ABSTRACT Title of dissertation: WHY DO REBELS SPLIT? EXAMINING THE CAUSES OF REBEL GROUP FRAGMENTATION Moran Stern Doctor of Philosophy, 2022 Dissertation directed by: Professor Shibley Telhami Department of Government and Politics Why do rebel groups undergo fragmentation? While extensive research about the consequences of rebel fragmentation exists, research on the process of fragmentation remains relatively nascent. This dissertation collects three papers on the causes of rebel group fragmentation. In the first paper, I develop a junior cadres-based explanation of fragmentation. I argue that in a centralized rebel group, factions will emerge when leaders block junior cadres? access to senior decision-making bodies. Junior cadres who want to influence the organization?s politics therefore face a choice between remaining within the rebel group and exiting it. Factionalizing is a way to redress grievances by aggrieved junior cadres who deem peaceful mechanisms for upward mobility ineffective. Using original datasets and personal interviews, I find strong evidence supporting my argument in the case of Palestinian Fatah. In the second paper, I argue that the solution to the question of fragmentation lies in rebel socialization?specifically, military training (MT). MT increases group cohesion by strengthening horizontal bonds among combatants; vertical bonds between combatants and commanders; and members? institutional bonds to the organization?s overall mission and esprit de corps. Members become mutually dependent, thus making splintering more costly and fragmentation less likely. I test this argument on a global sample of 83 rebel groups active between 1989 and 2010. I find that rebel groups that have recently conducted MT are less likely to fragment by about 75 percent. In the third paper, I explore the effect of foreign fighters (FFs) on rebel fragmentation, examining a number of mechanisms derived from previous research. First, I explore how reduced group dependency on local fighters, preference divergence, strategic disagreements, and member segregation increase the likelihood of fragmentation for rebel groups that recruit FFs. Second, I posit that if the foreignness of FFs in relation to local insurgents makes fragmentation more likely, then rebel groups that recruit coethnic FFs will be less likely to experience fragmentation. I test these arguments on a global sample of 227 rebel groups active between 1989 and 2011. I find that rebel groups that recruit FFs are significantly more likely to fragment, even after accounting for the endogenous choice of rebel groups to recruit FFs. Against my expectations, I find that the recruitment of coethnic FFs does not diminish the probability of fragmentation. This finding raises questions about the value of ethnic homogeneity in the context of FFs in particular. WHY DO REBELS SPLIT? EXAMINING THE CAUSES OF REBEL GROUP FRAGMENTATION by Moran Stern Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park, in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2022 Advisory Committee: Professor Shibley Telhami, Chair Professor David Cunningham Professor Kathleen Cunningham Professor Mark Lichbach Professor Arie Kruglanski (Dean?s representative) ? Copyright by Moran Stern 2022 Dedication In memory of Julia. In celebration of Emilia & Neomi. ii Acknowledgements Over the course of the past eight years there have been so many incredible individuals in my life without which this journey would have never become viable. First and foremost, I would like to thank Stephanie Eder, my partner in life, who has supported me unconditionally throughout this project. This dissertation is also hers. I also send my deepest love to my two daughters, Emilia and Neomi, who have made me put things into perspective every morning (and occasionally during the middle of night). This project would not have been possible without the support of my committee. Above all, I am truly grateful to Shibley Telhami who has been my advisor from the beginning of my time at the University of Maryland. Shibley believed in me from the start when he gave me the opportunity to join the doctoral program, and for that I will be always indebted to him. Over the years, Shibley has guided, mentored, intellectually disciplined, and treated me with the utmost patience necessary to complete this project. I also thank Kathleen Cunningham, whose research inspired my dissertation on rebel fragmentation and who has provided me incredible feedback on my work. I thank Mark Lichbach for the many hours of coursework and office hours ? all of which have been instrumental in helping me think like a political scientist. A special thank you goes to David Cunningham. David?s commitment to my research, close reading of my work, and investment in my intellectual development has been exemplary. Beyond my committee, many others at the University of Maryland have shaped my trajectory as a graduate student. Paul Huth, one of the sharpest minds I have ever met, contributed to my development as a scholar by shaping my work and the ways in which I think about global affairs. John McCauley, a mentor and a friend, who always iii found the time to pick up the phone and answer my questions. Mike Hanmer and Patrick Wohlfarthas, who served as the Director of Graduate Studies, were absolutely superb in providing the professional support I needed to successfully finish my dissertation. I deeply thank Ann Marie Clark, Elaine Rudder, Karmin Cortes, and Stefanie Drame for making the most complex administrative problems appear so simple. I am grateful for the generous donors and committee who selected my article for the 2021-2022 Don C. Piper Award. I am extremely thankful to two extremely bright and dear colleagues and friends of mine, Jori Breslawski and Trey Billing, for reading drafts of my work, brainstorming ideas with me, spending countless hours helping me solve methodological problems, and believing in my research. I thank Thair Abu-Rass for helping me preparing for my fieldwork. I am also thankful to all my students at UMD. Class discussions with them sparked new ideas and honed existing ones. Their support was key in receiving the 2021 Excellence in Teaching Award. Finally, I am grateful to Yoram Peri, Paul Scham, Avis Koeiman, Jaclyn Hopkins, and the Gildenhorn Institute for Israel Studies at the University of Maryland, who brought me on as a Graduate Fellow and provided me with the necessary financial, administrative, and logistical support. Outside of UMD, I am deeply thankful to the amazing individuals at Center for Jewish Civilization (CJC) at Georgetown University, which has been a second home to me for over a decade and has provided me a venue to hone my teaching skills. In particular, I am sincerely thankful to Directors Jacques Berlinerblau and Bruce Hoffman; to consecutive program managers Melissa Spence, Anna Dubinsky, and Brittany Fried; to the hundreds of students who have allowed me to be part of their iv undergraduate education; and critically, to three exceptional research assistants who have meticulously tied the knots of my research: Maddox Angerhofer, Marie-Ann Wells, and Michelle Fan. Special thanks go to Ambassador Dennis Ross, a mentor and a role model, whose office door has always been open for me; Michael Milshtein, whose expertise on all-things-Palestinian was invaluable for my research; and to Elior Levy for helping me navigate the maze of Palestinian politics. I thank Karim Mezran, a friend and mentor, who insisted on me pursuing a doctoral degree. I am thankful to the two anonymous reviewers at Civil Wars journal whose tough, yet much appreciated, critiques have pushed me to get my work published. I am extremely grateful to S. Daniel Abraham for so generously supporting my studies and research. I thank all those interviewed in the Palestinian Authority for their time, hospitality, courage, and willingness to share their stories and perspectives. I would also like to thank my dear friends John Probyn, Yoni Komorov, Ohad, Itamar, and Itay for listening, advising, and believing in my decisions. Lastly, I thank my mom, Esty, who has taught me the virtues of curiosity and knowledge; my dad, Yossi, who has supported my studies and constantly shares how proud I make him; and my brother, Ra?anan who is there for me whenever and wherever I need him. This dissertation is for them too. v Table of Contents Dedication ii Acknowledgements iii Table of Contents vi Introduction 1 1 Factionalization From Below: The Case of Palestinian Fatah 9 1.1 Introduction .................................................................................................... 10 1.2 Existing Explanations of Rebel Group Fragmentation .................................. 13 1.3 The Argument: Political Exclusion, Junior Cadres? Grievances, and Factionalization .................................................................................................... 19 1.4 The Data ......................................................................................................... 25 1.5 Assessing the Argument ................................................................................. 28 1.5.1 Fatah?s Centralized Leadership ............................................................. 28 1.5.2 Factionalization Despite Centralized Leadership .................................. 35 1.5.3 Determinants of Factionalization in Fatah ............................................ 37 1.5.4 Limitations of Alternative Explanations ............................................... 43 1.6 Conclusion...................................................................................................... 45 1.7 Appendix ........................................................................................................ 48 2 Rebel Socialization, Military Training, and Group Cohesion 68 2.1 Introduction .................................................................................................... 69 2.2 Rebel Socialization and Group Cohesion ...................................................... 72 2.2.1 Selective Recruitment ........................................................................... 75 2.2.2 Political Education (PE) ....................................................................... 78 2.2.3 Military Training (MT) ........................................................................ 80 2.3 A Theory of Military Training and Cohesion ................................................ 80 2.3.1 The Internal Structure of Rebel Groups ................................................ 81 2.3.2 How Military Training Builds Group Cohesion? Insights from Military Sociology ....................................................................................................... 84 2.3.3 MT and Horizontal Bonding ................................................................. 87 2.3.4 MT and Vertical Bonding ..................................................................... 89 vi 2.3.5 MT and Institutional Bonding ............................................................... 92 2.4 Hypothesis ...................................................................................................... 94 2.5 Data and Methods .......................................................................................... 94 2.5.1 The Dependent Variable ....................................................................... 95 2.5.2 The Independent Variable ..................................................................... 97 2.5.3 Control Variables and Alternative Explanations ................................... 99 2.6 Results .......................................................................................................... 106 2.7 Predicted Probabilities ................................................................................. 109 2.8 Robustness Checks and Further Analysis .................................................... 110 2.9 Conclusion.................................................................................................... 112 2.10 Appendix .................................................................................................... 114 3 Foreign Fighters and Rebel Fragmentation 121 3.1 Introduction .................................................................................................. 122 3.2 Defining Foreign Fighters (FFs) .................................................................. 124 3.3 FFs in Civil Wars ......................................................................................... 126 3.3.1 The Advantages of FFs ....................................................................... 130 3.3.2 The Disadvantages of FFs ................................................................... 133 3.4 The Argument: Foreign Fighters and Rebel Fragmentation ........................ 136 3.4.1 Reduced Dependence on Local Fighters ............................................. 137 3.4.2 Preference Divergence ........................................................................ 137 3.4.3 Strategic Disagreements ...................................................................... 139 3.4.4 Segregation .......................................................................................... 140 3.5 Hypothesis 1 ................................................................................................. 143 3.6 Hypothesis 2 ................................................................................................. 145 3.7 Data and Methods ........................................................................................ 145 3.7.1 The Dependent Variable ..................................................................... 146 3.7.2 The Independent Variable ................................................................... 147 3.7.3 Control Variables and Alternative Explanations ................................. 150 3.8 Results .......................................................................................................... 155 3.9 Predicted Probabilities ................................................................................. 161 3.10 Selection Effects......................................................................................... 163 3.11 Robustness Checks and Further Analysis .................................................. 166 3.12 Conclusion.................................................................................................. 167 3.13 Appendix .................................................................................................... 172 Conclusion 178 Bibliography 185 vii List of Tables 1.1 List of Fatah factions .............................................................................................. 48 1.2 Data on Fatah Central Committee (FCC) ............................................................... 49 1.3 Data on Fatah factions ............................................................................................ 51 1.4 Descriptive statistic ................................................................................................. 55 2.1 Descriptive statistics ............................................................................................... 95 2.2 Logit regression of MT on rebel fragmentation .................................................... 107 2.3 Cases of rebel fragmentation in the data sample .................................................. 115 2.4 Cross tabulations of the main variables ................................................................ 116 2.5 Annual fixed effects .............................................................................................. 116 2.6 Alternative MT lag structures (t-1 | t-2) ................................................................ 117 2.7 Alternative MT lag structures (t-1 | t-2 | t-3)......................................................... 118 2.8 Forced recruitment ................................................................................................ 119 2.9 Ideologically-motivated rebel groups ................................................................... 120 3.1 Descriptive statistics ............................................................................................. 146 3.2 Logit regression of FFs on rebel fragmentation .................................................... 156 3.3 Logit regression of coethnic FFs on rebel fragmentation ..................................... 158 3.4 Bivariate probit of FFs on rebel fragmentation..................................................... 165 3.5 Cases of rebel fragmentation in the data sample .................................................. 173 3.6 Cross tabulations of the main variables ................................................................ 173 3.7 Annual fixed effects .............................................................................................. 174 3.8 Alternative FFs lag structures (t-1 | t-2) ................................................................ 175 3.9 Alternative FFs lag structures (t-1 | t-2 | t-3)......................................................... 176 3.10 Ideologically-motivated rebel groups ................................................................. 177 viii List of Figures 1.1 Number of new Fatah factions per year .................................................................. 36 1.2 Justifications for factionalization by category ........................................................ 39 2.1 Distribution of rebel fragmentation ........................................................................ 96 2.2 Distribution of rebel MT ......................................................................................... 99 2.3 Predicted probabilities of rebel fragmentation ...................................................... 109 2.4 Alternative presentation of rebel fragmentation over time ................................... 114 3.1 Civil wars with foreign fighters by region ............................................................ 127 3.2 Civil wars with foreign fighters vs. total civil wars .............................................. 128 3.3 Distribution of rebel fragmentation ...................................................................... 147 3.4 Distribution of FFs ................................................................................................ 149 3.5 Distribution of FFs vs. coethnic FFs ..................................................................... 159 3.6 Predicted probabilities of rebel fragmentation ...................................................... 162 3.7 Proportion of rebel fragmentation in a given year over time ................................ 