Otto 2018; Woldemariam 2018). To measure this, I use the UCDP Conflict Actor List, which includes a dummy variable that “indicates whether a non-state actor was formed by breaking away from an actor that has also been registered in UCDP data” (Version 2.1-2010, 12). I also include an indicator for rebel groups that enjoy the material support of a state sponsor, as existing studies suggest that these groups may have lower levels of cohesion (Lidow 2016; Tamm 2016). Finally, there is some evidence that subcommanders may form competing factions as a function of local civilian victimization. For example, Ansaru split from Boko Haram, in part, due to the treatment of “moderate” or noncompliant Muslims in northeast and central Nigeria. Similarly, the Salafist Group for Preaching and Combat (GSPC) split from the Armed Islamic Group (GIA) under the direction of Hassan Hattab in direct response to the anti-civilian attacks ordered by GIA leader Antar Zouabri. Moreover, Willingham and Doctor (2015) indicate that certain leader types may be more likely to target civilians than others. As such, I control for rebel one-sided violence, the number of civilian casualties inflicted by the leader's rebel group in a given year. Information for this variable comes from the UCDP One-Sided Violence Dataset (Eck and Hultman 2007; Allansson, Melander, and Themner 2017). At the conflict and country-year levels, I control for several factors that are likely to inform a subcommander's decision to form a new militant organization. First, as a number of existing studies on rebel fragmentation find that increased levels of combatant casualties in a conflict dyad affect a group's probability of splitting, I issue control for conflict intensity. To code this variable, I use a group-year measure of battle-related deaths from the UCDP Battle-Related Deaths Dataset (Pettersson and Eck 2018). Second, to account for the increased cost posed by crowded conflict spaces and higher numbers of potential opponents, I control for the presence of foreign military intervention forces with a dichotomous indicator variable, as recent work indicates that these may increase the probability of splintering (Olson Lounsbery 2016). These forces include the official presence of troops from national armies, United Nations (UN) peacekeeping forces, and troops from the European Union, African Union, or subregional blocs, such as the Economic Community of West African States (ECOWAS). Similarly, all models control for multiparty conflict, indicating conflict-years in which more than one rebel group—as recorded in the UCDP Armed Conflict Dataset—are active in a given conflict. I expect that, as the conflict space becomes more crowded with armed parties, the perceived cost of starting a new faction increases and, thus, decreases the probability of fragmentation, all else equal. Finally, in systems with a history of electoral institutions, subcommanders may see additional postwar benefits from leading their own movements. As such, I control for democracy with a binary indicator, in which I code all countries with a Polity IV score of 7 or above as “1,” and “0” otherwise (Marshall, Jaggers, and Gurr 2014). Logistic Regression Results Table 2 displays the estimated coefficients. These express the expected change in the log-odds of fragmentation for a unit-change in a given independent variable, all else equal. Positive coefficients, accordingly, indicate an increase in the log-odds of fragmentation in a given rebel-year. Negative coefficients indicate an estimated decrease in the log-odds. Model 1 presents the estimated correlations between the main leader dimensions and fragmentation. Model 2 builds on this specification to include all group-level controls. In Model 3 and Model 4, I test the military and political indicators separately, respectively, with a full set of controlling variables. In Model 5, I include all five leader indicators with the full set of controlling variables. Additional model specifications and robustness checks are provided in the supplementary online appendix. Table 2. Leadership and rebel fragmentation in armed conflict, 1989–2014 . Rebel fragmentation . . (1) . (2) . (3) . (4) . (5) . Military training 1.856** 1.899** 1.869** 2.012** (0.753) (0.758) (0.825) (0.822) Military combat −1.425** −1.400** −1.583** −1.522** (0.709) (0.701) (0.747) (0.747) Military service −0.757 −0.738 −0.566 −0.748 (0.567) (0.587) (0.597) (0.623) Political office −0.225 −0.561 −0.657 −0.662 (0.661) (0.700) (0.684) (0.743) Political party 1.122* 1.578** 1.345* 1.605** (0.642) (0.685) (0.712) (0.757) Founder 0.644* 0.622* 0.595 (0.384) (0.375) (0.385) Splinter group −0.535 −0.029 −0.252 −0.374 (0.456) (0.452) (0.459) (0.498) State sponsor 0.410 0.385 0.420 0.498 (0.376) (0.403) (0.399) (0.410) Rebel one-sided violence 0.140** 0.128* 0.135** 0.130** (0.062) (0.065) (0.066) (0.066) Foreign military intervention −0.233 −0.0004 −0.273 (0.494) (0.449) (0.496) Conflict intensity 0.0001 0.00004 0.0001 (0.0001) (0.0001) (0.0001) Multiparty conflict 0.235 0.357 0.569 (0.711) (0.702) (0.759) Democracy 0.648 0.514 0.475 (0.519) (0.532) (0.540) Constant −3.990*** −4.491*** −5.647*** −5.652*** −6.117*** (1.541) (1.559) (1.813) (1.749) (1.804) Observations 1,411 1,411 1,411 1,411 1,411 Log likelihood −218.340 −213.420 −211.250 −212.629 −209.162 Akaike inf. crit. 580.681 576.841 584.501 585.257 584.324 . Rebel fragmentation . . (1) . (2) . (3) . (4) . (5) . Military training 1.856** 1.899** 1.869** 2.012** (0.753) (0.758) (0.825) (0.822) Military combat −1.425** −1.400** −1.583** −1.522** (0.709) (0.701) (0.747) (0.747) Military service −0.757 −0.738 −0.566 −0.748 (0.567) (0.587) (0.597) (0.623) Political office −0.225 −0.561 −0.657 −0.662 (0.661) (0.700) (0.684) (0.743) Political party 1.122* 1.578** 1.345* 1.605** (0.642) (0.685) (0.712) (0.757) Founder 0.644* 0.622* 0.595 (0.384) (0.375) (0.385) Splinter group −0.535 −0.029 −0.252 −0.374 (0.456) (0.452) (0.459) (0.498) State sponsor 0.410 0.385 0.420 0.498 (0.376) (0.403) (0.399) (0.410) Rebel one-sided violence 0.140** 0.128* 0.135** 0.130** (0.062) (0.065) (0.066) (0.066) Foreign military intervention −0.233 −0.0004 −0.273 (0.494) (0.449) (0.496) Conflict intensity 0.0001 0.00004 0.0001 (0.0001) (0.0001) (0.0001) Multiparty conflict 0.235 0.357 0.569 (0.711) (0.702) (0.759) Democracy 0.648 0.514 0.475 (0.519) (0.532) (0.540) Constant −3.990*** −4.491*** −5.647*** −5.652*** −6.117*** (1.541) (1.559) (1.813) (1.749) (1.804) Observations 1,411 1,411 1,411 1,411 1,411 Log likelihood −218.340 −213.420 −211.250 −212.629 −209.162 Akaike inf. crit. 580.681 576.841 584.501 585.257 584.324 Notes: Two-tailed test; * p < 0.10, ** p < 0.05. Standard errors in parentheses. Fixed effects for country and year included in each model. Cubic, quadratic, and linear terms included for leader tenure. Rebel OSV is transformed by the natural log to account for skew. Open in new tab Table 2. Leadership and rebel fragmentation in armed conflict, 1989–2014 . Rebel fragmentation . . (1) . (2) . (3) . (4) . (5) . Military training 1.856** 1.899** 1.869** 2.012** (0.753) (0.758) (0.825) (0.822) Military combat −1.425** −1.400** −1.583** −1.522** (0.709) (0.701) (0.747) (0.747) Military service −0.757 −0.738 −0.566 −0.748 (0.567) (0.587) (0.597) (0.623) Political office −0.225 −0.561 −0.657 −0.662 (0.661) (0.700) (0.684) (0.743) Political party 1.122* 1.578** 1.345* 1.605** (0.642) (0.685) (0.712) (0.757) Founder 0.644* 0.622* 0.595 (0.384) (0.375) (0.385) Splinter group −0.535 −0.029 −0.252 −0.374 (0.456) (0.452) (0.459) (0.498) State sponsor 0.410 0.385 0.420 0.498 (0.376) (0.403) (0.399) (0.410) Rebel one-sided violence 0.140** 0.128* 0.135** 0.130** (0.062) (0.065) (0.066) (0.066) Foreign military intervention −0.233 −0.0004 −0.273 (0.494) (0.449) (0.496) Conflict intensity 0.0001 0.00004 0.0001 (0.0001) (0.0001) (0.0001) Multiparty conflict 0.235 0.357 0.569 (0.711) (0.702) (0.759) Democracy 0.648 0.514 0.475 (0.519) (0.532) (0.540) Constant −3.990*** −4.491*** −5.647*** −5.652*** −6.117*** (1.541) (1.559) (1.813) (1.749) (1.804) Observations 1,411 1,411 1,411 1,411 1,411 Log likelihood −218.340 −213.420 −211.250 −212.629 −209.162 Akaike inf. crit. 580.681 576.841 584.501 585.257 584.324 . Rebel fragmentation . . (1) . (2) . (3) . (4) . (5) . Military training 1.856** 1.899** 1.869** 2.012** (0.753) (0.758) (0.825) (0.822) Military combat −1.425** −1.400** −1.583** −1.522** (0.709) (0.701) (0.747) (0.747) Military service −0.757 −0.738 −0.566 −0.748 (0.567) (0.587) (0.597) (0.623) Political office −0.225 −0.561 −0.657 −0.662 (0.661) (0.700) (0.684) (0.743) Political party 1.122* 1.578** 1.345* 1.605** (0.642) (0.685) (0.712) (0.757) Founder 0.644* 0.622* 0.595 (0.384) (0.375) (0.385) Splinter group −0.535 −0.029 −0.252 −0.374 (0.456) (0.452) (0.459) (0.498) State sponsor 0.410 0.385 0.420 0.498 (0.376) (0.403) (0.399) (0.410) Rebel one-sided violence 0.140** 0.128* 0.135** 0.130** (0.062) (0.065) (0.066) (0.066) Foreign military intervention −0.233 −0.0004 −0.273 (0.494) (0.449) (0.496) Conflict intensity 0.0001 0.00004 0.0001 (0.0001) (0.0001) (0.0001) Multiparty conflict 0.235 0.357 0.569 (0.711) (0.702) (0.759) Democracy 0.648 0.514 0.475 (0.519) (0.532) (0.540) Constant −3.990*** −4.491*** −5.647*** −5.652*** −6.117*** (1.541) (1.559) (1.813) (1.749) (1.804) Observations 1,411 1,411 1,411 1,411 1,411 Log likelihood −218.340 −213.420 −211.250 −212.629 −209.162 Akaike inf. crit. 580.681 576.841 584.501 585.257 584.324 Notes: Two-tailed test; * p < 0.10, ** p < 0.05. Standard errors in parentheses. Fixed effects for country and year included in each model. Cubic, quadratic, and linear terms included for leader tenure. Rebel OSV is transformed by the natural log to account for skew. Open in new tab First, I find mixed support for Hypothesis 1—that military experience will mitigate rebel leaders’ risk of experiencing fragmentation. The results suggest that the nature of a leader's military experience matters greatly. All else equal, controlling for other forms of military experience, leaders with prior military training are actually more prone to experience group fragmentation. In Model 5, the odds of splintering increase for these leaders by an estimated 648 percent; this is consistent across all models. Conversely, under the same conditions, prior combat experience decreases the odds of fragmentation in a given year by 358 percent (Model 5). I find no evidence that prior service in a state military affects the overall likelihood that a group splits into competing factions. Together, the results emphasize the importance of real battlefield experience in preventing fragmentation. Indeed, though the positive effect of military training cuts against my stated hypothesis, it also serves to illuminate the difference between the effects of simulated versus real military experience. Rebel leaders who only have military training generally enter a conflict with the same (or less) military experience as their subcommanders. As such, these subcommanders are likely to be more confident in their own capacities to succeed as independent conflict actors. Second, the results of this analysis support Hypothesis 2: rebel leaders with prior experience as head of a political party are more likely to experience fragmentation than those with no experience as political party leaders, all else equal. For groups led by former party leaders, the odds of fragmentation increase by 398 percent (Model 5). I find no evidence, however, that leaders with prior experience in political office are more or less prone to splinter than any other rebel leader type. I have argued that former politicians: (1) may struggle to manage military and logistical operations; and (2) are more likely to engage in political negotiations, sometimes against the preferences of more hardline subcommanders. The results indicate that political party leaders perpetuate these dynamics more consistently than those who become rebel leaders from bureaucratic, legislative, or executive political positions. This dynamic merits further qualitative assessment of the networks from which these two rebel leader types recruit and mobilize support. Based on this analysis, certain distinctions in rebel leadership, all else equal, are associated with predictable differences in the probability of fragmentation.14 In a given year, rebel groups managed by leaders with prior combat experience have a 0.07 lower probability of fragmentation as compared to those led by individuals with no military experience, all else equal. In contrast, militants led by leaders with prior military training, but no combat experience, have an 0.08 greater probability of fragmentation. Finally, all else equal, rebels have a 0.10 greater probability of splitting into factions when led by the former head of a national political party. Given the relatively fine-grained unit of analysis and the generally small number of fragmentation cases in the sample, these discrete changes in the probability of fragmentation constitute meaningful differences in the observed cohesion of rebel organizations. In sum, leadership offers important leverage over the dynamics of group cohesion. The estimates corresponding to a number of controlling variables also point to some interesting dynamics. First, highly predatory groups are more likely to fragment than those who engage in lower levels of violence against civilians. A unit increase in the natural log of rebel-inflicted civilian deaths results in an estimated 16 percent increase in the odds of fragmentation. Second, leaders who created their rebel organization may be more susceptible to fragmentation than those who assume leadership after conflict begins, though these correlations are not statistically discernible at traditional levels in a two-tailed test. I investigate this issue further in the following section. Leadership Selection A valid concern is the issue of selection bias—in particular, that pre-existing rebel groups that are predisposed to fragment may attract or appoint leaders who themselves have characteristics associated with discernible levels of fragmentation. While Models 3–5 do control for founding leaders, I conduct two robustness checks to address the important issue of leadership selection dynamics. In the first, I directly measure the nature of a leader's entry or appointment to