172 ix Introduction Since the end of the Second World War, civil wars have become the leading form of violent conflict, consisting of over 75 percent of armed conflicts (Pearlman & Cunningham, 2012). A majority of the world?s countries have undergone civil wars, which have also had massively destructive side effects with regards to human development and security (Blattman & Miguel, 2010). A history of insurgency reveals that rebel group fragmentation is a recurring phenomenon in intrastate conflict. For example, in 1968 two groups split off from the Popular Front for the Liberation of Palestine (PFLP) ? the Popular Front for the Liberation of Palestine-General Command (PFLP-GC), led by former leaders of the Palestine Liberation Front (PLF), and the Democratic Front for the Liberation of Palestine (DFLP), which consisted of other former members of the PFLP (Sayigh, 1997; Pearlman, 2011). The Irish Republican Army (IRA) in 1969 underwent fragmentation, splitting between the Official and the Provisional IRA, which itself later broke into the Continuity and the Real IRA (Moloney, 2002). The National Patriotic Front of Liberia (NPFL) also underwent two major fragmentation episodes over the course of its conflict with the Liberian government. First, in early 1990 a group of disgruntled commanders broke away from the NPFL and established the Independent National Patriotic Front of Liberia (INPFL). Then, in 1994, a unit of high-ranking NPFL members split from the NPFL and founded the Central Revolutionary Council (CRC-NPFL) in opposition to 1 the NPFL (UCDP - Uppsala Conflict Data Program, n.d.; Lidow, 2016). Most recently, the extreme fragmentation of rebel groups in the ongoing civil wars in the Middle East have sparked growing scholarly and policy interest in the process of fragmentation. Rebel group fragmentation is ?an event in which a segment of a rebel organization formally and collectively exits the existing rebel organization and establishes a new, independent rebel organization? (Doctor, 2020, p. 599). Why rebel groups undergo fragmentation is a puzzling question. Forming and maintaining a new rebel group is a costly, risky, and uncertain endeavor (Lichbach, 1995; Gurr, 2000; Byman et al., 2001; Freeman, 2012; Seymour, 2014; Clarke, 2015; Doctor, 2020). Groups already operate under resource constraints, and fragmentation leads to more inter-rebel competition and fighting over scare resources that could otherwise be devoted to the war with the government (Fjelde & Nilsson, 2012). Given that the conflict between the rebels and the government is so vital, why do rebels split? What makes rebel group fragmentation more likely to occur? Understanding why rebel groups fragment is key to understanding their internal dynamics and trajectory and, hence, intrinsically relevant to and consequential for both academics and policy-makers. Nevertheless, while extensive research about the consequences of rebel fragmentation exists (cf. Driscoll, 2012; K. Cunningham et al., 2012; K. Cunningham, 2013; Findley & Rudloff, 2012; Rudloff & Findley, 2016), research on the process of fragmentation remains relatively nascent. This dissertation contributes to a growing body of research on the causes of rebel fragmentation. In the chapters that follow I develop and evaluate three distinct theories on the causes of rebel fragmentation, elucidate multiple mechanisms by which rebel 2 groups splinter, generate original findings, and recommend several policies for governments fighting insurgents domestically and for international actors interested in conflict resolution. Chapter 1 examines faction emergence in a rebel group with a centralized leadership and decision-making structure by developing a junior cadres-based explanation of fragmentation. Departing from studies that treat junior cadres as actors who, for the most part, respond to external stimuli, (cf. Tamm, 2016; Ives, 2019; Nagel & Doctor, 2020), I contend that junior cadres are autonomous actors who have the agency (Jasper, 2004; Shapiro & Siegel, 2007; Shapiro & Siegel, 2012; Shapiro, 2013; Abrahms & Potter, 2015) to decide their present and future affiliation with the rebel group. I develop a theory on political exclusion, junior cadres? grievances, and factionalization. I argue that in a centralized rebel group, factions will emerge when the ability of junior cadres to advance in the organization through peaceful mechanisms is systematically blocked by the senior leadership. Excessive control and limited leadership turnover block junior cadres? access to leadership positions and, consequently, restrict their ability to affect the organization from within. Junior cadres who want to influence the organization?s politics therefore face a dilemma: remain within the rebel group, albeit without real prospects to access senior leadership positions, or exit it. Factionalizing is a way to redress grievances for junior cadres who perceive existing mechanisms for upward mobility ineffective. By investigating the motivations of junior cadres for exiting their organization I demonstrate how the 3 systemic exclusion of junior cadres from senior decision-making positions precipitates fragmentation even in rebel groups with a centralized leadership structure. The approach adopted in Chapter 1 is based on multiple studies that demonstrate how motivation and opportunity explanations for political action operate complementarily rather than competitively (Bara, 2014; Buhaug et al., 2008; D. Cunningham et al., 2017; also see Zaks, 2017). Specifically, the combination between junior cadres? grievances (motivation) and low political access to executive power (opportunity) is associated with a higher propensity for factionalization. I evaluate my theory through the case of Palestinian Fatah. I begin by establishing Fatah as an organization with a centralized leadership structure. Next, I show how Fatah experienced multiple instances of factionalization despite its centralized decision-making structure. I then proceed to evaluate the determinants of factionalization in Fatah. I show how growing centralization in decision-making and systemic exclusion of lower-tier members from senior leadership positions through existing mechanisms for upward mobility create ideal conditions for factionalization. Finally, I discuss the limitations of several alternative explanations to Fatah?s factionalization, namely access to arms and resources, interorganizational competition, and state repression. Three key findings in support of my theory emerge. First, a centralized decision- making structure does not immunize a rebel group to factionalization. This finding challenges the prevailing scholarly notion that rebel groups with centralized leadership have better mechanisms for inhibiting fragmentation (cf. Pearlman, 2011; Asal et al., 2012; Shapiro, 2013). Second, I find that in the majority of cases, factions are driven by 4 disenfranchised junior cadres and not by disagreements at the leadership level, as current literature argues (cf. Bakke et al., 2012; K. Cunningham et al., 2012; K. Cunningham, 2013; Plank, 2017). Third, I show that political opportunity structure and motivation work complimentarily to explain faction emergence. As I have continued researching the causes of rebel fragmentation, I realized that existing studies on fragmentation presuppose, albeit tacitly, that cohesion is the baseline for how we understand the internal dynamics of rebel groups. Rebel groups, in other words, begin as relatively cohesive organizations which, over the course of a conflict, deviate from cohesion. The relative rarity of rebel fragmentation1, however, suggests that it might be more fitting to ask, ?What do rebel groups do to prevent fragmentation and foster internal cohesion?? Chapter 2 brings forth cohesion ? as opposed to fragmentation ? as the puzzle in need of explanation. I propose that the solution to that puzzle is located in rebel socialization ? the different methods that rebel groups employ in order to socialize fighters into the organization and in order to generate a deeper change in one?s sense of self (Hoover Green, 2016, 2018; Gates, 2017; Cohen, 2017; Checkel, 2017). Although recent literature has begun to point towards the effects of rebel socialization on member retention and cohesion (Oppenheim et al., 2015; Forney, 2015; Gates, 2017; Cohen, 2017; Nagel & Doctor, 2020), the assertion that rebel socialization increases group cohesion and hinders fragmentation has not been tested directly. 1 For example, in the data used in Chapter 2, fragmentation occurred only 27 times, or in 4.5 percent of the observations. In the data used in Chapter 3, fragmentation occurred 20 times, or in 2.3 percent of the observations. 5 To fill this gap, I synthesize insights from existing literature on military sociology, military training (MT), and rebel group cohesion to develop a theory of MT and insurgents? cohesion. I argue that, compared to other forms of rebel socialization ? specifically, selective recruitment and political education ? MT is uniquely positioned to foster cohesion. Specifically, MT generates personal and professional bonding on all three levels of the rebel group: among combatants; between combatants and commanders at the combat unit level; and members? bonds to the overall organizational mission, as generated by group leaders. Combined, these bonds make splintering more costly. When combatants are socially and professionally tied to one another, the costs of exiting the group increases, and fragmentation is less likely. Estimating a series of binary logistic regression models with a sample of 83 rebel groups active in 47 armed intrastate conflicts between 1989 and 2010, I find that rebel groups that conduct MT are significantly less likely to split into multiple groups. In Chapter 3, I assess the effect of foreign fighters (FFs) on rebel group fragmentation. For the most part, academic research and policy have centered on the high-lethality attacks associated with FFs (Cordesman, 2005; Rich & Conduit, 2015) or the danger FFs pose when they return to their home countries, especially in the West (Bergen & Reynolds, 2005; Noonan & Khalil, 2014; Lindekilde et al., 2016; Barrett, 2017; A. Braithwaite & Chu, 2018; Malet & Hayes, 2020). Far less attention, however, has been given to questions about FFs? influence in insurgencies. Existing work presupposes that FFs present a tradeoff for the rebel group. On the one hand, FFs buttress the capabilities of domestic rebel groups through additional manpower, resources, and know-how (Bakke, 2014; Doctor, 2021). FFs are found to be 6 associated with a decreased likelihood of government victory and help insurgents achieve a negotiated settlement (Chu & Braithwaite, 2017). On the other hand, FFs disrupt the group?s internal dynamic, precipitate deep divisions, and pose a threat to the group?s internal cohesion (Bakke, 2014; Rich & Conduit, 2015; Doctor, 2021). In Chapter 3, I evaluate the assertion that the presence of FFs increases the probability of group fragmentation. I introduce an argument connecting FFs and fragmentation and highlight mechanisms that link the two. I utilize insights from an array of studies and cases on FFs in order to explain how FFs disrupt the internal dynamics of rebel groups to the point of fragmentation. I start by exploring how reduced group dependency on local fighters, preference divergence, strategic disagreements, and member segregation increase the likelihood of fragmentation for rebel groups that recruit FFs. I hypothesize that insurgent groups that recruit FFs will be more likely to fragment than groups with a predominantly local membership base. Next, I posit that if the foreignness of FFs in relation to local insurgents makes fragmentation more likely, than rebel groups that recruit coethnic FFs will be less likely to experience fragmentation. I test these hypotheses on a global sample of 227 rebel groups active between 1989 and 2011. The findings support the argument that rebel groups that recruit FFs are significantly more likely to fragment, even after accounting for the endogenous choice of rebel groups to recruit FFs. Against my expectations, I find that the recruitment of coethnic FFs does not diminish the probability of fragmentation. This finding raises questions about the value of ethnic homogeneity in the context of FFs and implies that fighters? ?foreignness? remains a decisive factor in trying to predict fragmentation. 7 Overall, this dissertation suggests that the internal dynamics of rebel groups are tightly associated with their trajectory. For all rebel groups, the risk of fragmentation is omnipresent. Creating a coherent unit is a challenge that all rebel groups face (Cohen, 2017; Nagel, 2021). Because rebel group fragmentation increases the likelihood of civil war incidence (K. Cunningham, 2013), shapes the intensity and duration of fighting (D. Cunningham, 2006; K. Cunningham et al., 2012; Findley & Rudloff, 2012), and the likelihood and durability of peace (Nilsson, 2010; Driscoll, 2012; Rudloff & Findley, 2016; K. Cunningham, 2017), it might be more effective to craft policies which are based on the causes, as opposed to the consequences, of fragmentation. Understanding why insurgent organizations fragment can help scholars predict those outcomes and help policymakers prevent them. 8 Factionalization From Below: The Case of 1 Palestinian Fatah Abstract This article examines the role of junior cadres in rebel group fragmentation. I argue that in a centralized rebel group factions will emerge when leaders block junior cadres? access to senior decision-making bodies. Junior cadres who want to influence the organization?s politics therefore face a choice between remaining within the rebel group and exiting it. Factionalizing is a way to redress grievances by aggrieved junior cadres who deem peaceful mechanisms for upward mobility ineffective. Using original datasets and personal interviews, I find strong evidence supporting my argument in the case of Palestinian Fatah. A published version of this chapter may be found at: Stern, M. (2022). Factionalization from Below: The Case of Palestinian Fatah. Civil Wars, 1-46. DOI: 10.1080/13698249.2022.2015194. 9 1.1 Introduction Researchers and policy makers have become increasingly interested in why rebel groups fragment. Greater numbers of rebel groups in civil wars directly affects conflicts? processes and outcomes (D. Cunningham, 2006; Sinno, 2011; Pearlman, 2011; Pearlman, & Cunningham, 2012; Findley & Rudloff, 2012; Bakke et al., 2012; Driscoll, 2012; K. Cunningham, 2013, 2014). Splinter factions often resort to violence, a costly and highly visible signal with lethal consequences for all warring sides (K. Cunningham et al., 2012; Breslawski & Ives, 2018). While extensive research about the consequences of rebel group fragmentation exists, we do not fully understand the process. Some scholars have argued that disagreements within the leadership structure causes splintering of factions from the organization. They contend that rebel groups with centralized leadership may have better mechanisms for maintaining internal cohesion (Pearlman, 2011; Asal et al., 2012; Shapiro, 2013). If researchers are correct, a centralized structure 2 should reduce factionalization.3 But this is not always the case. Fatah, the most historically prominent Palestinian rebel group, has a centralized leadership and decision-making structure coalesced in a central committee, which should have reduced the likelihood of factions. Yet, numerous factions have emerged with conflicting views on the best way to advance Fatah and Palestinian goals (Miller, 1983). How can researchers account for Fatah?s 2 I use the terms ?centralized leadership? and ?centralized decision-making? structure interchangeably. Having a centralized leadership inevitably means that decision-making is centralized and vice versa. 3 Various studies describe the term ?factionalization? as ?fragmentation? or ?splintering? and I use the terms interchangeably. 10 multiple instances of factionalization despite the rebel group?s centralized leadership structure? Although the distinction between centralized and decentralized leadership structure incorporates an important factor in the political processes which lead to factionalization, it discounts important dynamics which play out at lower levels of the organization. Specifically, in this article I argue that in rebel groups with a centralized leadership, factions are driven by disenfranchised junior cadres and not by disagreements among leadership as current literature argues. I argue that in a centralized rebel group factions will emerge when the ability of junior cadres to advance in the organization through peaceful mechanisms for upward mobility is systematically blocked by the senior leadership. Excessive control and limited leadership turnover block junior cadres? access to leadership positions and, consequently, restricts their ability to affect the organization from within. Junior cadres who want to influence the organization?s politics are therefore facing a dilemma: remaining within the rebel group, albeit without real prospects to access senior leadership positions, and exiting it. Factionalizing is a way to redress grievances by junior cadres who perceive existing mechanisms for upward mobility ineffective. The approach adopted in this article is based on multiple studies that demonstrate how motivation and opportunity explanations for political action operate complementarily rather than competitively (Bara, 2014; Buhaug et al., 2008; D. Cunningham et al., 2017; also see Zaks, 2017). Specifically, the combination between junior cadres? grievances and low political access to executive power is associated with a higher propensity for factionalization. 11 In Fatah, junior cadres ?represented a widespread grass-roots phenomenon? (Cobban, 1984) which have historically contributed to opposition to Fatah?s leadership (Sayigh, 1997). As this article demonstrates, Fatah?s junior cadres? justifications for factionalizing show that variation in the emergence of cadre-driven factions was a function of their respective ability to advance in the organization through peaceful mechanisms and procedures for upward mobility. When existing channels of political influence were blocked, factionalization increased. When junior cadres perceived existing political channels as effectively facilitating upward mobility, factionalization decreased. Using two original datasets, primary and secondary sources in English, Arabic and Hebrew, and personal interviews with Fatah junior cadres, this study contributes to literature studying rebel group fragmentation and the understanding of how and why rebel groups form (Staniland, 2014; Lewis, 2017; Larson & Lewis, 2018; Walter, 2019; Lewis, 2020; J. Braithwaite & Cunningham, 2020). Furthermore, whereas existing studies focus on conflict-level factors or group attributes to explain rebel fragmentation (e.g., Christia, 2012; Staniland, 2012; Doctor, 2020; Nagel & Doctor, 2020), this study explores how access to senior leadership positions within rebel groups shape the behavior of junior cadres in ways that promote factionalization. This article is structured as follows. First, I review the relevant literature with a focus on existing explanations for rebel group fragmentation. Second, I discuss the theory of junior-cadre lead fragmentation, followed by a description of data and methods. The final sections present empirical analysis and findings to discuss their implications for future research and policy. 12 1.2 Existing Explanations of Rebel Group Fragmentation Rebel group fragmentation presents a puzzle. Some studies show that fragmented rebel groups can offer incentives to the government to make strategic concessions to moderate factions, thereby making fragmentation advantageous to rebels (K. Cunningham, 2011; Driscoll, 2012). Others agree that fragmentation is both destructive and resource consuming. A higher number of rebel groups generate greater commitment and information problems thereby increasing the likelihood of civil war, incidence (K. Cunningham, 2013), and duration (D. Cunningham, 2006). More cohesive rebel groups are better structured to resolve internal disputes and direct a common strategy, while less cohesive ones are inflicted by infighting and spoiling that prevent strategic progress (Pearlman, 2011). Forming and maintaining a new rebel group is a costly, risky, and uncertain endeavor (Lichbach, 1995; Gurr, 2000; Byman et al., 2001; Freeman, 2012; Seymour, 2014; Clarke, 2015; Doctor, 2020). Splits also weaken the rebel group, whereas unity strengthens them (Staniland, 2014). Operating under resource constraints, fragmentation leads to more inter-rebel competition and fighting over scare resources that could otherwise be devoted to the war with the government (Fjelde & Nilsson, 2012). Given that the conflict between the rebels and the government is so vital, why do rebels split? What makes rebel groups? fragmentation more likely to occur? To address these questions, a growing number of researchers have turned to investigate the causes of rebel groups? cohesion and fragmentation. First, existing studies find that rebel groups formed around shared identity have advantages which promote cohesion whereas ideological differences are often 13 associated with fragmentation (Stedman, 1997; Seymour et al., 2016; Gade et al., 2019). Ideology, for example, specifies goals, objectives, and norms against which members can assess their leaders and organizational discipline promoted (Lichbach, 1995; Weinstein, 2007; Christia, 2012; Hoover Green, 2016, 2018). Ideology can also help leaders screen for the most committed-to-the-cause recruits thereby reducing the likelihood of future defections (Weinstein, 2007; Oppenheim et al., 2015). However, considering the stress of war, converting ideological convictions, as appealing as they might be, into cohesive rebel groups is a particularly challenging task (Staniland, 2014). Moreover, because in multiparty conflicts, several rebel groups often claim to represent the same constituency (Bloom, 2004, 2005), they might end up fighting each other to become the sole representative of that constituency (Fjelde & Nilsson, 2012). Hence, identity-centric arguments point in opposite directions, and it is difficult to extrapolate the impact of ideology on the patterns of rebel group fragmentation. Second, competitive dynamics which characterize civil conflict are key determinants of shaping fragmentation (Seymour et al., 2016). For example, splits are more likely to occur when extremist and moderate factions reach an internal impasse about whether to continue fighting or to negotiate a peace agreement (Stedman, 1997; Kydd & Walter, 2002; Findley & Rudloff, 2012; Duursma & Fliervoet, 2021). Factions may also emerge when inter-rebel competition over popular support intensifies. For instance, competition between Palestinian groups during the second Intifada drove members of existing organizations to factionalize and use suicide attacks in pursue of public support and recruits (Bloom, 2004). Conversely, inter-rebel competition can benefit rebels by highlighting differences between moderates and extremists factions 14 thus encouraging the government to make concessions to the more moderate ones (K. Cunningham, 2011). Third, state repression increases the costs of mobilization and provokes internal competition and splits among rival factions vying for power (Seymour et al., 2016). Sustained government counterinsurgency campaigns could lead protection-seeking rebels (Staniland, 2012) to collectively fragment in order to survive in the wake of intensified battlefield losses (Christia, 2012). However, the degree to which state repression is associated with fragmentation may be conditioned on the level of members? satisfaction with institutional arrangements before the onset of repression. When rebels are relatively satisfied with the institutional arrangements of their organization, repression will promote cohesion. When rebels are dissatisfied with their organization?s existing arrangements, repression will present an opportunity to act on that opposition; potentially leaving the rebel group more divided than beforehand (McLauchlin & Pearlman, 2012). Fourth, external support can generate both rebel unity and division. According to Carter (2012), rebel groups that receive support from foreign sponsors are less likely to internally dissolve because they are better able to compensate their members with selective benefits, which alleviates collective action problems and aids group cohesion (Carter, 2012). Other scholars, however, argue that external support introduces exogenous interests to the conflict and catalyzes intragroup divisive tendencies (Seymour, 2014; Tamm, 2016; Seymour et al., 2016; Walther & Pedersen, 2020). For example, when foreign sponsors distribute resources in order to maintain an imbalance of power in favor of the leader, the group is less likely to fragment. However, when 15 external support contributes to the balance of power between leaders and subcommanders it effectively undermines the leadership?s control, making the rebel group more likely to split (Tamm, 2016; Lidow, 2016). This is the case when multiple state sponsors with opposing political agendas seek to foster their own proxy clients through patronage. For instance, rebels in the Syrian civil war insist that chaos in the sources and distribution of money to the rebellion is the gravest cause of disunity within its ranks (Pearlman, 2013). Ives (2019) develops these theories by conditioning the impact of external support on fragmentation on ethnic links between the supporter and the rebel group. Specifically, when a rebel group receives support from an external actor with whom it shares ethnic linkages, commanders are more likely to perceive the preferences of the external supporter as similar. Consequently, intragroup trust between the commanders and the leaders increases, and fragmentation is less likely. Finally, scholars emphasize the relationship between group structure and fragmentation. According to Sinno (2011, p. 93), ?the management of intense factionalism is one of the more important considerations driving the structural choices of organizational leaders.? Preference divergence between leaders and subordinates (Shapiro, 2013; Abrahms & Potter, 2015) may result in acts of insubordination which increase the likelihood of defections and splits (Shapiro, 2013; Warren & Troy, 2015; Woldemariam, 2018). Hence, multiple studies argue that higher levels of centralization of command and decision-making reduce fragmentation (Asal et al., 2012; Hazen, 2013; Staniland, 2014; Doctor, 2020). Centralized rebel groups are better able to screen for recruits who will adhere to the group?s objectives and follow its leadership (Weinstein, 2007; Burch 16 & Ochreiter, 2020). The centralization of decision-making also means that leaders delegate less autonomy to junior cadres (Arquilla & Karasik, 1999; Stepanova, 2008; Shapiro, 2013; Abrahms & Potter, 2015). As a result, centralized rebel groups are better at monitoring, detecting, rewarding and punishing deviant members, inhibiting undesired outcomes (Weinstein, 2007; Shapiro, 2013; Abrahms & Potter, 2015; Hoover Green, 2016, 2018), and increasing the costs of splitting (Christia, 2012; Burch & Ochreiter, 2020; Nagel & Doctor, 2020). When fragmentation occurs, it is often preceded by rivalries between rebel leaders (Pearlman, 2011; Bakke et al., 2012; K. Cunningham et al., 2012; K. Cunningham, 2013, 2017; Plank, 2017). Specifically, competition between leaders allow for a plurality of potentially competing opinions, objectives, and priorities, making rebel groups more vulnerable to breaking apart under external stressors (Asal et al., 2012). Similarly, struggles for mobilization between leaders increase the likelihood of splits (Warren & Troy, 2015). Recently, Joo and Mukherjee (2021) demonstrated that while highly centralized rebel groups are better at fostering cohesion in their formative years, they become more vulnerable to fragmentation over time as intensified inter- leader competition over power drives alienated leaders to split. In sum, whereas existing explanations shed important light on the causes of rebel group fragmentation, there seems to be scholarly disagreement. Fragmentation studies often point in opposite directions, making it difficult to deduce theoretically the impact of each explanation on the causes of rebel groups? fragmentation. More specifically, existing scholarship suggests that rebel groups characterized by decentralized leadership and internal disagreements are more likely to factionalize amid growing external 17 pressures. Conversely, organizations with centralized leadership may have better mechanisms for maintaining internal cohesion and hence are more resistant to factionalization as a consequence of leadership discord. What can explain faction emergence in a rebel group with a centralized leadership structure like Fatah? Although differences in group attributes, leadership structure and external factors incorporate important variables in the political processes which lead to factionalization, they discount a key social stratum that drives faction emergence: the junior cadres. Studies that do incorporate junior cadres into their analysis treat them as actors who respond to external stimuli, ranging from external support (Tamm, 2016; Ives, 2019) to sexual violence (Nagel & Doctor, 2020). As such existing research insufficiently exploits junior cadres? agency. I offer a different approach. In trying to explain rebel group factionalization, I propose a junior cadres-based explanation of fragmentation. I contend that junior cadres are not actors who reflexively respond to external stimuli. Alternatively, junior cadres are autonomous actors who have the agency (Jasper, 2004; Shapiro & Siegel, 2007, 2012; Shapiro, 2013; Abrahms & Potter, 2015) to decide on their present and future affiliation with the rebel group. By investigating into the motivations of junior cadres to exit their organization I demonstrate how the systemic exclusion of junior cadres from senior decision-making positions precipitates fragmentation even in rebel group with a centralized leadership structure. 18 1.3 The Argument: Political Exclusion, Junior Cadres? Grievances, and Factionalization What explains Fatah?s multiple instances of factionalization despite the group?s centralized leadership? I argue that rebel group factionalization depends on a combination between the ability of junior cadres to access the organization?s highest decision-making bodies and their subsequent grievances against their leaders, who systematically block their access to senior leadership positions. Existing studies find support for the relationship between political and economic grievances and civil conflict (Gurr, 1970; Collier & Hoeffler, 1999; Collier & Hoeffler, 2002; Emirbayer & Goldberg, 2005; Humphreys & Weinstein, 2008; F. Stewart, 2008; Cederman et al., 2011; Dyrstad & Hillesund, 2020). Systemic political and economic exclusion motivate individuals to factionalize in order to rectify, what they consider, injustice. Considering the limitations of grievances to independently explain political action (Snyder & Tilly, 1972), opportunity structure theories emphasize the role of specific changes or events that may provide windows of opportunity for protesters in achieving the collective action necessary for political action (Fearon & Laitin, 2003; Collier & Hoeffler, 2004; Meyer & Minkoff, 2004; Gleditsch & Ruggeri, 2010). For example, the availability of resources, such as arms (Seymour, 2014) and funds (Collier & Hoeffler, 2004; Seymour, 2014), might explain factionalization because, unlike grievances, they generate tangible incentives for members to factionalize. Geographical concentration can foster cohesion because it facilitates interaction between group members which helps the group to collectively organize (Weidmann, 2009). 19 The difficulty to clearly operationalize political opportunity theories (Meyer & Minkoff, 2004) sometimes mean that they cannot easily be generalized to other settings, even within the same case. Moreover, differences across factions protesting against the same leadership across multiple structural circumstances suggest that the explanatory power political opportunity theories are limited. To address these shortcomings, I follow scholarship that illustrates how motivation and opportunity explanations for political violence complement rather than rival one another (Fuchs, 2001; Bara, 2014; Buhaug et al., 2008; D. Cunningham et al., 2017; Dyrstad & Hillesund, 2020), I argue that, ceteris paribus, a centralized rebel group will produce factions when the ability of junior cadres to advance in the organization through peaceful mechanisms and procedures for upward mobility is systematically blocked by low turnover in senior leadership. Political exclusion fosters a sense of groupness (Coser, 1957) among politically and economically disenfranchised junior cadres. Processes of social comparison and intergroup evaluation of junior cadres vis-?-vis their leaders breed political and material grievances (Gurr, 1970; Muller & Seligson, 1987; Cederman et al., 2011), undermine trust in the effectiveness of ordinary political channels for upward mobilization, and increase support for political action (Dyrstad & Hillesund, 2020). Junior cadres who want to influence the organization?s politics are therefore facing a choice between remaining within the rebel group, albeit without real prospects to access senior leadership positions, and exiting it. Factionalizing is a way to redress grievances by junior cadres who deem peaceful mechanisms for upward mobility ineffective. However, the decision of junior cadres who decide to factionalize involves a tradeoff. 20 Specifically, the new faction is not nearly as influential as a unitary rebel group would have been. I define a faction as a clear group of individuals within a rebel group who communicate aims or goals (K. Cunningham, 2013). These goals are characterized by the prioritization of those individuals? preferences, worldviews, and strategy over those of the current leadership (K. Cunningham et al., 2012; K. Cunningham, 2013; Breslawski & Ives, 2018). That is, for the rebel group a faction may advocate a change of leadership, strategy, ideology and so forth. For its members, a faction pursues increased benefits, voice (Pearlman, 2011), and opportunities within the rebel group; potentially at the expense of the current leadership or other factions in it. Factions often pursue the same goal on behalf of the same constituency (K. Cunningham, 2013). For example, claiming to represent the Palestinian people, all Fatah factions have sought to establish an independent Palestinian state. Nevertheless, Fatah factions have occasionally varied in terms of the mechanisms to achieve that goal. The Black September Organization (BSO) and Abu Nidal Organization (ANO) have engaged in international terrorism; Tanzim, the al-Aqsa Martyrs? Brigades (AMB), and the Popular Resistance Committees (PRC) have targeted Israeli civilian and soldiers as well as rival Palestinians; and al-Mustaqbal was a nonviolent political opposition to Fatah?s leadership. Factions may also vary in terms of the character of the final objective. For instance, while most Fatah factions have advocated for the establishment of a secular Palestinian state, a few factions, such as the PRC, have adopted a radical Islamist ideology and framed the struggle against Israel as a holy war (?The Popular Resistance 21 Committees,? 2011). Importantly, a faction is distinguished from individual defections. Although factionalization and defections may occur simultaneously (Ives, 2019, p. 10), defections refer to individuals who join competing groups or the government (Kalyvas, 2008, p. 3; Staniland, 2012; Shapiro, 2013; Seymour, 2014), while factions consist of a group of individuals who collectively challenge the incumbent leadership, often by exiting the group altogether to create a new rebel group (Christia, 2012; Doctor, 2020; Nagel & Doctor, 2020). Finally, for different methodological and data-related reasons, researchers have restricted organizational splits to major group splintering resulting in the establishment of a newly independent rebel group (Lidow, 2016; Tamm, 2016; Ives, 2019; Doctor, 2020; Nagel & Doctor, 2020; Duursma & Fliervoet, 2021). Smaller, weaker, less violent or nonviolent groups are excluded. In this study I take a slightly different approach. To better understand the deeper political processes of rebel group factionalization, I account for all Fatah factions regardless of their size, durability, use of violence, and their degree of splitting from the organization. Table 1.1 in the appendix lists all factions that emerged from Fatah and their respective year of emergence. Fatah?s factions vary considerably in composition. Some were dominated by military commanders (e.g., Fatah Uprising, ?Atallah ?Atallah?s Supreme Military Council) while others by civilian administration (e.g., the Maoist Tendency and the Vietnamese Line) or a mix of both (e.g., the democratic direction, AMB). This makes the disaggregation of factions according to the position of their members in Fatah?s hierarchy extremely convoluted. One common thread throughout the majority of Fatah?s 22 factions, however, is that their members were most often subordinates to the senior leadership. I collectively define them as junior cadres. In a rebel group, junior cadres are below leaders in the organization hierarchy (Shapiro & Siegel, 2007; Doctor, 2020; Nagel & Doctor, 2020). They are often members or ?in charge of subgroups, which are potentially independent groups in the making? (Ives, 2019). Fatah?s junior cadres were often a ?subordinate middle command used to obeying orders? from the senior leadership and were, overall, ?far from the political limelight, and seemingly devoid of political ambitions of his own? (Frisch, 1993). Although Fatah?s junior cadres are often younger than senior leaders, the category ?junior cadres? refers to members who are not part of Fatah?s senior leadership and decision-making echelon. According to a Fatah junior cadre, ?the main difference between the senior and junior leadership is not age,? but rather each stratum?s ?background and position in the organization. We had good positions in Fatah but ultimately it was all about the senior leaders? decisions? (personal communication, March 4, 2020). Whereas junior cadres exist in all rebel groups (Haer et al., 2011; Shapiro, 2013; Abrahms & Potter, 2015; Ives, 2019; Doctor, 2020; Nagel & Doctor, 2020), not all fragment. Considering the heavy costs associated with creating a new rebel group (Lichbach, 1995; Gurr, 2000; Byman et al., 2001; Freeman, 2012; Clarke, 2015), factionalization should be an unlikely response to junior cadres who believe they can access influential positions within the group through existing mechanisms of participation (Gurr, 1970; Gates et al., 2006; Hegre, 2014).4 Similar to nonviolent 4 Huntington (1968) has famously argued that the absence of mobility opportunities ?channels energies into politically deviant behavior? such as corruption. (Huntington, 1968). 23 organizations, rebel groups use promotion to increase members? performance, satisfaction, mutual trust, and compliance (Kosteas, 2011; Haer et al., 2011; Hays & Bendersky, 2015). When junior cadres feel that they are assessed fairly, they perceive promotion as a sign of trust on the part of their leaders and reciprocate with higher commitment (Haer et al., 2011; Haer & Banholzer, 2015). Conversely, lower-level personnel, if not promoted, become more likely to defect from the rebel group (P. Berman, 1974). Importantly, the ability of individuals to ascend to senior leadership positions through elections demonstrates broad support for the new rebel leader and the group and sends a strong signal of group cohesion (K. Cunningham & Sawyer, 2019). However, the existence of conventional political channels for change and participation are in and of themselves insufficient to ensure access to political power nor peaceful political change (Dyrstad & Hillesund, 2020). Even in a seemingly competitive political environment incumbents routinely abuse resources, predetermine opposition candidates, harass opposition supporters, and manipulate electoral results in order to keep their grip on power (Levitsky & Way, 2002). Elections often serve as a smokescreen to democracy both within sovereign states (Levitsky & Way, 2010) and rebel groups (Breslawski, 2018; K. Cunningham et al., 2021). Hence, as it happens, the existence of democratic institutions is not necessarily associated with real opportunities for political change nor does it warrant the promotion of individuals and groups to senior leadership positions (Vogt et al., 2015). Thus, my expectation is that in a rebel group with a centralized leadership and low turnover in senior leadership, like Fatah, factions will emerge when junior cadres find existing channels of political influence and upward mobility flawed or blocked. 24 Conversely, junior cadres are less likely to factionalize if they believe existing political channels effectively facilitate upward mobility. 