leadership with a nine-part categorical variable.15 Feasibly, not just the turnover of leadership, but also the process by which groups nominate or recognize new leadership may signal their latent proclivity to fragment in the first place. This follows from recent studies on the appointment and election of rebel leaders. For example, Cunningham and Sawyer (2019) argue that the process by which individuals are selected for positions of rebel leadership has lasting effects on the timing of peace negotiations. Ultimately, none of the tested leadership entry categories is associated with a discernible likelihood of fragmentation and the results of the main independent variables remain robust to this model specification. In a second test, I subset the sample of rebel leaders to only “founders,” or those who bring a new armed party into war. As these leaders are largely responsible for the mobilization, formation, and consolidation of these forces—as well as conflict initiation itself—their “type” is less likely to be endogenous to a conflict that they themselves are responsible for instigating. Moreover, recent studies have found that these actors may make certain decisions differently than those who claim a position of leadership during an ongoing conflict. In a pair of related studies, Prorok (2016, 2018) explains how rebel leaders’ respective degree of personal responsibility for conflict connects to observed differences in conflict outcomes. Prorok finds that rebel leaders who are responsible for initiating conflict—those in positions of authority at conflict onset—are less likely to settle peacefully (2016) and more likely to prolong conflict (2018) relative to those who step into positions of power after the conflict has begun. Again, the results presented in Table 2 are robust to this approach. Further details and the results of both tests are presented in the supplementary online appendix. The Rate of Rebel Fragmentation? Closely tied to the probability of rebel fragmentation is the issue of timing—at what rate do groups fragment? While the results from the logistic regression analysis show the probability that a group fragments in a given year as a function of its leadership, they tell us little about when fragmentation occurs. That is, the logistic regression models estimate leadership effects on the cumulative incidence of fragmentation. Moreover, the logistic models pick up cases in which the same group fragments multiple times. Indeed, of the rebel groups in RLCW that experienced fragmentation, 17 percent fragmented at least twice. To more clearly delineate the effects of leadership on a group's incidence rate of fragmentation, I employ a survival analysis. This approach investigates, based on the logic outlined in the theoretical framework above, whether subcommanders update their beliefs about the rebel leader at discernible rates, sooner or later in the conflict campaign. Increasingly, scholars have employed survival, or duration, analysis to study the dynamics and outcomes of armed conflict (Rudloff and Findley 2016; Aliyev 2017). To assess the effects of rebel leader characteristics on the timing of rebel group fragmentation, I use a series of Cox proportional-hazards models. The Cox model estimates the effect of covariates on the hazard rate, which specifies the probability that, if a rebel group “survives” to year t, it will experience fragmentation in the following year. This model—which does not require a specification of the baseline hazard rate—is appropriate as I have no strong reason to expect that rebel fragmentation occurs in a particular pattern.16 For this analysis, the dependent variable is the length of a rebel group's “survival” before splitting into competing factions, measured as time in years.17 In a sample of 156 rebel organizations, 33 percent experienced fragmentation at some point. I include the same set of independent and controlling variables as used in the logistic models, as well as a fixed-effect term for each country in the sample. The main coefficient estimates, which are expressed as hazard ratios, and their estimated Z-scores are shown in Table 3. Table 3. Cox proportional-hazards models, 1989–2014 . Rebel fragmentation . . (6) . (7) . (8) . (9) . Military training 7.199** 4.588** 5.501** (16.79) (12.46) (13.92) Military combat 0.702** 0.903 0.832 (2.77) (0.78) (1.38) Military service 0.036** 0.148** 0.125** (20.71) (13.66) (14.80) Political office 2.458** 2.873** 1.776** (5.06) (5.98) (3.81) Political party 6.081** 1.703** 2.823** (8.82) (2.59) (6.521) Founder 2.431** 3.199** 2.338** (5.58) (8.82) (5.37) Splinter group 1.634** 1.362** 1.593** (2.78) (2.14) (2.66) State sponsor 1.571** 1.342** 1.652** (3.33) (2.51) (3.71) Rebel one-sided violence 1.046 0.993 1.053* (1.58) (0.27) (1.87) Foreign military intervention 0.517** 0.702** 0.511** (5.01) (2.59) (5.09) Conflict intensity 1.000 1.000 1.000 (0.61) (0.49) (0.49) Multiparty conflict 1.537** 2.521** 1.591** (2.18) (5.55) (2.36) Democracy 5.385** 5.897** 5.005** (11.77) (15.08) (11.31) Observations 1,156 1,156 1,156 1,156 # of fragmentation cases 51 51 51 51 . Rebel fragmentation . . (6) . (7) . (8) . (9) . Military training 7.199** 4.588** 5.501** (16.79) (12.46) (13.92) Military combat 0.702** 0.903 0.832 (2.77) (0.78) (1.38) Military service 0.036** 0.148** 0.125** (20.71) (13.66) (14.80) Political office 2.458** 2.873** 1.776** (5.06) (5.98) (3.81) Political party 6.081** 1.703** 2.823** (8.82) (2.59) (6.521) Founder 2.431** 3.199** 2.338** (5.58) (8.82) (5.37) Splinter group 1.634** 1.362** 1.593** (2.78) (2.14) (2.66) State sponsor 1.571** 1.342** 1.652** (3.33) (2.51) (3.71) Rebel one-sided violence 1.046 0.993 1.053* (1.58) (0.27) (1.87) Foreign military intervention 0.517** 0.702** 0.511** (5.01) (2.59) (5.09) Conflict intensity 1.000 1.000 1.000 (0.61) (0.49) (0.49) Multiparty conflict 1.537** 2.521** 1.591** (2.18) (5.55) (2.36) Democracy 5.385** 5.897** 5.005** (11.77) (15.08) (11.31) Observations 1,156 1,156 1,156 1,156 # of fragmentation cases 51 51 51 51 Notes: Two-tailed test; * p < 0.10, ** p < 0.05. Coefficients expressed as hazard ratios. Z-scores shown in parentheses. Fixed effects for country included in each model. Cubic, quadratic, and linear terms included for leader tenure level to address temporal dependency. Open in new tab Table 3. Cox proportional-hazards models, 1989–2014 . Rebel fragmentation . . (6) . (7) . (8) . (9) . Military training 7.199** 4.588** 5.501** (16.79) (12.46) (13.92) Military combat 0.702** 0.903 0.832 (2.77) (0.78) (1.38) Military service 0.036** 0.148** 0.125** (20.71) (13.66) (14.80) Political office 2.458** 2.873** 1.776** (5.06) (5.98) (3.81) Political party 6.081** 1.703** 2.823** (8.82) (2.59) (6.521) Founder 2.431** 3.199** 2.338** (5.58) (8.82) (5.37) Splinter group 1.634** 1.362** 1.593** (2.78) (2.14) (2.66) State sponsor 1.571** 1.342** 1.652** (3.33) (2.51) (3.71) Rebel one-sided violence 1.046 0.993 1.053* (1.58) (0.27) (1.87) Foreign military intervention 0.517** 0.702** 0.511** (5.01) (2.59) (5.09) Conflict intensity 1.000 1.000 1.000 (0.61) (0.49) (0.49) Multiparty conflict 1.537** 2.521** 1.591** (2.18) (5.55) (2.36) Democracy 5.385** 5.897** 