1.4 The Data To test my arguments, I created two datasets on Fatah. The first dataset, FCC 2016, is yearly data that records membership composition changes in FCC between 1967 and 2016. Each observation is a year. The data includes 50 observations and 24 variables (see Table 1.2 in the Appendix for a summary). I collected information about FCC membership using multiple sources in English and Arabic. Specifically, I surveyed history accounts about Fatah, press releases, and academic and policy papers that list FCC membership and change between 1967 and 2016, the last year Fatah held elections to its governing institutions. The second dataset is a faction-year dataset that records all Fatah factions since the group?s foundation in 1959 up to 2019 (see Table 1.3 in the Appendix for a summary). The dataset provides detailed information on each faction. Because in several instances multiple factions emerged at the same year, that same year appears multiple times. For example, three factions - AMB, Tanzim, and PRC - emerged in 2000. Therefore, there are three entries for 2000. Consequently, the data includes 71 observations years. The dataset records a total of 29 different factions. The analysis presented in this study focuses on the source (i.e., junior cadres) and reasons for factionalization, not the timing of faction emergence. I operationalize a faction if a group of Fatah members communicate clear ambitions to amend the organization?s program. More specifically, a group of Fatah members is considered a faction if they express a need to change Fatah?s leadership, 25 ideology, political, strategic, or religious program, advocate institutional reform, call for the redistribution of economic benefits, and, crucially, adopt or referred to with a clear name. I gathered information about Fatah factions using a rich variety of primary and secondary sources in English, Arabic and Hebrew. I used Factiva and open electronic sources to survey newspapers, press statements about the faction, intelligence and policy reports, press statements and interviews made by faction leaders themselves, as well as factions? websites or academic articles and historical accounts about the faction. I manually analyzed each source to determine a set of 15 possible justifications (r_1 to r_15) for each faction?s emergence. The list of justifications also includes existing explanations for the causes of fragmentation (e.g., inter-leadership disputes). All data was coded manually. For example, if the reason for Fatah members to factionalize involved disagreements regarding the use of violence, it is coded 1 in the Armed Struggle variable (r_4) and 0 otherwise. In cases where multiple justifications were identified, they were all coded accordingly. To determine the relative weight of each justification for faction emergence to all other justifications, I first added the number of times each justification (r_n) appeared in the dataset and then divided the sum of each justification by 132, which is the sum of all justifications (see Table 1.4 in the Appendix for a summary). I operationalize junior cadres as any Fatah member who was not a member of the senior leadership, coalesced in FCC, at the time of the faction?s emergence. To measure whether junior cadres drove factions, I traced key individuals in each faction by name and checked whether or not they were members of FCC during the year of the faction?s emergence. A faction is considered driven by junior cadres if none of its 26 members is also an FCC member at the time of the faction?s emergence. A faction is driven by leaders if at least one of its members was also an FCC member at the time of the faction?s emergence. For example, neither Muhammad Abdel-Ghaffar nor Sabri al-Banna, leaders of the Arab Revolutionary Brigades and the ANO, respectively, were FCC members when they factionalized. In contrast, two members of FCC, Muhammad Dahlan and Jibril Rajoub, were also members of the al-Mustaqbal faction when it emerged in 2005 (Yaghi, 2009; B2. Election Results for the Fatah Central Committee, 2009). The al- Mustaqbal faction is therefore considered a faction not driven by junior cadres while the Arab Revolutionary Brigades and ANO are.5 Of Fatah?s 29 factions (1959-2019), 25 (or 86%) were driven by cadres and 4 (or 4%) by leaders. Finally, I conducted five in depth, several-hours-long, semi-structured personal interviews with five Fatah junior cadres who were leaders and combatants in three different factions which emerged in the 2000s: Tanzim, AMB, and al-Mustaqbal.6 All interviews were conducted in the Palestinian Authority (PA) in the West Bank. As in 5 That being said, all members of al-Mustaqbal, including the two who were also members of FCC when the faction emerged, are considered to be part of Fatah?s younger leadership. Insights I gained through interviews with several members of al-Mustaqbal point to evidence in support of my argument. 6 In the process of scheduling the interviews, I faced several obstacles due to political events and the outbreak of COVID-19. Specifically, some of the prospective interviewees were arrested by Palestinian or Israeli security forces just days before a scheduled interview. In addition, timing the interviews was particularly challenging. Unrest increased in Palestine following the release of the Trump administration?s policies towards the PA, as well as Israeli announcements about its intention to annex parts of the West Bank. Finally, during my fieldwork the COVID-19 pandemic broke out and travel between Israel and Palestine was restricted. Amid growing uncertainty, I had to leave the region earlier than planned. Combined, these factors inhibited my ability to gather a larger sample of interviewees. 27 the practice of semi-structured interviews, I refrained from strictly following a formal list of questions. Instead, interviews were based on open-ended questions and allowed for a discussion with the interviewee rather than a straightforward question and answer format (Galletta, 2013), even if that resulted in not all question being asked, requesting for further clarifications, or asking additional new questions. However, the strength of the evidence gathered through these interviews is limited for two main reasons. First, interviews included members of only three Fatah factions, which operated during a relatively short episode in Fatah?s history. Second, junior cadres who led a faction have an incentive to misrepresent their true motivations by holding their reasons for factionalizing private (Weinstein, 2007; Humphreys & Weinstein, 2008). Junior cadres may, for example, emphasize grievances and portray their actions in altruistic terms (Hoffman, 2017), while downplaying other factors affecting their decision to factionalize. Therefore, interviews were neither designed nor capable to give insight to every faction that emerged. Alternatively, interviews? purpose was to gain understanding of broader mechanisms of factionalization, not to validate every coding decision made. Consequently, the descriptive statistics presented in this study are not based on data collected through interviews. 1.5 Assessing the Argument 1.5.1 Fatah?s Centralized Leadership Fatah?s centralized leadership structure should have reduced the likelihood of factionalization. However, Fatah deviates from existing theories (Levy, 2008) regarding the drivers of rebel group factionalization. As early as five years after Fatah?s 1959 founding, intra-Fatah factions challenged the organization?s senior leadership. 28 According to my dataset, 29 different factions emerged within Fatah between 1959 and 2019. Throughout its history Fatah was characterized by stable leadership which ?contributed to its influence and staying power? (Miller, 1983). According to Cobban (1984), ?The stability which marked the composition of Fatah leadership over the decades?stood in stark contrast to notions generally held to in the West about the ?fractiousness? and even ?fissiparousness? of the organization.? Over the years Fatah?s leaders commanded the organization ?from the center? facilitating an ?extremely strong? continuity between Fatah?s founders and the organization?s leadership (Cobban, 1984). To secure control over the organization, Fatah?s founders ?constituted themselves into a Central Committee, which was to be the seat of the organization?s greatest day-to-day power? since the late 1960s (Cobban, 1984; Rasgon, 2015). As Fatah?s most senior institution and its executive arm (Hijazi & Lovatt, 2018), FCC is Fatah?s nerve system. FCC develops and implements the organization?s strategic vision and initiatives (Rasgon, 2015). Even today, FCC members hold some of the most important and sensitive portfolios in Palestinian politics (Rasgon, 2015). Since Fatah?s establishment in 1959, its core leaders7 were in the ?middle of the Fatah web? and have continued to dominate FCC and Fatah since then (Cobban, 1984). Fatah?s decision-making has remained hierarchical (Khalil Shikaki, personal 7 Yasser Arafat, Salah Khalaf (Abu Iyad), Khalil Wazir (Abu Jihad), and Khaled al- Hassan are Fatah?s founding fathers. The incumbent Palestinian president Mahmoud Abbas, Farouk al-Qaddumi, and Mohammed Ghoneim are prominent members who joined Fatah shortly after the organization?s establishment and are considered to be Fatah?s cofounders. Together, Fatah?s founders and cofounders are often referred to as Fatah?s core leadership. 29 communication, March 5, 2020) and initiatives to affect the organization led by younger activists have historically been discouraged (H?igilt, 2016). Veteran members of Fatah who joined the organization before 1968 were reluctant to delegate power or responsibility to newer members (Cobban, 1984). Consequently, by the early 1980s, ?only very few activists could hope to wield much real power inside Fatah unless they already had a solid record of over 15 years of continuous service in the movement? (Cobban, 1984). Fatah?s most pressing decisions were made in secret consultation by core leadership (Pearlman, 2011). While Fatah?s other decision-making bodies - the General Conference and Revolutionary Council (FRC) - were tasked with checking the powers of FCC, in practice they had little effective power. Cobban (1984) concludes that ?the development of Fatah since its inception has remained overwhelmingly in the hands of its Central Committee,? controlled by the core leadership. Fatah?s centralized decision-making and the systemic exclusion of junior cadres from senior leadership positions are evident by the group?s low turnover in senior leadership. Between 1967 and 2016, a total of 57 individuals were members of the FCC. On average, an FCC member served 14.7 years on the committee. However, this figure is misleading as the turnover of Fatah?s senior leadership was very low and power was concentrated in the hands of the same few individuals. Specifically, 14 individuals averaged 33 years on the committee, compared with an average of 11 years served by the remaining 43 members. Viewed from a different perspective, a minority of 14 individuals together served 55% of the total years of FCC membership between 1967 and 2016, whereas a majority of 43 individuals served 45% of the total years of FCC 30 membership. Commenting on Fatah?s structure, a former Palestinian diplomat reflected that ?Fatah is a system that doesn?t rejuvenate itself? (personal communication, October 29, 2019). Fatah?s leaders provided central direction and leadership to the Palestinian national movement (Miller, 1983). Specifically, the domination of Fatah?s senior leaders over Palestinian politics extended to the two central organs of the Palestinian national movement: the Palestinian Liberation Organization (PLO) and the PA, which were founded in 1964 and 1994, respectively. As Salah Khalaf, one of Fatah?s founding fathers said, whatever happens ?within Fatah reflects, one way or another, on the entire resistance movement?And if Fatah falls the entire Palestinian movement will do the same? (Maksoud, 1973). Founded in 1964 by the Arab League as an instrument to control and monitor Palestinian nationalism, the PLO was meant to serve as an umbrella organization for all Palestinian political and guerilla organizations.8 Since 1974, the PLO has increasingly gained international recognition as the representative of the Palestinian people, serving as the de facto Palestinian representative body. Within the PLO, Fatah has always constituted the dominant group (Miller, 1983; Cobban, 1984; Sayigh, 1997). Fatah leaders have historically controlled the PLO?s most important institutions and staffed its diplomatic posts abroad (Miller, 1983). When Yasser Arafat, Fatah?s strongman, was elected PLO chairperson on February 1969, Fatah had effectively gained control over the organization (Br?ning, 2013). In the years that followed, Fatah strengthened ?its hold on all parts of the PLO apparatus, while Fatah?s own?vigor and resilience 8 Historically, Fatah has blocked some prominent Palestinian Islamist rebel groups, such as Hamas and Palestinian Islamic Jihad (PIJ), from becoming members of the PLO. 31 expanded the PLO?s hold over all aspects of Palestinian public life? (Cobban, 1984). By the late 1970s, Fatah?s senior leaders dominated the Palestinian body politics in near totality. Fatah leaders ensured that the majority of seats in all PLO bodies were occupied by Fatah loyalists and, crucially, Fatah?s core leadership often decided secretly on matters related to the PLO more broadly (Pearlman, 2011), allowing them to ?steer the PLO in the direction they wanted? (Cobban, 1984). External financial and logistical support to the PLO by Arab states, estimated at over half a billion dollars during the 1970s and 1980s, was channeled by Fatah?s leaders to secure their dominance over the organization (Cobban, 1984). Funds were often used to counter the grassroots organizational opposition (Brynen, 1995). Fatah?s leaders did not respond to episodes of dissent and factionalization by democratizing the organization, but rather by further centralizing it. To maintain their grip over the organization, Fatah?s senior leaders prized personal loyalty over competence and corruption by loyal commanders was overlooked in the interest of power (Brynen, 1995). The outbreak of the first Palestinian Intifada in December 1987 surprised Fatah leaders. Concerned that they were about to lose their dominant position in the Palestinian national movement to grassroots activists, Fatah leaders sought negotiations with Israel as a way to regain control over Palestinian politics. Shortly after the signing of the Oslo Accords between Israel and the PLO in September 1993, the PA was founded as an interim phase towards the creation of an independent Palestinian state in 32 the West Bank and the Gaza Strip. With the creation of the PA, Fatah and PLO leaders left Tunis9 and settled in the territories designated for a Palestinian state. For Fatah?s seniors, the PA has simply served as a new framework to maintain their power (Br?ning, 2013; H?igilt, 2016). Fatah entrenched in the PA so deeply that distinction between the organization and the PA institutions is virtually nonexistent (Br?ning, 2013; H?igilt, 2016), making Palestinian politics into a one-party regime (Hilal, 1998). Indeed, since the creation of the PA, 17 Palestinian governments have been formed, and Fatah dominated every one of them (Baroud, 2019). All important decisions were in the hands of PA President and Fatah leader, Arafat, and his cabinet, while democratically elected institutions, such as the Palestinian Legislative Council (PLC), were marginalized.10 According to Husam Khader, a Fatah member, and one of the founders of the AMB faction in Nablus, Palestinian political institutions have no real power. ?I don?t want to trick myself into thinking that we have institutions, laws and rules,? he said in 1998. ?We argue, we bargain. We have procedures and votes and lobbyists of a kind. Yes, the council [PLC] dances beautifully. But in the dark? (Sontag, 1998). The vesting of power in the hands of the same few individuals has resulted in a system of authoritarian government characterized by systemic corruption and personalized politics in both Fatah and Palestinian politics (Robinson, 1998). Instead of relying on institutions and their procedures to make decisions and elect competent 9 Initially, the PLO?s main headquarters were in Jordan. Following King Hussein?s war against the PLO in 1970-1971, the organization relocated to Lebanon. The 1982 Israeli invasion of Lebanon and its subsequent siege over Beirut, led to the removal of Fatah and the PLO?s senior leadership to Tunis, where they were located until the establishment of the PA in 1994. 10 According to the Oslo Accords, the PLC is, in fact, supposed to be the single most important decision-making body in the PA (Robinson, 1998). 33 individuals, Fatah and PA seniors have resorted to personalized power around Arafat (Robinson, 1998; Pearlman, 2011). Moreover, Fatah?s laws stipulate that elections to the organization?s governing bodies, including FCC, should be convened every five years. In reality, however, elections have been only held seven times over the past five decades (al-Omari, 2016; Sayigh, 2016). Access to decision-making and resources were determined through patronage, not by competence and merit. Those lacking ?connections? within Fatah?s core leadership have found themselves unable to access resources or influence policy (Brynen, 1995; Sayigh, 2016). Growing internal authoritarianism and near complete lack of political renewal have left Fatah largely out of touch with its base, including many younger members who stopped seeing any path for advancement within the organization (al-Omari, 2016). After six years of tight senior leadership control over the PA, Fatah?s junior cadres felt completely marginalized (Khalil Shikaki, personal communication, March 5, 2020). Arafat continued to head Fatah, the PLO, and the PA simultaneously until his death on November 2004. In 2005, he was replaced by another founding member of Fatah, the incumbent Palestinian president, Mahmoud Abbas. Fatah members who hoped that Abbas would bring a different leadership style were quickly disillusioned. According to a Fatah junior cadre, ?Abbas convinced us that he is completely different from Arafat. But later we learned that Abbas is different from Arafat in a negative way. Arafat at least listened to people. He made us feel like he is everyone?s dad. Abbas is only his kids? father. He is more authoritarian than Arafat. We ended up with a dictator? (personal communication, March 4, 2020). Fatah?s concentrated power has developed 34 the executive into an authoritarian proto-state with the organization?s ?aging old guard monopolizing top positions? (Brown, 2010). Like his predecessor, Abbas dominates Fatah and the PLO. Because Abbas?s presidential term was supposed to end in 2009, he is an unelected president (H?igilt, 2016). 1.5.2 Factionalization Despite Centralized Leadership Growing centralization in decision-making and systemic exclusion of lower-tier members from access to senior leadership positions through existing mechanisms for upward mobility create ideal conditions for factionalization. A former AMB commander aggrieved that, ?it was unfair that the same leaders from the 1960s were still there. The old leaders refused change and kept sitting on their chairs until their death. On paper Fatah says that it is a democratic movement. In reality, it was governed by only two men: Arafat and then Abbas? (personal communication, March 10, 2020). More broadly, in 19 years (or 32%) of Fatah?s total 60 years of existence, at least 1 faction emerged, with 3 to 4 new factions some years (Figure 1.1). Crucially, factions were significantly more often driven by junior cadres, not by disagreements among senior leadership. Of Fatah?s 29 factions, 86% were driven by junior cadres while only 14% by senior leaders,11 a finding that challenges existing leadership-centric arguments (Pearlman, 2011; Asal et al., 2012; Doctor, 2020). 11 As the case of the al-Mustaqbal faction demonstrates, Fatah?s senior leaders had profound disagreements, occasionally resulting in factionalizations. But for the most part, inter- leadership disputes ?have not resulted in formal breaks? (Miller, 1983). Even as disputes have intensified in the FCC, they rarely resulted in a leader giving up his position in the committee or splintering from Fatah altogether. The incorporation of Fatah?s junior cadres into the process of factionalization explains the numerous faction emergences. 