5.005** (11.77) (15.08) (11.31) Observations 1,156 1,156 1,156 1,156 # of fragmentation cases 51 51 51 51 . Rebel fragmentation . . (6) . (7) . (8) . (9) . Military training 7.199** 4.588** 5.501** (16.79) (12.46) (13.92) Military combat 0.702** 0.903 0.832 (2.77) (0.78) (1.38) Military service 0.036** 0.148** 0.125** (20.71) (13.66) (14.80) Political office 2.458** 2.873** 1.776** (5.06) (5.98) (3.81) Political party 6.081** 1.703** 2.823** (8.82) (2.59) (6.521) Founder 2.431** 3.199** 2.338** (5.58) (8.82) (5.37) Splinter group 1.634** 1.362** 1.593** (2.78) (2.14) (2.66) State sponsor 1.571** 1.342** 1.652** (3.33) (2.51) (3.71) Rebel one-sided violence 1.046 0.993 1.053* (1.58) (0.27) (1.87) Foreign military intervention 0.517** 0.702** 0.511** (5.01) (2.59) (5.09) Conflict intensity 1.000 1.000 1.000 (0.61) (0.49) (0.49) Multiparty conflict 1.537** 2.521** 1.591** (2.18) (5.55) (2.36) Democracy 5.385** 5.897** 5.005** (11.77) (15.08) (11.31) Observations 1,156 1,156 1,156 1,156 # of fragmentation cases 51 51 51 51 Notes: Two-tailed test; * p < 0.10, ** p < 0.05. Coefficients expressed as hazard ratios. Z-scores shown in parentheses. Fixed effects for country included in each model. Cubic, quadratic, and linear terms included for leader tenure level to address temporal dependency. Open in new tab The main coefficients of interest are expressed as hazard ratios in Table 3. Hazard ratios offer an intuitive means of interpretation, where values less than 1 indicate a relative drop in the hazard of fragmentation, all else equal. Conversely, hazard ratios greater than 1 suggest an increased hazard of fragmentation. Like the logistic model, the results of the survival analysis provide mixed support for my argument that leaders with prior military experience are at lesser risk of fragmentation. All else equal, rebel leaders who have military training operate at a considerably higher risk of fragmentation, with the expected hazard increasing by 450 percent relative to leaders with no former training. In contrast to the logistic regression estimate, a leader's prior combat experience has no discernible effect on the timing of fragmentation once group and environmental controls are included. This suggests that the leadership strengths that come with prior combat experience make rebel organizations more cohesive overall, but do not further insulate them from the pressures that cause fragmentation to occur in particular stages of an insurgency. Conversely, rebel leaders who previously served as a ranking member of a state military are associated with an expected hazard 88 percent lower than those who did not, all else equal. Overall, however, the logistic regression model indicates that these leaders will experience fragmentation as often as those with no prior military service, all else equal. This suggests that experience in a state military may equip rebel leaders with the know-how necessary to quickly consolidate a commanding control over their organizations, thus delaying any episodes of fragmentation. The estimated hazard rates corresponding with leaders’ pre-war political experiences tell an equally compelling story. Political experience seems to matter. The Cox model results indicate that individuals who hold national political office prior to becoming a rebel leader experience fragmentation at a higher rate. Specifically, Model 9 estimates that these leaders are associated with a 78 percent increase in the hazard of fragmentation. Similarly, based on the estimates from the same model, the expected hazard of fragmentation increases by 182 percent for rebel leaders who previously acted as heads of a political party, all else equal. In tandem with the results of the logistic analysis, this finding indicates that former political party leaders are not only more likely to experience fragmentation, but also they will do so at a higher rate. That is, political leaders experience fragmentation both more often and sooner in the campaign. This supports the expectation articulated in Hypothesis 2, and my argument that these leaders may struggle to send credible signals that they are prepared to manage the political and military operations necessary to maintain a viable insurgency. Given the mechanisms I describe as connecting leaders’ prior political experiences to fragmentation, it makes sense that the estimated effect of political experience is larger for former party leaders than for former holders of political office. These individuals should wield considerable capacity for mobilization, but in the anarchic field of war, the benefits of political savvy are marginal and short-lived in the absence of military acumen. The results from Table 3 are also displayed in Figure 2, which presents the estimated survival curves for each of the main independent variables as they relate to rebel fragmentation. Figure 2. Open in new tabDownload slide Estimated survival curves, 1989–2014 Figure 2. Open in new tabDownload slide Estimated survival curves, 1989–2014 Figure 2 displays a rebel group's estimated probability of “survival” (i.e., not experiencing fragmentation) in a given year according to variation in the characteristics of its leadership. Using the estimates from Model 9 shown in Table 3, the displayed survival functions are calculated by holding all controlling variables at their means (continuous variables) or modal categories (discrete variables) while switching the specified variable of interest from “0” to “1.” In short, the figure plots the probability of groups surviving—or not fragmenting—over the duration of a leader's tenure. The solid black line in each panel represents the baseline survival function; the grey lines depict the survival functions associated with a positive value of the leadership variable of interest. A few trends emerge from this analysis. First, in all five panels, the estimated probabilities of survival for each group only begin to separate after the first four years of a leader's tenure. This suggests that, regardless of who is charge, most rebel groups enjoy similarly high levels of cohesion in the earliest phases of an insurgency. Second, the effect of prior military experience is intensified as rebel groups continue operations. For example, rebel groups managed by leaders with prior service in a state military are quite resilient to fragmentation, with a 0.95 probability of survival at ten years of fighting. By comparison, leaders with no prior military service have only a 0.60 probability of surviving without splitting at ten years, a considerable difference of 0.35. The size of the gap between these groups widens with time; the difference in survival probabilities increases to 0.50 at twelve years of fighting and 0.60 at fifteen years. In addition, those leaders who receive military training are more likely to experience fragmentation sooner than those without training, controlling for other forms of military experience as well as other factors. After six years of activity, leaders with military training but no other forms of military experience have a 0.71 probability of survival, as compared to 0.90 for all others. At year ten, the probability gap between these types grows