35 Figure 1.1: Number of new Fatah factions per year, 1959 to 2019. For instance, in summer 1983, members of Fatah?s officer corps splintered from the organization in protest against Arafat?s leadership. Led by Fatah?s military officer, Said al-Muragha, the officers protested a number of issues including, corruption, strategy, and Arafat?s autocratic leadership-style (Strindberg, 2000). Members of the faction complained that ?Arafat behaved like an autocrat. He short-circuited the democratic process and organizational structures? (Khalidi & Stork, 1983). Junior cadres also decried the appointment of individuals to key positions based on their personal loyalty to the core leadership as opposed to competence. During the 1980s, complaints arose when Arafat appointed two of his loyalists, Colonel Ghazi ?Atallah and Colonel Haj Isma?il, to command Fatah forces in Lebanon. The men in their units 36 considered Arafat?s loyalists to be underqualified and, indeed, when Israel invaded Lebanon, the two deserted their units (Schiff & Ya?ari, 1984). More than two decades later, in 2005, members of the PRC?s armed wing, the Salah al-Din Brigades, dragged Moussa Arafat, Arafat?s cousin and appointee, out of his villa in Gaza and executed him for alleged corruption (Newton, 2015). Revisiting Fatah under Arafat, a junior cadre reported that they ?were unsatisfied with the Arafat system. He was a symbol, but allowed corruption and surrounded himself by incompetent and corrupt men? (personal communication, March 5, 2020). Commenting on the Oslo process, a member of AMB explained that ?many AMB members did not like the way the process was unfolding. The whole agreement was negotiated in secret. The leaders did not share anything with us. They returned to Palestine and simply presented the agreement. Arafat, Abbas, and their negotiators were thinking only about themselves and their cronies; not about the simple Palestinians? (personal communication, March 10, 2020). 1.5.3 Determinants of Factionalization in Fatah As discussed previously, considering the limitations of grievances and opportunity explanations to independently explain political action, the approach adopted in this article demonstrates how motivation and opportunity explanations for political violence operate complementarily rather than competitively (Bara, 2014; Buhaug et al., 2008; D. Cunningham et al., 2017). To do so, I assess political opportunity in terms of access to senior leadership positions and trace members? grievances around each episode of factionalization. When access to leadership positions is blocked, aggrieved Fatah junior cadres have justified their decision to factionalize with their exclusion from senior 37 leadership positions which restricted their ability to influence Fatah from within. Conversely, when junior cadres believe existing political channels effectively facilitate upward mobility, factionalization decreased significantly. I identified 15 repeated justifications for factionalization, such as authoritarian leadership; disagreements concerning strategy; leaders? corruption and so forth (see Appendix). Among these reasons, I also accounted for the existing argument that inter- leadership disagreements drive factionalization (see e.g., Asal et al., 2012; K. Cunningham, 2017). Next, I consolidated the 15 justifications into 5 categories, presented in Figure 1.2 below. For example, disputes concerning the use of armed struggle or diplomacy were collapsed into one category, ?strategy.? Calls for internal reform, democratic practices (e.g., elections), condemnations of corrupted appointments and so forth were now aggregated into a single category, ?access to decision-making.? As Figure 1.2 illustrates, nearly 60% of the justifications for all factions? emergence concern disagreements between cadres and leaders. More specifically, cadres? grievances against the leadership are followed by junior cadres? grievances concerning access to decision-making (20%). Marginally, only 4% of the grievances involve disputes between the organization?s senior leadership. 38 Figure 1.2: Justifications for factionalization by category, 1959 to 2019. While junior cadres? grievances against Fatah?s leadership are diverse, more specific grievances concern Fatah?s highly centralized decision-making structure and limited access to senior leadership positions. Strategic and ideological disagreement seem more marginal. According to a member of one of Fatah?s factions, the reason for forming the faction ?was not about ideology nor strategy. At the time, we agreed with the senior leadership on nonviolence and on continuing with the peace process? (personal communication, March 4, 2020).12 Highlighting limited access to positions of influence and grievances that followed, another Fatah junior cadre commented that, ?In 12 According to a member of one of Fatah?s factions, the reason for forming the faction ?was not about ideology nor strategy. At the time, we agreed with the senior leadership on nonviolence and on continuing with the peace process.? 39 the decision-making structure we felt that we were Fatah type B. We were left outside of all central decision-making? (personal communication, March 5, 2020). According to a Fatah junior cadre in Ramallah, in comprising the Fatah list towards the Palestinian national elections in 2006, ?we agreed with [President] Abbas on meaningful representation for the junior cadres which will emphasize the organization?s democratization and institutional reform.? When Abbas did not implement what was agreed on, ?we had no choice, but to split? (personal communication, March 4, 2020). ?Otherwise,? as another Fatah junior cadre maintained, ?the leaders would have not accepted us? (personal communication, March 5, 2020). Husam Khader, one of AMB?s founders in Nablus, argued that the leadership ?have broken the ladders to prevent us from rising? (Abu Toameh, 2008). According to AMB commander in Ramallah, ?junior cadres were angry at the leadership. We didn?t get the right to be in FCC. Only the Tunisians13 were influential. The local commanders were excluded. The old leaders controlled Fatah and didn?t give us the opportunity to be in the leadership? (personal communication, March 5, 2020). It is true that the creation of the PA is a unique experience that has further contributed to the Fatah?s authoritarian leadership style (Robinson, 1998; Hilal, 1998; Schulz, 2002; Milshtein, 2004) exacerbated junior cadres grievances, and opened a new opportunities for factionalization (Pressman, 2003; Cheong, 2012). However, its explanatory power seems limited in explaining faction emergence over time. Junior cadre-driven splits have been a recurring event in Fatah?s history, a function of flawed 13 The ?Tunisians? or ?returnees? are popular synonyms often used to describe Fatah and PLO leaders and their loyalists. The terms are based on the fact that following the 1982 Israeli invasion to Lebanon and its subsequent siege over Beirut, senior Fatah and the PLO leaders withdrew to Tunis, where they were located until the establishment of the PA in 1994. 40 channels for upward mobility. Moreover, Fatah?s transition from a revolutionary movement to a ruling political party in the 1990s raise a legitimate question on whether Fatah can be considered a rebel group after the creation of the PA. However, the examination of Fatah through its different developmental phases affords an opportunity to generalizable organizational characteristics such as leadership structure and junior cadres? access to leadership positions in order to improve our understanding of how these organizational features galvanize or inhibit factionalization. Evidently, the exclusion of newer members from power gained traction since the late 1960s and has been further exacerbated since the 1980s (Cobban, 1984; al- Omari, 2016). If the expectations set here are correct, when personnel turnover in the FCC increases, faction emergence should decrease. Elections for the FCC in 1971 did not bring about meaningful change (Sayigh, 1997, p. 294; al-Ghani, 2016). Fatah?s core leaders continued to dominate FCC and of the three members who lost their seat that year, two died of unnatural causes.14 By the early 1980s, ?only very few activists could hope to wield much real power inside Fatah unless they already had a solid record of over 15 years of continuous service in the movement? (Cobban, 1984). Frustration about the lack of access to decision-making was manifested in a spike in faction emergence during the 1980s. The so-called ?Fatah rebellion,? which broke out in 1983, was one of the most challenging moments in Fatah?s history. Leaders of the Fatah Uprising faction, which emerged that year, argued that given Arafat?s disregard for the democratic councils of Fatah, the grassroots had no alternative but to leave the organization and use 14 Walid Nimr was executed by the Jordanian forces in July 1971 (Sayigh, 1997, p. 279); Mamduh Saydam died of cancer in July 1971 (Sayigh, 1997, p. 294). Salim al-Za?anoun lost his seat (Sayigh, 1997, p. 294). 41 force to redress their grievances (Khalidi & Stork, 1983; Strindberg, 2000; Pearlman, 2011, p. 89). Between 1980 and 1986 five new factions emerged and the total number of factions doubled between the late 1970s and 1987. By 1987 the total number of Fatah factions has reached a historical record of 10. Amid growing dissent within its base Fatah?s leaders announced elections to all Fatah governing institutions, including FCC. FCC was expanded from 15 to 22 members with 11 new members serving in FCC for the first time. Between 1988 and 1999 only one new faction emerged and the total number of Fatah factions has gradually decreased to 6 (1988-1990) and then to 5 (1991- 1999). After years of fairly dormant faction emergence, factions started emerging more rapidly in the early 2000s. Intensified marginalization of Fatah?s junior cadres (al- Omari, 2016; Sayigh, 2016) has resulted in the emergence of well-organized but loosely led factions which used violence to promote their interests (H?igilt, 2016). Most prominently, in September 2000 disenfranchised members of Fatah formed the AMB and turned Tanzim against Israel in order to establish their political influence vis-?-vis Fatah and the PA leaderships on matters regarding political and economic opportunities as well as strategy. According to Pearlman (2011), Fatah?s junior cadres ?tended to feel shut out of meaningful political participation? and forming these factions, as well as taking up arms, gave them the political voice they lacked, enabling them to exert pressure on Fatah elites to take them seriously (Shikaki, 2001, 2002; Pearlman, 2011). According to Shikaki, ?They saw Arafat and his cronies withholding from them a role to shape the future of Fatah and PA. Violent escalation with Israel opened the 42 opportunity for a rebellion that was underway for some time? (Khalil Shikaki, personal communication, March 5, 2020). 1.5.4 Limitations of Alternative Explanations The availability of weapons (Seymour, 2014) does not explain the decision of junior cadres to factionalize. In her study of the Palestinian national movement, Pearlman (2011) argues that, ??the leap from weapons endowment as a fact to weapons endowment as a causal explanation is insufficient? (p. 152). For instance, access to weapons does not explain why Fatah junior cadres factionalized during the second Intifada but did not in the Oslo years or between 2009 and 2019,15 nor does it account for the emergence of multiple nonviolent factions. Similarly, the availability of funds alone does not explain the decision of junior cadres to factionalize. Some of Fatah?s factions, such as the Arab Revolutionary Brigades and ANO, have been heavily dependent on the financial and logistical support of external sponsors, such as Libya and Iraq (Friedman, 1986; Sayigh, 1997, p. 311), whereas the AMB emerged in the absence of such support. Fatah?s competition with other Palestinian rebel groups, most conspicuously Hamas (Bloom, 2004), have certainly affected the calculations of Fatah junior cadres to factionalize, as acknowledged by several interviewees. But factions emerged throughout Fatah?s history, oftentimes completely disassociated from competitive dynamics. Finally, while state repression may increase the costs of mobilization and provoke internal splits among rival factions vying for power (Seymour et al., 2016), 15 Between 2009 and 2019 only one nonviolent faction emerged: Fatah ? Democratic Reform Bloc or al-Mustaqbal. 43 factionalization in the case of Fatah occurred numerous times outside of the Palestinian Territories, where Israeli or other state repression was out of reach. Hence, rather than discounting existing explanations, I claim that their scope limits their applicability to a category of rebel groups characterized by low turnover in leadership positions that increase the likelihood of factionalization. It was the lacking de jure power that caused Fatah?s junior cadres to accuse the senior leadership of excluding them from leadership positions, cronyism, and corruption (Shikaki, 2001; Aly et al., 2013). Jamal Abu Samhadana, who served as the PRC?s leader until his death in 2006 (?The Popular Resistance Committees,? 2011), described Fatah?s decision-making style as one in which ?Everything is determined by the decision of the ?headmaster.? I believe Arafat is a dictator? (MacFarquhar, 1993). To change this, several Fatah?s junior cadres sought to remove the senior leadership (Shikaki, 2002; Aly et al., 2013). Blocked from influencing Fatah from within, Fatah junior cadres factionalized and resorted to armed struggle. In 2000, members of Fatah founded their own independent armed group, the AMB (Aly et al., 2013). A Fatah junior cadre in Ramallah claims that, ?we believed that after the intifada, we would reap the fruit, which was to be in the decision-making and shape Palestine the way want it: clean from corruption, democratic, humble, and egalitarian? (personal communication, March 5, 2020). According to a close companion of Marwan Barghouti, leader of Tanzim and AMB, ?Barghouti saw the activity of Fatah?s grassroots as a way to neutralize the FCC, a body he considered anachronistic and irrelevant. Marwan saw members of his rank and the grassroots as the authentic representatives of the people unlike those [senior leaders] who were living like princes in the [Persian] Gulf, completely detached from 44 the people? (Fatah junior cadres and member of the Palestinian Legislative Council, personal communication, March 5, 2020). While the 2009 Fatah conference did not expand FCC membership, it was profoundly reshuffled. New leadership emerged as the majority of the newly elected FCC were first-time members, including Marwan Barghouti (B2. Election Results for the Fatah Central Committee, 2009). Faction emergence slowed down again. The only faction to emerge after 2009, was Muhammad Dahlan?s Fatah- Democratic Reform Bloc which appeared in 2016.16 ?We took our right through the intifada,? the Ramallah AMB commander argued, ?and by 2009 we could be part of the decision-making? (personal communication, March 5, 2020). 1.6 Conclusion This article argues that centralized leadership with low personnel turnover in high-level decision-making bodies galvanizes faction emergence among aggrieved junior cadres who are systematically excluded from senior leadership positions. Using two original datasets on Fatah factions and leadership structure, this article demonstrates three key findings. First, a centralized decision-making structure does not immunize a rebel group to factionalization. Second, in the majority of cases factions are driven by disenfranchised junior cadres and not by disagreements at the leadership level as current literature argues. Third, political opportunity structure and motivation work complimentarily to explain faction emergence. These findings further elucidate the 16 In 2011, Dahlan was expelled from Fatah after falling out with President Mahmoud Abbas and in 2016 the faction emerged, following the failure of reconciliation between Dahlan and Abbas (Rasgon, 2015). The following individuals were added to FCC by a decision of FCC and endorsed by the FRC instead of Mohammed Dahlan: Nabil Abu Rdainah, Zakaria Al-Agha, Sakhr Bseiso, Amal Hamad (al-Ghani, 2016). 45 processes by which rebel groups splinter and contributes to the broader discussion on the roles of motivation, structure and agency in shaping the dynamics of rebel group fragmentation. But research remains incomplete. First, future research should incorporate junior cadres and the rebel groups? decision-making structure into large-N studies in order to assess the theoretical arguments presented here on a larger number of rebel group cases. Second, factionalization does not necessarily spell the rebel group?s complete disintegration. Despite the emergence of multiple factions throughout its history, Fatah not only survived, but it maintained its dominant position in the Palestinian national movement until today. This insight should take us back to Fatah?s leaders. Future researchers might consider how leaders respond to junior cadres? grievances as a way to prevent further factionalization. Why would the same rebel leadership hold on to their seats for such a long time, refusing to share power (or at least the spoils of power) better with junior cadres, especially if doing so will lead to factionalization? Do senior leaders design democratic procedures and form institutions to contain grievances short of factionalization? And if so, what determines their success or failure over time? Finally, and more specifically in the case of Fatah, was there a structural factor that made factionalization a threat or problem negligible enough for the senior leadership not to undertake meaningful political reforms? From a policy perspective, this study makes several suggestions. First, understanding how rebel groups are able to retain members effectively is an indispensable factor for governments that are fighting rebels because it can facilitate the 46 development of effective tools for detaching members from the rebel group (Haer et al., 2011). Second, existing research finds that states are more likely to negotiate with rebel leaders that come to power through local selection processes such as elections (K. Cunningham & Sawyer, 2019). If the exclusion or inclusion of junior cadres from senior leadership positions tell us something about rebel groups cohesion, governments interested in negotiating with the rebels should not only follow how leadership has been reshuffled but also whether junior cadres were incorporated in the process in a meaningful way and whether they gained access to senior leadership positions. Third, junior cadres? expressions of their grievances are an invaluable testimony to the level of their satisfaction with existing institutional arrangements. Because factionalization is mostly driven by aggrieved junior cadres, monitoring their grievances can help policy analysts to forecast organizational splits. Relatedly, grievances expressed by group members offer governments immaculate information to assess in- group differences. Depending on the government?s objectives, such insights can help it to identify which individuals constitute an opportunity for dialogue, isolate extremists, as well as exacerbate intragroup divisions in order to create fissures. 47 1.7 Appendix Factionalization From Below: The Case of Palestinian Fatah Table 1.1: List of Fatah factions (1964 to 2016) Source: Fatah Year Faction Dataset Number Faction Name Year of Emergence 1 The Madmen (al-Majanin) 1964 2 The Rational Wing (al-?Aqalaniyyun) 1964 3 The Dissenting Wing (al-Janah al-Munshaq) 1967 4 The Democratic Direction (al-Ittijah al-Dimuqrati) 1968 5 The Black September Organization (BSO) 1971 6 The Soviet Group 1973 7 The Vietnamese Line 1973 8 The Maoist Tendency 1973 9 Arab Revolutionary Brigades 1973 10 Abu Nidal Organization (ANO) 1974 11 Palestinian Communist Workers Party (PCWP) 1978 12 Arab Socialist People?s Movement 1978 13 Fatah Movement / March Correction 1980 14 Leftist Corrective Movement 1983 15 Fatah al-Intifada (Fatah Uprising) 1983 16 Second Corrective Movement 1984 17 The Supreme Military Council of the Fatah Movement 1986 18 Abu-Rish Brigades 1991 19 Tanzim17 2000 20 The al-Aqsa Martyrs? Brigades (AMB) 2000 17 Although Arafat established the Tanzim as an armed wing of Fatah in 1995, multiple sources report that Fatah?s ability to rein in the Tanzim has been gradually eroded. The pick comes in 2000 with the beginning of the Second Intifada as Tanzim members often acts independently from the orders of Fatah?s leaders. Indeed, multiple sources refer to Tanzim as internal opposition within Fatah (see e.g., Usher, 2000; Schenker, 2000; Fishman, 2015). I have decided to code the Tanzim as a faction starting in 2000 (beginning of the Second Intifada) as the group acts in defiance of Fatah?s central leadership. 