to 0.55. Third, there is no discernible, independent effect of a leader's prior combat experience on the risk or timing of group fragmentation. In conjunction with the results of the logistic regression analysis, this result indicates that combat veterans are better at mitigating threats to group cohesion but, in comparison to other leaders, are no more effective at certain periods of their tenure than others. As demonstrated in Table 2, in contrast to former military servicemen, rebel leaders who rely solely on military training are actually at greater risk of fragmentation. This strongly suggests that simulated military experience may actually serve to weaken group cohesion. At year five, leaders with military training have an estimated 0.84 probability of survival, while those who have no military training have a 0.98 probability of survival, a 0.14 difference. Over the following years, this gap increases rapidly to roughly 0.70. What does this suggest about military experience? Taking a look under the hood, in RLCW, many of the rebel leaders whose only prior military experience is military training are trained ad hoc—i.e., they are trained specifically for the purposes of launching an insurgency—and often do so alongside their initial corps of recruits. It is quite plausible that this makes subcommanders—who have seen as much of war as their leader—more confident in their abilities to do as well or better as leaders of their own factions. This is a compelling dynamic that merits further investigation. Rebel leaders who previously led a political party are more prone to see cases of splintering, all else equal. After six years, rebel leaders who previously acted as heads of a political party are more likely to experience fragmentation than not. By year ten, former party heads have only a 0.25 probability of avoiding fragmentation, as compared to an estimated 0.65 probability for those commanders with no experience as party leaders, a probability difference of 0.40. Similarly, former holders of political office have a 0.20 higher probability of fragmentation at this same point in their tenure than those with no experience in political office. Overall, Figure 2 further demonstrates that the experiences that rebel leaders bring with them into conflict often carry important implications for their patterns of management and, subsequently, the risk of rebel fragmentation over time. Conclusion Rebel fragmentation is a common feature of civil conflict. This study provides a leader-based explanation for why some rebel organizations split into competing factions while others remain coherent. Together, the logistic and survival analyses offer needed insight into the microfoundations of rebel fragmentation and raise important questions for future analysis. In sum, I find that variation in rebel leadership offers empirical leverage over the puzzle of rebel fragmentation. First, in many cases, rebel leaders with prior military experience are better equipped to maintain group cohesion. The models demonstrate that leaders who are combat veterans experience fragmentation less often and those with a history of military service are able to maintain cohesion for a longer period of time. Conversely, this relationship does not stand for leaders that only have prior military training; indeed, rebel groups led by such leaders are actually more prone to fragmentation. This stresses the importance of real over simulated military experience. In addition, rebel leaders with previous experience as head of a political party—who may find it especially difficult to manage the military operations essential to a successful insurgency—are more prone to fragmentation. This offers further evidence that, though insurgency is a joint political-military campaign, a leader's demonstrated capacity for military logistical, organizational, and operational effectiveness strongly informs subcommanders’ confidence in the leader's ability to achieve victory against state and rival forces. This study makes a number of significant contributions to the conflict literature, bearing important implications for theoretical approaches to rebel leadership and rebel fragmentation. First, existing theories have struggled to identify “different characteristics and tendencies of rebel organizations that may help to explain cross-sectional differences in fragmentation” (Woldemariam 2018, 287). This study speaks to this gap, offering an organizational theory of fragmentation based on variation in rebel leadership. While external pressures—e.g., imbalance of military capacity, pressure from external supporters, or a dearth of extractable resources—certainly influence a subcommander's decision to create a splinter faction, leaders play a critical role in this process by either mitigating or accelerating the factionalizing influence of these dynamics. Indeed, I find evidence here that some leaders are better equipped than others to maintain cohesion in the face of such pressures. Second, this study makes a broader contribution to the literature by offering the first description of how rebel leadership characteristics affect the dynamics of civil war. Long emphasized in explanations of interstate conflict, leadership receives far less direct attention in the conflict and security literature. In particular, I describe how rebel leaders draw on their pre-war experiences to make decisions in the war stage. Decisionmakers have long been interested in rebel leadership. Indeed, the US Army and Marine Corps Counterinsurgency Field Manual (2007, 1.13) emphasizes the following: Leadership is critical to any insurgency. An insurgency is not simply random violence; it is directed and focused violence aimed at achieving a political objective. It requires leadership to provide vision, direction, guidance, coordination, and organizational coherence. Successful insurgent leaders make their cause known to the people and gain popular support. Their key tasks are to break the ties between the people and the government and to establish credibility for their movement. Their education, background, family and social connections, and experiences contribute to their ability to organize and inspire the people who form the insurgency. I take this perspective seriously and offer foreign policy and security practitioners a set of tools to use towards understanding how personal characteristics affect the decisions rebel leaders make in the management of insurgent warfare. Overall, this study offers a generalizable and accessible framework with which policymakers can improve their expectations about the probability of observing fragmentation in a rebel movement of strategic interest. This study also reveals some exciting opportunities for future inquiry. First, the influence of leaders’ prior military experiences on rebel fragmentation varies, suggesting the need for additional investigation into how leaders draw on specific forms of military experience both to maintain group cohesion and to sustain a viable insurgency. Second, it is plausible that leaders have access to unique recruitment pools during the mobilization stage, indicating the relevance of leaders’ pre-war origins and networks to the cohesion and strength of a rebel organization (Staniland 2014). Future studies should further investigate this dynamic: Do certain rebel leader types also bring with them subcommanders and fighters of a certain type or quality?18 Third, in this study, I have offered a generalizable theory of rebel leadership and fragmentation. An interesting and worthwhile extension would investigate and model directly how each leader type navigates the common flashpoint events supposed to be associated with fragmentation, e.g., military losses or the onset of peace talks. Fourth, I theorize the effects of a leader's military and political experience separately and independently. As such, while the results of this study do not indicate that the effects of these separate leadership dimensions are conditional on one another, it is still possible that the demonstrated effect of either military or political experience may “wash out” the other. That is, because some of these experiences work in opposite directions, a rebel leader with both military and political backgrounds may not have a discernible likelihood of fragmentation relative to those leaders with neither of these pre-war experiences. This issue warrants additional inquiry in a later study. Finally, while I take a quantitative approach to test my theoretical hypothesis, more qualitative work is needed to illuminate and trace the mechanisms that connect variation in rebel leadership to the management of armed rebel organizations. In addition to explaining why rebel organizations splinter into competing factions, this study speaks to the broader issue of structure and agency in civil war. The results suggest the utility of leveraging variation in rebel leadership to explain rebel behavior and performance. Rebel leaders are not created equal; neither do they emerge at civil war onset with the same capacities for leadership. Instead, leaders draw on their past experiences to manage their organizations on and off the battlefield. And this study indicates that the patterns of leadership corresponding with these characteristics have lasting implications on a rebel group's risk of fragmentation. Acknowledgements A version of this article was presented at the 2018 Annual Meeting of the American Political Science Association in Boston, MA. For valuable feedback, I wish to thank Chad Clay, John Willingham, Amanda Murdie, Danny Hill, Cas Mudde, Ryan Bakker, Stephen Bagwell, Jess Braithwaite, Sirianne Dahlum, and the article's anonymous reviewers. All remaining errors are my own. Footnotes 1 Similarly, Woldemariam (2018, 25) defines fragmentation as “an event where a segment of a rebel group formally and collectively exits that rebel organization and either (a) establishes a new rebel organization, (b) joins an existing rebel organization, or (c) joins the incumbent government.” I consider these three categories of fragmentation to embody meaningfully distinct processes and, thus, choose to focus on explaining the first of these types, the establishment of a new rebel organization. 2 There are also a number of excellent studies on the causes and consequences of rebel movement proliferation and fragmentation. For notable examples, see Cunningham (2011), Bakke, Cunningham, and Seymour (2012), McLauchlin and Pearlman (2012), Dowd (2015), Seymour, Bakke, and Cunningham (2016), Mosinger (2018), and Fjelde and Nilsson (2018). 3 Relatedly, some studies suggest that successful leadership targeting attacks may increase the probability of fragmentation—though the generalizability of this relationship remains untested. For instance, Staniland (2014, 47) argues that, if counterinsurgent forces are “able to regularly arrest or kill key leaders, central processes will decay and perhaps collapse altogether,” with fragmentation soon to follow. 4 Gates (2017) comes to a similar conclusion based on a study on forced recruitment and rebel allegiance. 5 In a related working paper, for instance, Willingham and Doctor (2015) find that the level of violence inflicted by a rebel group against civilians in a given year corresponds with the group leader's military experiences and motives for conflict. 6 Notable exceptions include research by Tiernay (2015), Hoover Green (2016), Prorok (2016), and Cunningham and Sawyer (2019). My theory builds on these studies to demonstrate how rebel leaders’ individual experiences and characteristics shape the management of armed rebel groups in civil war. 7 Relatedly, in a study on militant loyalties in the Central African Republic, Debos (2008, 236) writes, “Political and military actors are continuously reconsidering their tactical loyalty to the regime. Their decision to join a rebellion or not depends on political calculations based on their perception of the situation and on their own ambition to achieve a better position within the security forces.” 8 Lidow (2016, 38) makes a similar point: “The effects of leader charisma and credibility also trickle down to the soldiers. When soldiers are inspired by the leader and believe in her long-term promises, their commanders feel increased pressure to follow the leader's orders…Credibility and charisma create positive incentives for discipline at every level of the organization, and decrease the opportunities for defection among the top commanders.” 9 A related explanation might focus on the charisma of the rebel leader, where the leader enjoys higher levels of support and obedience due to their inherent ability to win the affection of their fighters. Indeed, some individuals seem to possess a greater capacity for endearing themselves to others—and many such individuals emerge in positions of leadership. The concept of “charisma,” however, is quite vague and difficult to measure systematically. Moreover, a leader's degree of charisma may itself be a function of the leadership dimensions investigated in this study. As such, I focus on more concrete, observable, and measurable experiences that may shape how leaders make decisions and, thus, cultivate the loyalty of their subordinates. 10 Privately, Taylor's recruitment pitch was less ingenuous. In addition to guaranteeing an ample supply of weapons for the offensive, Taylor promised each of his early subcommanders—who he called his “special forces”—a large house, a $75,000 reward, and positions of prominence in his post-conflict administration (Lidow 2016, 116). 11 A shipment of arms from Cote d'Ivoire was delayed by several weeks because the NPFL lacked the necessary funds to hire trucks to transport the munitions. Moreover, the first attack was planned as a coordinated and simultaneous assault from three locations: Nimba County, Monrovia, and the Guinea- Liberia border. The assault at the tactical front in Nimba County was successful, but twenty-two of the NPFL fighters positioned in Monrovia were arrested and the forces sent to Guinea experienced a number of logistical issues such that they missed the attack altogether (Ellis 1999; Lidow 2016). 