48 21 Popular Resistance Committees (PRC) 2000 22 ?Abd al-Qadir al-Husayni Brigadas 2001 23 Shaheed Nabil Masoud Unit 2002 24 The Nidal al-?Amoudi Battalion 2002 25 The Ayman Juda Brigades 2002 26 al-Mustaqbal 2005 27 al-Ahrar 2007 28 The al-Mujahideen Brigades 2008 29 Fatah ? Democratic Reform Bloc 2016 Data on Fatah Central Committee (FCC) Table 1.2 Descriptive statistic Variable N Mean St. Dev. Min. Max. Elections 50 0.14 0.35 0 1 Expand 50 0.08 0.274 0 1 Total members 50 17.7 4.59 9 22 Change in total members 50 0.163 1.389 -3 3 Number of members survived 50 9.122 2.315 4 12 Share of members survived 50 0.542 0.176 0.181 1 Number of members that did not survive 50 0.633 2.288 0 15 Number of new members 50 0.959 2.806 0 14 Share of new members 50 0.053 0.15 0 0.736 Elections: A dichotomous variable which records whether elections to Fatah congress took place in a given year. Expand: A dichotomous variable which records whether FCC expanded in terms of the total number of members in a given year. Total members: A count variable which records the total number of FCC members in a given year. 49 Change in total members: A count variable which records the change in the total number of FCC members in a given year compared with the previous year. For example, in 1967 the total number of FCC members (Total members) was 11 and in 1968 the total number of FCC members dropped to 9. Therefore the value for Change in total members in 1968 is -2. Number of members survived: A count variable which records the number of FCC members from the previous year which continue to serve on FCC in the following year. For example, in 1988 eleven FCC members who served on the committee in 1987 also served on the committee in 1988. Therefore the value for Number of members survived in 1988 is 11. Note that the variable accounts for members who continue to serve on FCC regardless of whether elections for the committee took place in a given year. Share of members survived: A continuous variable which records the proportion that those who continue to serve in the committee comprise of the total members of FCC in a given year. The Share of members survived is calculated as: Number of members survived / Total members = Share of members survived. Number of members that did not survive: A count variable which records the number of FCC members from the previous year who did not serve on the committee in the following year. Note that the variable accounts all possible reasons of end of membership, including: voted out during elections, expulsion, death (including assassination or death on the battle field) etc. Number of new members: A count variable which records the number of new FCC members who joined (or rejoined) the committee in a given year. For example, following Fatah?s congress on December 2016, six new members joined FCC. Therefore, the values for Number of new members in 2016 is 6. Share of new members: A continuous variable which records the proportion that new FCC members comprise of the total members of FCC in a given year. The Share of new members is calculated as: Number of new members / Total members = Share of new members. 50 Sources on FCC membership: al-Ghani al-Shami, A., 2016. Conferences of the ?Fatah? Movement Since its Inception [online], Quds Press International News Agency. Available from: http://www.qudspress.com/index.php?page=show&id=25753. (2009) B2. Election Results for the Fatah Central Committee, Bethlehem, 8 August 2009. Journal of Palestine Studies, 39 (1), 170-171. Cobban, H., 1984. The Palestinian Liberation Organization: People, Power, and Politics. Cambridge University Press. ?Fatah Congress elects Central Committee and Revolutionary Council members, Ramallah,? 2016. WAFA News Agency, 5 Dec. Available from: https://english.wafa.ps/page.aspx?id=hWTZzja51779170212ahWTZzj. Rasgon, A., 2015. Fatah Central Committee Profiles, The Washington Institute for Near East Policy. Available from: https://www.washingtoninstitute.org/policy- analysis/fatah-central-committee-profiles. Sayigh, Y., 1997. Armed Struggle and the Search for State: The Palestinian National Movement, 1949-1993. Oxford University Press. Data on Fatah factions Table 1.3 Descriptive statistic Variable N Mean St. Dev. Min. Max. New faction 71 0.408 0.495 0 1 Number of new 71 0.831 1.23 0 4 factions Total factions 71 6.887 3.616 0 12 Faction by cadres 71 0.862 0.351 0 1 Leadership- 71 0.172 0.384 0 1 leadership dispute (r_1) Leadership-cadre 71 0.759 0.435 0 1 dispute (r_2) Cadre-cadre dispute 71 0.034 0.186 0 1 (r_3) Armed struggle 71 0.448 0.506 0 1 (r_4) Diplomacy (r_5) 71 0.276 0.455 0 1 51 Cadres? explicit 71 0.69 0.47 0 1 complaints about the leadership (r_6) Authoritarian or 71 0.379 0.494 0 1 centralized leadership style (r_7) Complaints against 71 0.517 0.509 0 1 Yasser Arafat and Mahmud Abbas (r_8) Leadership- 71 0 0 0 1 leadership ideological dispute (r_9) Leadership-cadre 71 0.345 0.484 0 1 ideological dispute (r_10) Internal reform 71 0.172 0.384 0 1 (r_11) Democracy (r_12) 71 0.172 0.384 0 1 Corruption (r_13) 71 0.345 0.484 0 1 Governance (r_14) 71 0.034 0.186 0 1 Access to decision- 71 0.207 0.412 0 1 making (r_15) New faction: A dichotomous variable which records whether a new faction emerged in a given year. In total, 29 Fatah factions have emerged between 1959 and 2019. Number of new factions: A count variable which records the total number of new factions emerged in a given year. Faction by cadres: A dichotomous variable which records whether a faction was driven by Fatah?s junior cadres or not. A faction is considered driven by cadres (coded ?1?) if none of its members is also an FCC member at the time of the faction?s emergence. A faction is driven by leaders (and coded ?0?) if at least one of its members was also an FCC member at the time of the faction?s emergence. Of Fatah?s 29 factions (1959- 2019), 25 were driven by cadres and 4 by leaders. Total factions: A count variable which records the total number of factions in a given year. The count includes both old factions which still exist (i.e., factions that survived) and new factions (i.e., factions that emerged in that year). 52 r_1 to r_15: For each faction in the dataset, a set of dichotomous variables which records the justifications for its emergence. r_1: Leadership-leadership dispute: If the faction was the result of a dispute between at least two senior Fatah members, measured by FCC membership at the time when the faction emerged (see Faction by cadres). r_2: Leadership-cadre dispute: If the faction was the result of a dispute between Fatah?s junior cadres and the organization?s senior leadership (see Faction by cadres). r_3: Cadre-cadre dispute: If the faction was the result of a dispute between Fatah?s junior cadres. r_4: Armed struggle: If the justification for factionalization involved disagreements regarding the use of violence, arms, and other forms of political violence. r_5: Diplomacy: If the justification for factionalization involved disagreements regarding the engagement in diplomatic, non-military strategies such as negotiations with Israel; different forms of political settlement (e.g., the ?phased political program?)18; the recognition of Israel and so forth. r_6: Cadres? explicit complaints about the leadership: If the justifications for factionalization involved complaints made by junior cadres against the leadership. For example consider the following quote by a member of Fatah Uprising, which splintered from Fatah in 1983: ?Arafat behaved like an autocrat. He short-circuited the democratic process and organizational structures.? (Quoted in Khalidi & Stork, 1983). r_7: Authoritarian or centralized leadership style: If the justifications for factionalization involved issues such as lack of political accountability; leadership failure to consult; lack of representation (in decision-making processes); aversion to sharing power and so forth. For example, see Naji ?Allush?s Arab Socialist People?s Movement and his opposition to the paternalistic and eclectic modes of leadership in Fatah (Sayigh, 1997, p. 352). r_8: Complaints against Yasser Arafat and Mahmud Abbas: If the justifications for factionalization refer explicitly to two (and only Fatah) chairpersons, Yasser Arafat and Mahmoud Abbas. 18 Fatah and the PLO?s ?phased program? was a political program to negotiate a political settlement with Israel in order to recover the West Bank and the Gaza Strip, as a preliminary phase towards the ultimate goal of liberating all Palestine. The program alarmed many members of the different Palestinian guerrilla groups as well as noncombatant Palestinian refugees who were concerned that, if concluded, the program would leave them in exile permanently. As a result, many of the refugees looked for more militant nationalist factions (usually leftist) within Fatah and the PLO (Sayigh, 1997, p. 353). 53 r_9: Leadership-leadership ideological dispute: A more specific category of leadership-leadership dispute (r_1) designed to capture specific ideologically-based ? as opposed to strategic ? disputes. r_10: Leadership-cadre ideological dispute: A more specific category of leadership- cadre dispute (r_2) designed to capture more ideologically-based ? as opposed to strategic ? disputes. For example, the PRC is known to have a radical Islamist ideology close to that of Hamas (?The Popular Resistance Committees,? 2011) and members of its armed wing, Liwa al-Tawheed depict the conflict with Israel in Islamic religious terms (al-Tawheed, 2021). r_11: Internal reform: If the justifications for factionalization involved demands for internal political or organizational reforms in Fatah. For example, in a document published by members of the AMB they demanded from Fatah?s leadership genuine reform that would secure an increase in the representation of the younger cadres (Cheong, 2012; BBC, 2004). r_12: Democracy: If the justifications for factionalization included calls for more democracy in Fatah; including elections. For example, calls for expanding democracy were expressed by members of several Fatah factions, most notably Fatah Uprising, Tanzim, and Mohammad Dahlan Fatah Democratic Reform Bloc. r_13: Corruption: If the justifications for factionalization included accusations of corruption; such as of transparency or the promotion of loyalty over competent commanders. r_14: Governance: If the justifications for factionalization included demands for a better governance. r_15: Access to decision-making: If the justifications for factionalization included demands for a greater access to decision-making in Fatah or complaints concerning the absence of access to decision-making. This may involve disputes concerning influence over policy making, mobilization and promotion, representation, voice, exclusion from senior levels of leadership and so forth. Figure 1.2: Justifications for factionalization by category, 1959 to 2019. Cadres? grievances against leadership: (1) Leadership-cadre dispute (r_2) (2) Complaints about the leadership (r_6) (3) Authoritarian; centralized leadership (r_7) (4) Yasser Arafat; Mahmoud Abbas (r_8) (5) Leadership-cadre ideological dispute (r_10) 54 Access to decision-making: (1) Internal reform (r_11) (2) Democracy (r_12) (3) Corruption (r_13) (4) Governance (r_14) (5) Access to decision-making (r_15) Strategy: (1) Armed struggle (r_4) (2) Diplomacy (r_5) Inter-leadership disputes: (1) Leadership-leadership dispute (r_1) (2) Leadership-leadership ideological dispute (r_9) Cadre-to-cadre disputes: (1) Cadre-to-cadre disputes (r_3) Table 1.4 Descriptive statistic Variable N Mean St. Dev. Min. Max. Cadres? 71 0.538 0.5 0 1 grievances against leadership (r_2, r_6, r_7, r_8, r_10) Access to 71 0.186 0.391 0 1 decision- making (r_11, r_12, r_13, r_14, r_15) Strategy (r_4, 71 0.362 0.485 0 1 r_5) Inter- 71 0.086 0.283 0 1 leadership disputes (r_1, r_9) Cadre-to-cadre 71 0.1724 0.384 0 1 disputes (r_3) 55 Sources on Fatah factions Faction?s name: The ?rational wing? (al-?aqlaniyyun) Sayigh, Y., 1997. Armed Struggle and the Search for State: The Palestinian National Movement, 1949-1993. Oxford University Press, p. 106-128. Faction?s name: The ?madmen? (al-majanin) Sayigh, 1997, pp. 106, 123. Faction?s name: The ?dissenting wing? (al-janah al-munshaq) Sayigh, 1997, p. 156. Faction?s name: The ?democratic direction? (al-ittijah al-dimuqrati) Ganim, A., 2010. The Politics of Reform in the Palestinian National Authority. In Palestinian Politics After Arafat: A Failed National Movement. Indiana University Press, p. 96. Sayigh, 1997, pp. 224, 294. Faction?s name: The Black September Organization (BSO) Byman, D., 2011. The Rise of the PLO and the Birth of International Terrorism (1956? 1970s). In A High Price: The Triumphs and Failures of Israeli Counterterrorism. Oxford University Press, pp. 29-57. Cobban, H., 1984. The Palestinian Liberation Organization: People, Power, and Politics. Cambridge University Press, p. 54. Jubran, M., 1990. Abu Nidal: Portrait of a Renegade. The Washington Report on Middle East Affairs, 22 Feb, p. 9. Morris, B., 2001. Righteous Victims: A History of the Zionist-Arab Conflict, 1881-2001. First Vintage Books, pp. 378-383. Naor, D., 2014. Spring of Youth? in Beirut: The Effects of the Israeli Military Operation on Lebanon. Israel Affairs, 20 (3), 410?425. Sayigh, 1997, pp. 306-307. Faction?s name: The ?Soviet Group? Sayigh 1997, pp. 349-352. Faction?s name: The ?Vietnamese Line? Sayigh 1997, pp. 351-352. 56 Faction?s name: The ?Maoist Tendency? (also self-dubbed as the ?correct line?) Sayigh 1997, p. 352-353. Faction?s name: Arab Revolutionary Brigades Badwan, A., 2006. Fatah?s revolutionary council committed assassinations to reiterate its political positions [in Arabic], Al-Bayan Foundation for Press. Available from: https://web.archive.org/web/20071122232833/http:/www.albayan.ae/servlet/Satellite? c=Article&cid=1164201518878&pagename=Albayan%2FArticle%2FFullDetail. Cobban, 1984, p. 55. Sayigh, 1997, pp. 311, 355. Tanner, H., 1975. P.L.O. Declares It Will Punish Hijackers. The New York Times. Available from: https://www.nytimes.com/1975/01/30/archives/plo-declares-it-will- punish-hijackers.html. Faction?s name: The Abu-Nidal Organization (ANO) Abu Nidal Organization (ANO), aka Fatah Revolutionary Council, the Arab Revolutionary Brigades, or the Revolutionary Organization of Socialist Muslims [online], 2009. The Council on Foreign Relations. Available from: https://www.cfr.org/backgrounder/abu-nidal-organization-ano-aka-fatah- revolutionary-council-arab-revolutionary-brigades. Fisk, R., 2008, Abu Nidal, notorious Palestinian mercenary, ?was a US spy,? Independent, 25 October, sec. Middle East. Available from: https://www.independent.co.uk/news/world/middle-east/abu-nidal-notorious- palestinian-mercenary-was-us-spy-972812.html. Friedman, T. L., 1986. Abu Nidal: Life of a P.L.O. Renegade, The New York Times, 1 Jan, p. 4. Jubran, M., 1990. Abu Nidal: Portrait of a Renegade. The Washington Report on Middle East Affairs, 22 Feb, p. 9. Morris, B., 2001. Righteous Victims: A History of the Zionist?Arab Conflict, 1881? 1999. New York: Knopf Doubleday. Sayigh, 1997, pp. 354-355. Steinberg, M., 1988. ?The Radical Worldview of the Abu Nidal Faction,? Jerusalem Quarterly, 48, 88-104. Sullivan, Colleen. Abu Nidal Organization. In Gus Martin ed. The SAGE Encyclopedia of Terrorism, 2011. SAGE, pp. 5-6. Terrorist Group Profiles, 1988. Washington, D.C.: Department of Defense, pp. 3-8. 57 Faction?s name: Palestinian Communist Workers Party (PCWP) Abu Khalil, A., 1987. Internal Contradictions in the PFLP: Decision-Making and Policy Orientation. Middle East Journal, 41 (3), 361-378. Badwan, 2006. Sayigh, 1997, p. 434. Faction?s name: Arab Socialist People?s Movement; Arab Popular Socialist Movement Sayigh, 1997, pp. 352, 355. Faction?s name: Fatah movement / march (rally) correction / Fatah Correction Path Badwan, 2006. Shriteh, F., 2014. Gaza: The bleeding wound. Xlibris Corporation. Faction?s name: Leftist ?corrective movement? Cobban, 1984, p. 62. Sayigh, 1997, pp. 558-9, 561-7. Faction?s name: Fatah al-Intifada (Fatah Uprising) Ahmad, E., Paul, J., & Stork, J., 1983. Yasser Arafat?s Nightmare. MERIP Reports, (119), pp. 18-23. BBC 2014. ?Commander Gives One-Week Ultimatum to Yarmouk Camp Militants.? CIA Report-1983 Fatah split, p. 5. Fatah al-Intifada, Terrorism Research & Analysis Consortium. Available from: https://www.trackingterrorism.org/group/fatah-al-intifada. Khalidi, R., & Stork, J., 1983. Behind the Fateh Rebellion. MERIP Reports, (119), pp. 6-12. Pearlman, W., 2011. Violence, Nonviolence, and the Palestinian National Movements. Cambridge University Press, p. 89. Rollins, T., 2017. Palestinian-Syrian Militarization in Yarmouk [online], The Atlantic Council. Available from: https://www.atlanticcouncil.org/blogs/syriasource/palestinian-syrian-militarization-in- yarmouk/. Sheikh Hassan, I., & Hanafi, S., 2010. ?(In)Security and Reconstruction in Post-conflict Nahr al-Barid Refugee Camp,? Journal of Palestine Studies, 40 (1), p. 27-48. Sheikh Hassan, I., & Hanafi, S., p. 32. 58 Strindberg, Anders, 2000. The Damascus-Based Alliance of Palestinian Forces: A Primer. Journal of Palestine Studies, 29 (3), 60-76. Faction?s name: Second ?corrective movement? Sayigh, 1997, p. 598. Faction?s name: (Atallah?s) Second ?corrective movement? Hijazi, I. A., 1986, EX-AIDE TO ARAFAT, NOW A FOE, SAYS P.L.O. HAS A FORTUNE IN BANKS. The New York Times, 29 June, p. 16 sec. 1. Available from: https://www.nytimes.com/1986/06/29/world/ex-aide-to-arafat-now-a-foe-says-plo- has-a-fortune-in-banks.html. HUSSEIN?S HAND SEEN IN ANTI-ARAFAT ACTIVITY. The New York Times, 9 April, 1986 p. 12 sec. A. Available from: https://www.nytimes.com/1986/04/09/world/hussein-s-hand-seen-in-anti-arafat- activity.html. Salam, K., Black September. Europa & Middle East News. Available from: http://www.kawther.info/wpr/memories/black-september. Sayigh, 1997, p. 588. Faction?s name: Abu-Rish Brigades Anderson, J. W., 2004, Air Strike Kills Israeli Militant. The Washington Post, 30 July, sec. World in Brief. Available from: https://www.washingtonpost.com/archive/politics/2004/07/30/world-in- brief/02f80b4b-6a3c-49b2-acee-0d4fa49c266a/?noredirect=on. Myre, G., 2004. Israel Adjusts Route of West Bank Barrier to Obey Its Court. The New York Times, 30 July. Available from: https://www.nytimes.com/2004/07/30/world/israel-adjusts-route-of-west-bank-barrier- to-obey-its-court.html. Rollins, 2017. Sela, Y., 2008. Gaza: Abu Rish Brigades Organization [online], The Middle East Forum. Available from: http://www.mideast.co.il/p-2_a-176/. Splinter Group?s Loyalty Limited to What They Want to Hear [online], 2004. The Age Available from: https://www.theage.com.au/national/splinter-groups-loyalty-limited- to-what-they-want-to-hear-20040807-gdyerk.html. Faction?s name: The al-Aqsa Martyrs? Brigades (ABM) Al-Aqsa Martyrs Brigades (AMB) ? Fatah [online], Mapping Palestinian Politics. Available from: 59 https://ecfr.eu/special/mapping_palestinian_politics/al_aqsa_martyrs_brigades_amb_f atah/. BBC News, 2004. Al-Aqsa Sets Out Palestinian Reform Plan, Disengagement Deal Conditions, 28 July BBC News, 28 July. Cheong, D. D., 2012. Fatah?s al-Aqsa Martyrs? Brigades: Armed struggle as a response to a crisis of legitimacy. Melbourne Journal of Politics, 35, 39-59. Fletcher, H., 2009. Al-Aqsa Martyrs Brigade [online], Council of Foreign Affairs. Available from: https://www.cfr.org/backgrounder/al-aqsa-martyrs-brigade. Levitt, M., 2002. Designating the al-Aqsa Martyrs Brigades, The Washington Institute for Near East Policy. Available from: http://www.washingtoninstitute.org/policy- analysis/view/designating-the-al-aqsa-martyrs-brigades. Pearlman, W., 2011. Violence, Nonviolence, and the Palestinian National Movements. Cambridge University Press, pp. 167-169. Shahar, Y., 2002. The al-Aqsa Martyrs Brigades - A political tool with an edge [online], International Institute for Counter-Terrorism. Available from: https://www.ict.org.il/Article.aspx?ID=822#gsc.tab=0. Zilber, N., & Al-Omari, G., 2018. State with No Army, Army with No State: Evolution of the Palestinian Authority Security Forces, 1994-2018 [online], The Washington Institute for Near East Policy. Available from: https://www.washingtoninstitute.org/policy-analysis/state-no-army-army-no-state- evolution-palestinian-authority-security-forces-1994. Faction?s name: Tanzim Byman, D., 2011. The Rise of the PLO and the Birth of International Terrorism (1956? 