12 As a starting point for the coding process, the set of rebel leaders featured in the RLCW data was determined by the sample of rebel organizations featured in the Uppsala Conflict Data Program (UCDP) One-Sided Violence Dataset (Allansson, Melander, and Themner 2017). Importantly, the set of group-year observations in our sample does not include only those in which anti-civilian violence is observed, but it is limited to those which are recorded as inflicting violence against civilians at some point in their rebel campaign. Efforts to expand the set of rebel actors in the RLCW data are underway. In the meantime, we have little reason to believe that this selection process biases our model estimates. Moreover, we control for variation in rebel one-sided violence in our main analysis. Coding guidelines are detailed in the supplementary online appendix. 13 According to the RLCW coding guidelines, in theory, rebel leaders can update their combat experience during war—it is quite rare, however, in practice. This experience is not automatically assigned to leaders as a function of active conflict. Rather, to make this coding decision, RLCW coders searched for specific evidence that a given leader was actively and systematically present on the battlefield in a given conflict-year. 14 I use the coefficients found in Model 5 to predict the probability of fragmentation based on changes in specified predictors while holding all other variables at their mean values (continuous) and modal categories (discrete). 15 These nine categories are: group election, direct seizure of power (internal coup), founder–faction, founder–original, executive appointment, military succession, accord collapse, coalition merger, and unknown. 16 An analysis of Grambsch and Therneau's (1994) “global test” and subsequent calculations of Harrell's rho indicate that the proportional hazards assumption is violated by three of the controlling variables: “splinter group,” “group age,” and “foreign military intervention.” The main model estimates, however, are resilient to the included interaction of a logged time trend with these variables (Box-Steffensmeier and Jones 2004), allowing me to report the estimated results with sufficient confidence in the model choice. 17 Of course, many rebel groups never experience fragmentation at all during a conflict. Accordingly, I account for all cases of “right censoring”—i.e., rebel groups that never experience fragmentation within the timeframe of the dataset (1989–2014)—with a binary indicator. Unaccounted for, right censoring introduces a form of selection bias: in effect, selecting on the dependent variable (Box-Steffensmeier and Jones 2004, 16–18). 18 One promising research project on this topic is the Foundations of Rebel Group Emergence (FORGE) project, Jessica Maves Braithwaite and Kathleen Gallagher Cunningham, “When Organizations Rebel: Introducing the Foundations of Rebel Group Emergence Dataset,” https://www.jessicamaves.com/forge.html. References Ahlquist John S. , Levi Margaret . 2011 . “ Leadership: What It Means, What It Does, and What We Want to Know about It .” Annual Review of Political Science 14 : 1 – 24 . Google Scholar Crossref Search ADS WorldCat Aliyev Huseyn . 2017 . “ Pro-Regime Militias and Civil War Duration .” Terrorism and Political Violence 0 ( 0 ): 1 – 21 . Google Scholar Crossref Search ADS WorldCat Allansson Marie , Melander Erik , Themner Lotta . 2017 . “ Organized Violence, 1989–2016 .” Journal of Peace Research 54 ( 4 ): 574 – 87 . Google Scholar Crossref Search ADS WorldCat Al-Muqrin Abd al-Aziz . 2008 . Al-Qa’ida’s Doctrine for Insurgency: Abd al-Aziz al-Muqrin’s ‘A Practical Course for Guerrilla Warfare’ . Translated by Norman Cigar . Washington, DC : Potomac Books Inc. Asal Victor , Brown Mitchell , Dalton Angela . 2012 . “ Why Split? Organizational Splits Among Ethnopolitical Organizations in the Middle East .” Journal of Conflict Resolution 56 ( 1 ): 94 – 117 . Google Scholar Crossref Search ADS WorldCat Bacon Tricia , Arsenault Elizabeth Grimm . 2017 . “ Al Qaeda and the Islamic State's Break: Strategic Strife or Lackluster Leadership? ” Studies in Conflict & Terrorism : 1 – 35 . OpenURL Placeholder Text WorldCat Bakke Kristin M. , Cunningham Kathleen Gallagher , Seymour Lee . 2012 . “ A Plague of Initials: Fragmentation, Cohesion, and Infighting in Civil Wars .” Perspectives on Politics 10 ( 2 ): 265 – 83 . Google Scholar Crossref Search ADS WorldCat Blattman Christopher , Miguel Edward . 2010 . “ Civil War .” Journal of Economic Literature 48 ( 1 ): 3 – 57 . Google Scholar Crossref Search ADS WorldCat Box-Steffensmeier Janet M. , Jones Bradford S. . 2004 . Event History Modeling: A Guide for Social Scientists . Cambridge : Cambridge University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Christia Fotini . 2012 . Alliance Formation in Civil Wars . Cambridge, MA : Cambridge University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Cunningham Kathleen Gallagher . 2011 . “ Divide and Conquer or Divide and Concede: How Do States Respond to Internally Divided Separatists? ” American Political Science Review 105 ( 2 ): 275 – 97 . Google Scholar Crossref Search ADS WorldCat Cunningham Kathleen Gallagher , Sawyer Katherine . 2019 . “ Conflict Negotiations and Rebel Leader Selection .” Journal of Peace Research 56 ( 5 ): 619 – 34 . Google Scholar Crossref Search ADS WorldCat Cunningham Kathleen Gallagher , Bakke Kristin M. , Seymour Lee . 2012 . “ Shirts Today, Skins Tomorrow: Dual Contests and the Effects of Fragmentation in Self-Determination Disputes .” Journal of Conflict Resolution 56 ( 1 ): 67 – 93 . Google Scholar Crossref Search ADS WorldCat Day Christopher R. 2017 . “ ‘Survival Mode’: Rebel Resilience and the Lord's Resistance Army .” Terrorism and Political Violence 31 ( 5 ): 966 – 86 . Google Scholar Crossref Search ADS WorldCat Debos Marielle . 2008 . “ Fluid Loyalties in a Regional Crisis: Chadian ‘Ex-Liberators’ in the Central African Republic .” African Affairs 107 ( 427 ): 225 – 41 . Google Scholar Crossref Search ADS WorldCat Dowd Caitriona . 2015 . “ Actor Proliferation and the Fragmentation of Violent Groups in Conflict .” Research & Politics 2 ( 4 ). https://doi.org/10.1177/2053168015607891. OpenURL Placeholder Text WorldCat Driscoll Jesse . 2012 . “ Commitment Problems or Bidding Wars? Rebel Fragmentation as Peace Building .” Journal of Conflict Resolution 56 ( 1 ): 118 – 49 . Google Scholar Crossref Search ADS WorldCat Eck Kristine , Hultman Lisa . 2007 . “ One-Sided Violence Against Civilians in War: Insights from New Fatality Data .” Journal of Peace Research 44 ( 2 ): 233 – 46 . Google Scholar Crossref Search ADS WorldCat Ellis Stephen . 1999 . The Mask of Anarchy: The Destruction of Liberia and the Religious Dimension of an African Civil War . London : Hurst Publishers . Findley Michael , Rudloff Peter . 2012 . “ Combatant Fragmentation and the Dynamics of Civil Wars .” British Journal of Political Science 42 ( 4 ): 879 – 901 . Google Scholar Crossref Search ADS WorldCat Fjelde Hanne , Nilsson Desiree . 2018 . “ The Rise of Rebel Contenders: Barriers to Entry and Fragmentation in Civil Wars .” Journal of Peace Research 55 ( 5 ): 551 – 65 . Google Scholar Crossref Search ADS WorldCat Gates Scott . 2017 . “ Membership Matters: Coerced Recruits and Rebel Allegiance .” Journal of Peace Research 54 ( 5 ): 674 – 86 . Google Scholar Crossref Search ADS WorldCat Grambsch Patricia M. , Therneau Terry M. . 1994 . “ Proportional Hazards Tests and Diagnostics Based on Weighted Residuals .” Biometrika 81 ( 3 ): 515 – 26 . Google Scholar Crossref Search ADS WorldCat Hazen Jennifer M . 2013 . What Rebels Want: Resources and Supply Networks in Wartime . Ithaca, NY : Cornell University Press . Hoover Green Amelia . 2016 . “ The Commander's Dilemma: Creating and Controlling Armed Group Violence .” Journal of Peace Research 53 ( 5 ): 619 – 32 . Google Scholar Crossref Search ADS WorldCat Horowitz Michael C. , Stam Allan C. , Ellis Cali M. . 2015 . Why Leaders Fight . Cambridge, MA : Cambridge University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Jones Seth G . 2017 . Waging Insurgent Warfare: Lessons from the Vietcong to the Islamic State . Oxford : Oxford University Press . Kalyvas Stathis N. 2006 . The Logic of Violence in Civil War . Cambridge, MA : Cambridge University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Kenny Paul . 2010 . “ Structural Integrity and Cohesion in Insurgent Organizations: Evidence from Protracted Conflicts in Ireland and Burma .” International Studies Review 12 ( 4 ): 533 – 55 . Google Scholar Crossref Search ADS WorldCat Lidow Nicholai Hart . 2016 . Violent Order: Rebel Organization and Liberia's Civil War . Cambridge, MA : Cambridge University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mahoney Charles W. 2017 . “ Splinters and Schisms: Rebel Group Fragmentation and the Durability of Insurgencies .” Terrorism and Political Violence . https://www.tandfonline.com/doi/full/10.1080/09546553.2017.1374254 . OpenURL Placeholder Text WorldCat Marshall Monty G. , Jaggers Keith , Gurr Ted Robert . 2014 . “ Polity IV Annual Time-Series, 1800–2013 .” Center for International Development and Conflict Management at the University of Maryland, College Park . OpenURL Placeholder Text WorldCat McLauchlin Theodore , Pearlman Wendy . 2012 . “ Out-Group Conflict, In-Group Unity? Exploring the Effect of Repression on Intramovement Cooperation .” Journal of Conflict Resolution 56 ( 1 ): 41 – 66 . Google Scholar Crossref Search ADS WorldCat Mosinger Eric S. 2018 . “ Brothers or Others in Arms? Civilian Constituencies and Rebel Fragmentation in Civil War .” Journal of Peace Research 55 ( 1 ): 62 – 77 . Google Scholar Crossref Search ADS WorldCat Nagel Robert U. , Doctor Austin C. . 2019 . “ Conflict-Related Sexual Violence and Rebel Fragmentation .” Presented at the Annual Meeting of the International Studies Association in Toronto, Canada . OpenURL Placeholder Text WorldCat Nilsson Desiree . 2010 . “ Turning Weakness into Strength: Military Capabilities, Multiple Rebel Groups and Negotiated Settlements .” Conflict Management and Peace Science 27 ( 3 ): 253 – 71 . Google Scholar Crossref Search ADS WorldCat Olson Lounsbery Marie . 2016 . “ Foreign Military Intervention, Power Dynamics, and Rebel Group Cohesion .” Journal of Global Security Studies 1 ( 2 ): 127 – 41 . Google Scholar Crossref Search ADS WorldCat Otto Sabine . 2018 . “ The Grass is Always Greener? Armed Group Side Switching in Civil Wars .” Journal of Conflict Resolution 62 ( 7 ): 1459 – 88 . Google Scholar Crossref Search ADS WorldCat Perkoski Evan James . 2015 . “ Organizational Fragmentation and the Trajectory of Militant Splinter Groups PhD .” Publicly Accessible Penn Dissertations, 1943. https://repository.upenn.edu/edissertations/1943 . OpenURL Placeholder Text WorldCat Pettersson Therése , Eck Kristine . 2018 . “ Organized Violence, 1989–2017 .” Journal of Peace Research 55 ( 4 ): 535 – 47 . Google Scholar Crossref Search ADS WorldCat Prorok Alyssa K. 2016 . “ Leader Incentives and Civil War Outcomes .” American Journal of Political Science 60 ( 1 ): 70 – 84 . Google Scholar Crossref Search ADS WorldCat Rudloff Peter , Findley Michael G. . 2016 . “ The Downstream Effects of Combatant Fragmentation on Civil War Recurrence .” Journal of Peace Research 53 ( 1 ): 19 – 32 . Google Scholar Crossref Search ADS WorldCat Seymour Lee J.M. , Bakke Kristin M. , Cunningham Kathleen G. . 2016 . “ E Pluribus Unum, Ex Uno Plures: Competition, Violence, and Fragmentation in Ethnopolitical Movements .” Journal of Peace Research 53 ( 1 ): 3 – 18 . Google Scholar Crossref Search ADS WorldCat Shapiro Jacob N. 2013 . The Terrorist's Dilemma: Managing Violent Covert Organizations . Princeton, NJ : Princeton University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Sinno Abdulkader H. 2008 . Organizations at War in Afghanistan and Beyond . Ithaca, NY : Cornell University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Staniland Paul . 2014 . Networks of Rebellion: Explaining Insurgent Cohesion and Collapse . Ithaca, NY : Cornell University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tamm Henning . 2016 . “ Rebel Leaders, Internal Rivals, and External Resources: How State Sponsors Affect Insurgent Cohesion .” International Studies Quarterly 60 ( 4 ): 599 – 610 . Google Scholar Crossref Search ADS WorldCat The United States Department of the Army . 2007 . The U.S. Army/Marine Corps Counterinsurgency Field Manual. Chicago : University of Chicago Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tiernay Michael . 2015 . “ Killing Kony: Leadership Change and Civil War Termination .” Journal of Conflict Resolution 59 ( 2 ): 175 – 206 . Google Scholar Crossref Search ADS WorldCat Waugh Colin M. 2011 . Charles Taylor and Liberia: Ambition and Atrocity in Africa's Lone Star State . Chicago, IL : Zed Books . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Weinstein Jeremy M. 2007 . Inside Rebellion: The Politics of Insurgent Violence . Cambridge, MA : Cambridge University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Williams Paul D. 2016 . War and Conflict in Africa , 2nd ed. Cambridge, UK : Polity Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Willingham John D. 2017 . “ Enter Stage Right: Rebel Leader Type and Conflict Duration in Intrastate War .” Theses and Dissertations, University of Georgia . OpenURL Placeholder Text WorldCat Willingham John D. , Doctor Austin C. . 2015 . “ Follow the Leader: Rebel Leaders and Anti-Civilian Violence in Intrastate War .” Presented at the Annual Meeting of the Peace Science Society–International, Oxford, Mississippi . OpenURL Placeholder Text WorldCat Woldemariam Michael . 2016 . “ Battlefield Outcomes and Rebel Cohesion: Lessons from the Eritrean Independence War .” Terrorism and Political Violence 28 ( 1 ): 135 – 56 . Google Scholar Crossref Search ADS WorldCat Woldemariam Michael . 2018 . Insurgent Fragmentation in the Horn of Africa: Rebellion and its Discontents . Cambridge, MA : Cambridge University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC © The Author(s) (2020). Published by Oxford University Press on behalf of the International Studies Association. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - A Motion of No Confidence: Leadership and Rebel Fragmentation
JF - Journal of Global Security Studies
DO - 10.1093/jogss/ogz060
DA - 2020-05-21
UR - https://www.deepdyve.com/lp/oxford-university-press/a-motion-of-no-confidence-leadership-and-rebel-fragmentation-IOxuXuO7wh
DP - DeepDyve
ER -