1970s). In A High Price: The Triumphs and Failures of Israeli Counterterrorism. Oxford University Press, pp. 81-82, 118, 120-121, 314. Durns, S., 2017. Backgrounder: Tanzim [online], COMMITTEE FOR ACCURACY IN MIDDLE EAST REPORTING AND ANALYSIS. Available from: https://www.camera.org/article/backgrounder-tanzim/. Fishman, A., 2015. Fatah on the Verge of Eruption. Ynetnews.com, 23 Oct. Available from: https://www.ynetnews.com/Ext/Comp/ArticleLayout/CdaArticlePrintPreview/0,2506, L-4715319,00.html. Gross, J., 2016. Israel Arrests Head of Fatah Armed Wing. The Times of Israel, 15 Feb. Available from: https://www.timesofisrael.com/liveblog_entry/israel-arrests-head-of- fatah-armed-wing/. 60 Kushner, A., 2004. Disclosed: Inside the Palestinian Authority and the PLO. United States: Pavilion Press, pp. 138-141. Marwan Barghouti, Fatah-Tanzim, and the Escalation of the Intifada, 2002, Jerusalem Center for Public Affairs, 1 (16). Available from: http://jcpa.org/article/marwan- barghouti-fatah-tanzim-and-the-escalation-of-the-intifada/. Milshtein, M., 2004. Between Revolution and Statehood: Fatah and the Palestinian Authority (Hebrew). The Moshe Dayan Center. Pina, A. D., 2005. Palestinian Factions. (CRS Report No. RS21235), p. 4. Available from: https://fas.org/sgp/crs/mideast/RS21235.pdf. Rubin, B., 2003. Yasser Arafat: A Political Biography. Oxford University Press, pp. 258-259. Schenker, D., 2000. Inside the Fatah Tanzim [online], The Washington Institute for Near East Policy. Available from: https://www.washingtoninstitute.org/policy- analysis/view/inside-the-fatah-tanzim-a-primer. Usher, G., 2000. Fatah?s Tanzim: Origins and Politics. Middle East Report, (217), 6-7. Faction?s Name: Popular Resistance Committees (PRC) Al-Aqsa Martyrs Brigades (AMB) ? Fatah [online], Mapping Palestinian Politics. Available from: https://ecfr.eu/special/mapping_palestinian_politics/al_aqsa_martyrs_brigades_amb_f atah/. al-Tawheed, L., 2021. Database: Identifiers of Designated Islamic Terrorist Organizations [online], Jihad Intel. Available from: https://jihadintel.meforum.org/group/174/liwa-al-tawheed. BBC 2011. ?Fatah Popular Committees Reject Palestinian Authority?s Dissolution Order.? BBC Monitoring Middle East - Political. Durns, S., 2018. Popular Resistance Committee Backgrounder [online], COMMITTEE FOR ACCURACY IN MIDDLE EAST REPORTING AND ANALYSIS. Available from: https://www.camera.org/article/popular-resistance-committee-backgrounder- 2018/. Halevi, J. D., 2006. The Popular Resistance Committees: Hamas? New Partners? Jerusalem Center for Public Affairs ? Jerusalem Issue Brief, 5 (24). Available from: https://www.jcpa.org/brief/brief005-24.htm. Israeli air strike kills chief of Gaza?s PRC group. Reuters, 18 Aug 2011. Available from: https://www.reuters.com/article/uk-israel-egypt-gaza-militants/israeli-air-strike-kills- chiefs-of-gazas-prc-group-idUKTRE77H50120110818. 61 Newton, C., 2015. The Rocket Men Who Start Gaza?s Wars. The Daily Beast. Available from: https://www.thedailybeast.com/with-irans-rocket-men-who-start-gazas-wars. Palestinian Terror Groups: Popular Resistance Committees (PRC), Jewish Virtual Library. Available from: https://www.jewishvirtuallibrary.org/popular-resistance- committees-prc. Popular Resistance Committees ? Palestine, Terrorism Research & Analysis Consortium. Available from: https://trackingterrorism.org/group/popular-resistance- committees-palestine/. The Popular Resistance Committees: Portrait of the Terrorist Organization Responsible for the Series of Combined Terrorist Attacks North of Eilat, Israel?s Southernmost City, 2011. The Meir Amit Intelligence and Terrorism Information Center. Available from: https://www.terrorism-info.org.il/Data/pdf/PDF_11_182_2.pdf. Ya?ari, E.. 2006. ?Collision Course,? The Jerusalem Report. Faction?s name: ?Abd al-Qadir al-Husayni Brigadas Mitchell, G. R., 2018. Face to face with the armed masked men of the Gaza Strip. The Press and Journal, 4 March. Available from: https://www.pressandjournal.co.uk/fp/uncategorized/1426717/face-to-face-with-the- armed-masked-men-of-the-gaza-strip/. News of Terrorism and the Israeli-Palestinian Conflict, 2017. The Meir Amit Intelligence and Terrorism Information Center. Available from: https://www.terrorism- info.org.il/Data/articles/Art_21202/E_088_17_1359585666.pdf. Faction?s name: Shaheed Nabil Masoud Unit Regular, A., Liss, J., & Hasson, N., 2011. The IDF assassinated two other terrorist operatives in the Gaza [in Hebrew]. Haaretz, 29 Aug. Available from: https://www.haaretz.co.il/misc/1.1065059. Faction?s name: The Nidal al-?Amoudi Battalion Fatah-linked Gaza group shows off mortar shell production. The Times of Israel, 6 Feb 2018. Available from: https://www.timesofisrael.com/fatah-linked-gaza-group-shows- off-mortar-shell-production/. Sones, M., 2019. Gaza Fatah splinter group performs simulated IDF kidnapping. Arutz Sheva, Available from: http://www.israelnationalnews.com/News/News.aspx/257852. 62 Faction?s name: The Ayman Juda Brigades Nahmias, R., 2009. Egyptians mulling death sentence for head of terror cell. Ynetnews.com, 19 April, Available from: https://www.ynetnews.com/articles/0,7340,L- 3702959,00.html. Faction?s name: Al-Mustaqbal Barghouti apologizes for Fatah mistakes, failures. The Jerusalem Post, 30 Dec 2021. Available from: https://www.jpost.com/Middle-East/Barghouti-apologizes-for-Fatah- mistakes-failures. Bicakci, S., 2007. The Palestinian Parliamentary Legislative Elections 25 January 2006. Journal of Southern Europe and the Balkans, 9 (1), 65-78. Confusion in Fatah after Primaries, Final Round of Palestinian Municipal Elections, Judicial Law Overturned [online], 2008. Carnegie Endowment for International Peace. Available from: https://carnegieendowment.org/2008/08/20/confusion-in-fatah-after- primaries-final-round-of-palestinian-municipal-elections-judicial-law-overturned-pub- 21061. Ellingwood, K., 2005. Times Fatah Split Could Benefit Hamas in Palestinian Poll. Los Angeles Times, 16 Dec. Available from: https://www.latimes.com/archives/la-xpm- 2005-dec-16-fg-fatah16-story.html. Erlanger, S., 2005. Main Palestinian Faction Splits Sharply Ahead of Election. The New York Times, 15 Dec. Available from: https://www.nytimes.com/2005/12/15/international/middleeast/main-palestinian- faction-splits-sharply-ahead-of.html. Fatah rebels insist on split, Al-Jazeera, 15 Dec 2005. Available from: https://www.aljazeera.com/news/2005/12/15/fatah-rebels-insist-on-split. Fatah splits before key election. BBC News, 15 Dec 2005. Available from: http://news.bbc.co.uk/2/hi/middle_east/4530206.stm. Fishman, B., & Yogi, M., 2006. Fatah?s ?young guard? faces a multitude of old problems. The Daily Star. Gutman, M., 2005. Fatah members warn of war within party. The Jerusalem Post, 30 Dec. Available from: https://m.jpost.com/middle-east/fatah-members-warn-of-war-within- party/amp/. Rumley, G., & Tibon, A., 2017. The Last Palestinian. Prometheus Books, p. 119. Yaghi, M., 2009. Fatah Congress: A Victory for Abbas, The Washington Institute for Near East Policy. Available from: https://www.washingtoninstitute.org/policy- analysis/view/fatah-congress-a-victory-for-abbas. 63 Faction?s name: Al-Ahrar BBC Monitoring Middle East, 2009. ?Egyptian website interviews leader of new Fatah faction Al-Ahrar? Faction?s name: The Al-Mujahideen Brigades BBC 2011. ?Palestinian Al-Mujahidin Brigades Group Forms Political Wing.? A Look at the Gaza Attack Tunnels of the Al-Mujahideen Brigades [online], 2018. The Algemeiner. Available from: https://www.algemeiner.com/2018/02/19/a-look-at-the- gaza-attack-tunnels-of-the-al-mujahideen-brigades/. A network of former Fatah operatives in the Gaza Strip (Al-Ahrar) supported by Hamas announced the establishment of a military wing [online], 2010. The Meir Amit Intelligence and Terrorism Information Center. Available from: https://www.terrorism- info.org.il/en/18051/ [Accessed 22 March 2021]. Shin Bet Nabs Palestinian Suspected of Recruiting Terrorists in Cairo, The Times of Israel, 6 March 2016. Available from: https://hamodia.com/2016/03/06/shin-bet-nabs- terrorist-who-recruited-palestinians-in-egypt/. Faction?s name: Fatah ? Democratic Reform Bloc Abu Toameh, K., 2018. PA stages ?campaign of arrests? against Fatah dissidents, critic. Times of Israel, 28 March. Available from: https://www.timesofisrael.com/pa-stages- campaign-of-arrests-against-fatah-dissidents-critics/. Fatah ? Democratic Reform Bloc [online], Mapping Palestinian Politics. Available from: https://www.ecfr.eu/mapping_palestinian_politics/detail/fatah_democratic_reform_blo c Issacharoff, A., 2016. Palestinian leadership may choose Abbas?s successor, and tear itself apart. The Times of Israel, 20 Nov. Available from: https://www.timesofisrael.com/palestinian-confab-may-choose-abbass-successor-and- tear-itself-apart/. Rasgon, A., 2015. Fatah Central Committee Profiles, The Washington Institute for Near East Policy, p. 5. Available from: https://www.washingtoninstitute.org/policy- analysis/fatah-central-committee-profiles. 64 Questions for interviews with Fatah junior cadres First theme: General (1) Why have you decided to become involved in Palestinian politics? (2) Have you always seen yourself involved in politics? (3) What made you become politically active/ involved? Was there any specific event that affected your decision to be involved politically? (4) What are your main/ most important recollections from your early years in politics? What were you striving to achieve? (5) Why did you join Fatah and not other Palestinian organizations? Second theme: Why Factionalize? (1) Why did the Palestinians need an organization like the AMB? Wasn?t Fatah good enough? (2) Can you tell me, with as many details as you can (in terms of people, times, and places) what you think has led supporters of Fatah to form or join the AMB? (3) Was the AMB founded overnight or was it a process? (4) What has made someone like Marwan Barghouti (leader of Tanzim and AMB) so attractive to many Fatah men? (5) Was Barghouti?s leadership more about his ideology or actions? (6) Do you think Oslo was a mistake? What were your views about the peace process 20 years ago? (7) At that time (mid-1990s, 2000), what were you thinking about Yasser Arafat? About his leadership style? The Oslo process? (8) Why was there a need for two separate organization like the AMB and Tanzim? Where is Arafat in the story? Did he like the idea of having these groups? Did it serve his goals in any way? (9) The AMB and Tanzim?s activity occurred even under the iconic leadership of Yasser Arafat. How was that possible? I was certain everyone would have been 100% unified around him. 65 (10) Why is it that with Abbas, who is far less charismatic than Arafat, groups like AMB and Tanzim are inactive? (11) Was the creation of AMB something that was developing before the outbreak of the intifada? (12) If the interviewee points to several years before 2000, ask: why do you think it only occurred in 2000? If he refers to Israel/peace process, ask him: were there any additional reasons? Say, within Fatah? Among Fatah leaders? Supporters? (13) What was the feedback/ response you got from Fatah?s senior leadership? Were there people in Fatah who were trying to curtail the activity of the AMB? Why? What were their arguments? (14) Did AMB organizers believe the faction would hold over time? (15) Has its structure changed over time? What caused the change? Third theme: Grievances: (1) What was it specifically about Fatah that people, in particular Fatah supporters and members, did not like? How did that affect the decision of these individuals to form/ join AMB and other armed factions before/ during/ after the intifada? (2) In your view, how much of the decision of Fatah supporters to form/ join AMB and start Tanzim was about:19 (a) Leadership-cadre disputes: Their frustration with Fatah?s leadership? (b) Access to decision-making: Frustration from their inability to influence Fatah/ direction of the movement? (c) Reform: Was it about trying to push Fatah leaders to reform the organization to have better governance? (d) Greed: Was it about pushing Fatah leadership to get them jobs? (e) External: Was it about the Palestinian leadership?s response to Israel? Israeli pressure/ oppression? Fourth theme: Who Factionalized? (1) If you had to describe the profile of an individual who joined the AMB, how would you describe him? For example, his affiliation with Fatah? Opinion about Fatah? Position in Fatah (e.g., supporter, member, activist, employee etc.)? Age? 19 I presented to the interviewee one reason at a time and wait for his reply. When needed, I asked follow up questions to extract as many details and examples as possible. 66 From a refugee camp? Town? Village? Occupation (i.e., employed, unemployed? If employed, what is his profession?)? Education? (2) Can you identify/ point towards something (s) they have all have in common? 67 Rebel Socialization, Military Training, and 2 Group Cohesion Abstract Why do rebel groups undergo fragmentation? I argue that the solution to this question lies in rebel socialization?specifically, military training (MT). MT increases group cohesion by strengthening horizontal bonds among combatants; vertical bonds between combatants and commanders; and members? institutional bonds to the organization?s overall mission and esprit de corps. Members become mutually dependent, thus making splintering more costly and fragmentation less likely. I test this argument on a global sample of 83 rebel groups active between 1989 and 2010. This study finds that rebel groups that have recently conducted MT are less likely to fragment by about 75 percent. This finding contributes to our understanding of rebel socialization and rebel group fragmentation. 68 2.1 Introduction For all rebel groups, the risk of fragmentation is omnipresent. Creating a coherent unit is a challenge that all rebel groups face (Cohen, 2017; Nagel, 2021). Because rebel group fragmentation increases the likelihood of civil war incidence (K. Cunningham, 2013), shapes the intensity and duration of fighting (D. Cunningham, 2006; K. Cunningham et al., 2012; Findley & Rudloff, 2012), and the likelihood and durability of peace (Nilsson, 2010; Driscoll, 2012; Rudloff & Findley, 2016; K. Cunningham, 2017), it is intrinsically relevant to and consequential for policy-making. Why, then, are some rebel groups more successful in fostering internal cohesion, while others splinter during intrastate conflict? Existing work on rebel fragmentation presupposes, albeit tacitly, that cohesion is the baseline for how we understand the internal dynamics of rebel groups. Rebel groups, in other words, begin as relatively cohesive organizations which, over the course of a conflict, deviate from cohesion. Indeed, some studies go as far as defining rebel group fragmentation as a process by which the rebel group ??loses [emphasis added] its cohesiveness? (Burch & Ochreiter, 2020, p. 48). Consequently, a growing number of researchers have turned to investigating the causes of rebel group fragmentation, which include preference divergence between leaders and lower-ranking members (Bakke et al., 2012; Shapiro, 2013; Warren & Troy, 2015; Woldemariam, 2018; Burch & Ochreiter, 2020), leadership structure (Asal et al., 2012; Stern, 2022a), ideological differences (Stedman, 1997; Seymour et al., 2016), strategic disagreements (Kydd & Walter, 2002; Nilsson & Kovacs, 2011; Findley & Rudloff, 2012; Duursma & Fliervoet, 2021), state repression (McLauchlin & Pearlman, 69 2012; Staniland, 2012; Schubiger, 2015) and external support (Carter, 2012; Tamm, 2016; Ives, 2019). However, the rarity of rebel fragmentation20 suggests that it might be more fitting to ask, ?What do rebel groups do to prevent fragmentation and foster internal cohesion?? The thought that cohesion, and not fragmentation, should be perplexing for scholars of civil conflict signifies a positive development in the way we approach studying the inner workings of rebel organizations. In this article I propose that the solution to the puzzle of cohesion is located in rebel socialization. The different methods that rebel groups employ in order to socialize fighters into the organization are intended to generate a deeper change in one?s sense of self (Hoover Green, 2016, 2018; Gates, 2017; Cohen, 2017; Checkel, 2017). Rebel socialization is widely understood as a key aspect of insurgencies and, hence, of national security. A considerable body of research has demonstrated that the methods rebel leaders employ to socialize their members into the organization shape patterns and types of violence (Wood, 2009; Hoover Green, 2016, 2018; Cohen, 2017; Cantin, 2021). Although recent literature has begun to point towards the effects of rebel socialization on member retention and cohesion (Oppenheim et al., 2015; Forney, 2015; Gates, 2017; Cohen, 2017; Nagel & Doctor, 2020), the assertion that rebel socialization increases group cohesion and hinders fragmentation has not been tested directly. In attempting to fill this gap, this article argues that, as a form of rebel socialization, military training (MT) is uniquely positioned to foster interpersonal bonds 20 The data used in this study include a sample of 83 rebel groups active in 47 armed, intrastate conflicts between 1989 and 2010. Of the 588 rebel-year observations, fragmentation occurred only 27 times, or in 4.5 percent of the observations. 70 among combatants, between combatants and commanders at the combat unit level, and between the rebel group members and the overall organizational mission and its esprit de corps. Combined, these bonds, which are cultivated through MT, make splintering more costly. When combatants are socially and professionally tied to one another, the costs of exiting the group increase, and fragmentation is less likely. I contribute to the existing literature on rebel fragmentation through my research on the effects of different methods of rebel socialization on the probability of fragmentation. In particular, I make the theoretical and empirical connection between MT and fragmentation explicit. I stress the need to fully grasp the structure of rebel groups beyond the simple leader-combatant model and identify the unique attributes of MT in fostering organizational cohesion. To test my theory, I estimate a series of binary logistic regression models with a sample of 83 rebel groups active in 47 armed intrastate conflicts between 1989 and 2010. My findings support my argument that rebel groups that conduct MT are less likely to split into multiple groups. This article is structured as follows. First, I review existing explanations of rebel socialization and explain how it helps rebel groups prevent fragmentation and build cohesion. I distinguish between informal and formal methods of rebel socialization and assess each method?s advantages and shortcomings for group cohesion. Second, I introduce a theory of MT and cohesion and consider the internal structure of a rebel group. I utilize insights from the field of military sociology in order to explain how MT fosters rebel group cohesion. Third, I present my research design and the results of my analysis. Finally, I conclude with a discussion of the results and their implications for future research and policy. 71 2.2 Rebel Socialization and Group Cohesion Rebel socialization refers to ?the process through which members adopt the norms and rules of a given community, the endpoint of which is internalization? (Dawson et al., 1977, p. 9; Checkel, 2017, p. 594). Socialization, in other words, is a process whose intended result is not simply behavioral adaptation, but a deeper change in one?s sense of self (Checkel, 2017; H?ppner, 2017). Socialization helps align a combatant?s preferences or, at the very least, his behavior with that of the group (San?n & Wood, 2014). On the group level, socialization facilitates the creation of social bonds where they are absent, strengthens people?s confidence in each other, and generates a sense of collective duty that diminishes attempts to desert or mutiny (Cohen, 2017). Indeed, regardless of how members are recruited, the social integration of these new fighters demands special effort. A rebel group often includes a number of different cleavages within its ranks and the way in which the group decides to manage this variety of actors will affect the likelihood of fragmentation (Cohen, 2017; Burch & Ochreiter, 2020). This raises the double issues of not only having to retain existing fighters but also needing to foster confidence in the organization and to develop social cohesion between older members and their new cohorts (Hoover Green, 2016, 2018; Gates, 2017; Nagel, 2021). Rebel socialization includes a set of formal and informal mechanisms designed to mediate between the group and the fighters? preferences (Hoover Green, 2016, 2018). Informal mechanisms for rebel socialization are unofficial norms and practices that may contribute to the building of cohesion between group members (Checkel, 2017; Cohen, 2017). They are neither formally ordered by group leaders, nor designed to regulate 72 members? behavior, nor are they implemented organizational policies (Wood & Toppelberg, 2017). Rather, informal socialization mechanisms are mostly restricted to the combat unit level. They develop through peer learning and imitation (Rogoff et al., 2003; Rodgers & Muggah, 2009; Checkel, 2017) and may include hazing and other forms of rituals (Checkel, 2017; Wood & Toppelberg, 2017; Cohen, 2017). Violent forms of hazing have been found to create high costs of entry to a group and create more committed recruits (Gerard & Mathewson, 1966; Cohen, 2017). For example, in the LRA, new recruits engaged in violence against their home village, purposely raising the cost of deserting and returning home, as well as creating an insider versus outsider identity (Gates, 2017). In other cases, the LRA socialized new recruits through spiritual ceremonies of cleansing, rebirth (Gates, 2017), or even forced marriage (Baines, 2014). Group violence against rivals or other group members has been found to be useful for the development of internal cohesion (Wood & Toppelberg, 2017; Cohen, 2017; Gates, 2017). Rebel groups may incentivize violent behavior in order to integrate new recruits (Humphreys & Weinstein, 2008) and engender a sense of belonging among members (Goldman et al., 2014; Littman & Palluck, 2015; Rodgers, 2017). Those who fail to engage in violent behavior may be punished, shamed, or outcast by their peers. Moreover, violence is a particularly strong signal for newer and lower-ranking members who seek recognition, respect, trust, and status from other group members (Littman & Palluck, 2015; McLauchlin, 2015). Hence, individuals may often choose to partake in group violence over continued estrangement from their peers (Cohen, 2017). 73 For example, as a stigmatizing and public act of violence, the perpetration of sexual violence promotes cohesion because group members that engage in acts of sexual violence prove their willingness to pay the price for being a member and develop a sense of commitment to one another (Cohen, 2013, 2016, 2017; Wood, 2009, 2018; Nagel & Doctor, 2020; Nagel, 2021). Rape poses a direct health risk to the perpetrators ? as any contracted sexually transmitted diseases would most likely go untreated during conflict ? but because of this, it doubles as a signal of loyalty and a willingness to undergo personal peril for the sake of being a part of the group, making it an especially useful device for building cohesion (Cohen, 2017). Because gang rape communicates standards of masculinity, virility, and brutality, as well as mutual regard and commitment, it is used as means for creating and sustaining armed groups (Franklin, 2004; Cohen, 2017). Indeed, research shows there is a greater chance of gang rape spreading within a rebel group when it is undergoing episodes of low levels of cohesion, such as during an influx of new recruits (Cohen, 2017; Checkel, 2017; see also Doctor, 2021). However, violent socialization is likely to have a detrimental effect for combat units that remain stagnant for a relatively long time (Cohen, 2017; Nagel, 2021). Moreover, there is scant evidence of leaders commanding their constituents to participate in sexual violence or of groups viewing it as an element of a tactical repertoire (Cohen, 2016; Wood, 2018; Nagel & Doctor, 2020). It remains unclear whether fighters commanded, urged, authorized, or simply tolerated sexual violence, nor is it possible to say definitively what the reasoning behind sexual violence is (Cohen, 2017; Wood, 2018; Nagel & Doctor, 2020). In fact, research indicates that because 74 informal socialization methods compromise official policies and values, group leaders may take action to prohibit them (Wood & Toppelberg, 2017). Therefore in this article I focus on formal socialization mechanisms. Formal socialization mechanisms are typically top-down processes designed to foster fealty to the greater organization (Hoover Green, 2016, 2018; Gates, 2017; Wood & Toppelberg, 2017). They primarily include selective recruitment, political education (PE), and military training (MT) (Hoover Green, 2016, 2018; Gates, 2017; Checkel, 2017). I argue that when cohesion is apparent, it is probably due to how rebel groups formally socialize fighters into the organization. More specifically, whereas the end goal of all socialization methods is identical ? that is to foster group cohesion ? differences in each method?s inner mechanisms may yield variations in group cohesion. I contend that compared to other forms of rebel socialization ? particularly, selective recruitment and PE ? MT is uniquely positioned to foster group cohesion. MT, which includes drills exercises as well as the cultivation of comradeship, plays an important role in member socialization and is directly associated with internal group cohesion (King, 2006, 2007, 2013; Siebold, 2007, 2011). 2.2.1 Selective Recruitment All rebel groups must recruit members. For the recruited, recruitment introduces them to the group?s expectations and social customs (Hoover Green, 2018). For the group, the necessity of recruitment presents a tradeoff. On the one hand, recruitment is inherently an issue of survival for rebel groups (Faulkner & Doctor, 2021). Without recruiting a sufficient number of individuals, a rebel group cannot overcome the 75 collective action problem, survive, evolve, defeat or extract concessions from the government (Weinstein, 2007; Mosinger, 2019; Lewis, 2020). A significant caveat to recruitment, however, is associated with recruits? tendency to misrepresent their true motivations for fighting by holding their reasons for joining the rebel group private (Weinstein, 2007; Humphreys & Weinstein, 2008; San?n & Wood, 2014). Recruiting opportunistic or uncommitted individuals can produce a rebel group made up of individuals with dissimilar inclinations who may prefer to split from the larger group rather than submit to strategies and goals with which they disagree (Burch & Ochreiter, 2020; Faulkner & Doctor, 2021). Thus, by attempting to screen for the most committed recruits, selective recruitment is a method of socialization intended to account for the risks of adverse selection (Weinstein, 2007; Faulkner & Doctor, 2021). By selectively recruiting the right kind of individuals, rebel leaders try to ensure that new members are committed to the group?s cause and uncommitted ones are screened out, thereby reducing the likelihood of splits and buttressing group cohesion (Weinstein, 2007; Shapiro, 2013; San?n & Wood, 2014; Forney, 2015; Eck, 2014; Hoover Green, 2018; Faulkner & Doctor, 2021). For example, rebel groups that recruit from excluded ethnic groups have stronger collective solidarity (Wucherpfennig et al., 2012) compared to more ethnically diverse rebel groups (Humphreys & Weinstein, 2006). Consequently, rebel groups that vet members based on their ethnicity are less likely to experience splits (Gates, 2002; Fjelde & Nilsson, 2018). Rebel groups that screen new recruits are also better positioned to control and to police noncooperation (Kalyvas, 2006; Fjelde & Nilsson, 2018). Less selective or forced recruitment may help rebel groups inflate their ranks rapidly. 76 However, compromising on the vetting of applicants tends to result in groups of individuals who often had no choice but to join (Cohen, 2017; Gates, 2017), leading to a higher probability of fragmentation (Faulkner & Doctor, 2021). In El Salvador, the Frente Farabundo Marti de Liberacion Nacional (FMLN) focused on the recruitment of ideologically committed individuals (Peterson, 1996; Wood & Jean, 2003; San?n & Wood, 2014). The ability of Hamas?s leaders to carefully screen recruits has contributed to the development of impressive organizational cohesion (Abu-Amr, 1993; E. Berman, 2009). Similarly, in Algeria, the AIS recruited from a network of local mosques, which provided a vast recruitment base for the organization (Ruedy, 1992, p. 253; Kepel, 2002, pp. 167, 169; Hill, 2009; Hafez, 2000). Common to the FDLR in the Democratic Republic of the Congo (DRC), the LRA in Uganda, and LURD in Liberia is their heavy reliance on forced recruitment (Doom & Vlassenroot, 1999; Pugel, 2010; Cohen, 2017; Gates, 2017). All three groups have also experienced fragmentation. A study comparing voluntary and abducted recruits in the DRC found that voluntary recruits felt a greater sense of autonomy over their own lives and saw the group as less threatening than their abducted counterparts (Hecker et al., 2013; Cohen, 2017). Whereas Palestinian Fatah has never forced recruitment, the group was not particularly selective in terms of members? recruitment either. According to a former member of Fatah, recruits ?came from all walks of life. Throughout the years, students, teachers, refugees, white collar professionals, Islamists, and guerilla fighters have all joined Fatah? (personal communication, September 10, 2018). Fatah?s relatively lax recruitment practices are a factor that has potentially contributed to the group?s multiple instances of fragmentation (Stern, 2022a). 77 Even when rebel leaders engage in selective recruitment, their inability to observe combatants? underlying preferences hinders their capacity to screen for qualified recruits and risks fragmentation (Nagel & Doctor, 2020). Consequently, the effects of recruitment methods are heavily supplemented by other socialization practices, which occur after recruitment (Cohen, 2016; Hoover Green, 2018; Nagel & Doctor, 2020). 2.2.2 Political Education (PE) PE is the ??formal instruction that explains specific social or political purposes of a particular conflict, and connects conflict purposes to specific behavioral norms? (Hoover Green, 2016, p. 624). It is a classic form of socialization that many rebel groups use to indoctrinate their members (J. Herbst, 2000; Weinstein, 2007, p. 136; Hoover Green, 2016, 2018; Gates, 2017). PE promotes cohesion by fostering members? identification with group goals (Bartov, 1992; Mackenzie, 1992; Posen, 1993; Edwards, 2005; Weinstein, 2007; Oppenheim et al., 2015; Gates, 2017). Existing research shows that rebel groups with PE programs tend to be more cohesive (Oppenheim et al., 2015; Fjelde & Nilsson, 2018; Walther & Pedersen, 2020). Evidence shows that rebel leaders often perceive PE as a significant galvanizer for their members (J. Herbst, 2000). For example, the Eritrean Peoples? Liberation Army, Algeria?s FLN, South Africa?s ANC and PAC, Zimbabwe?s ZANU and ZAPU, and the FMLN have all invested heavily in the political indoctrination of their soldiers and adherents, reflecting the notion that their leadership did not see the costs of indoctrination as a hollow endeavor (Pateman, 1998, p. 126; J. Herbst, 2000; Hoover 78 Green, 2018). In both Colombia and Nepal, rigorous ideological indoctrination by FARC and Maoist rebels, respectively, was an important factor in reducing combatants? desertion rates (Ortiz, 2002; Oppenheim et al., 2015; Eck, 2010). Whereas PE is an important method to cultivate internal cohesion, it has several serious shortcomings. First, PE is a costly activity. It is both time-consuming and labor- intensive (Weinstein, 2007, p. 52; Eck, 2014; Gates, 2017). Under the stress of war, rebel groups are exponentially constrained in terms of the time and resources they can devote to educational activities. Consequently, their priorities are likely to lie with activities that are more directly related to the battlefield. For example, as the war between the Nepalese government and Maoist rebels intensified in the early 2000s, time devoted to PE decreased significantly (Cowan, 2010; Eck, 2014; Gates, 2017), leading to diminished levels of combatant retention (Gates, 2017). Second, PE is theoretical and does not involve practical application. Rebel leaders have little way of knowing whether fighters have internalized the PE materials instructed them, and members have incentive to misrepresent their level of ideological internalization. Under such circumstances the possibility of deviance remains relatively large. Third, while a common ideology may encourage group members to trust each other, it may also designates enemies within the rebel group, resulting in divisions (McLauchlin, 2020, pp. 3-4). Relatedly, PE may open a Pandora?s Box of ideological debates, which can create divisions among group members (Seymour, 2014; Seymour et al., 2016; Gade et al., 2019). Finally, circumstances may alter ideological convictions (Seymour, 2014). Indeed, it is often the case that preferences drive ideology and not the other way around (Christia, 2012). 79 Instances of fragmentation therefore reflect shifts in the ideological preferences of factional leaders (Seymour, 2014). 2.2.3 Military Training (MT) Existing studies on rebel group socialization often reduce MT to the technical operation of weapons and the transformation of ordinary individuals into instinctive killers (cf. Hoover Green, 2016, 2018; Gates, 2017). I argue, however, that as a socialization method, MT is uniquely positioned to foster group cohesion. It is during MT that individuals begin to identify themselves internally as members of the group. Considering the violent nature of civil wars, comparatively little scholarly attention has been given to the role MT plays in inhibiting rebel fragmentation and fostering group cohesion. In the next section, I develop an argument of how and why MT is uniquely positioned to foster internal cohesion. 2.3 A Theory of Military Training and Cohesion In this section, I articulate an argument on why MT increases internal cohesion and decreases the likelihood of rebel group fragmentation. I first describe the organization of rebel groups. I distinguish between three organizational levels: leaders, commanders, and combatants. I briefly explain how the dynamics in each level is associated with fragmentation. I then draw on studies in military sociology to discuss the relationship between MT and cohesion. Finally, I explain horizontal, vertical, and institutional bonding in a rebel group and theorize how MT impacts interpersonal and professional bonds on all three levels, resulting in increased levels of organizational cohesion overall. Building on existing works on military cohesion, I define ?cohesion? as the process of associating combatants and commanders together in such a way as to sustain 80 their will and commitment to each other, group leaders, and the larger rebel group to which they belong, despite combat or mission-induced stress (Siebold, 2007; Van Epps, 2008). It is true, however, that MT is not the only socialization process that generates cohesion. Group cohesion is often developed over the long, uneventful stretches of time outside training and combat, during which soldiers exchange stories about their private lives (Henderson, 1985; Wong et al., 2003; Siebold, 2007). However, such informal interactions are difficult to observe, whereas actual MT is relatively easier to capture. Therefore, this study also understands cohesion as the product of MT, over the course of which soldiers coordinate their actions ever more closely (King, 2006, 2007, 2013). 2.3.1 The Internal Structure of Rebel Groups Recent studies on rebel groups have moved beyond the basic framework of the principal-agent model and emphasize a more elaborate distinction between leaders, commanders, and combatants (cf. Staniland, 2014; Hoover Green, 2018; Ives, 2019; Nagel & Doctor, 2020; Doctor, 2020). A leader21 is at the top of the rebel group (Prorok, 2016; Ives, 2019; Nagel & Doctor, 2020; Doctor, 2020). He issues the final verdict in the organization with respect to political goals, ideology, strategy, tactical selection, funding, commanders, and combatants (Prorok, 2016; Lidow, 2016; Nagel & Doctor, 2020; Doctor, 2020). The way in which rebel leaders arrange their fighting force will have significant ramifications for the group?s ability to persist as a cohesive unit (Staniland, 2014; Burch 21 Leadership structure varies across rebel groups. For example, in Fatah, the top leadership has historically coalesced into an executive committee (Stern, 2022a). 81 & Ochreiter, 2020). Rebel leaders are those who initially mobilize rebellion, unify all recruits, and manage the armed forces, and as such, they are instrumental in affecting the conduct and durability of their groups in combat situations (Weinstein, 2007; Staniland, 2014; Doctor, 2020). Leaders provide a general sense of purpose and meaningfulness that is directly linked to the cause, for which individuals join the rebel group to begin with. All else equal, to develop and maintain cohesion, leaders? and members? preferences need to align (Shapiro, 2013). The ideology, recruitment strategies, training, and political objectives that leaders set are constitutive of the social integration of members with the larger rebel group. Leaders? conduct and decisions affect the credibility of the organization in the eyes of their subordinates. Rebel groups might develop around a particular individual who becomes the group?s leader (J. Braithwaite & Cunningham, 2020). In some cases, the leader?s aura is the reason individuals join the rebel group. According to a member of Fatah, for instance, the main reason he joined the group was the charismatic personality of Fatah?s venerated leader, Yasser Arafat (personal communication, March 5, 2020). Leadership disunity and disputes between leaders and their subordinates are potential sources of fragmentation (Asal et al., 2012; Ives, 2019; Stern, 2022a). A commander is situated at an intermediate level of authority in the rebel group, between the rebel leader and the footsoldiers. Commanders are responsible for the daily management of the fighting forces. Because commanders are those in charge of training combatants and leading them in war (Lidow, 2016; Hoover Green, 2018; Nagel & Doctor, 2020; Nagel, 2021), they are considered an integral part of the combat unit 82 (Siebold, 2007). Typically, combatants know their commander on a closer level ? as nuanced individuals with a unique backstory, identity, and personal characteristics, beyond the name, title, or rank they hold (Siebold, 2007). Their responsibilities and, critically, their frequent personal interactions with their combatants often create strong bonds between the two. This may result in strong loyalty to their commander (Cockerham, 1978; Siebold, 2007; Christia, 2012; Ives, 2019; Nagel & Doctor, 2020), with adverse implications for group cohesion. Strong bonds between commanders and their combatants can become a source of opposition to the greater organization (Wesbrook, 1980; King, 2013, p. 73; Nagel & Doctor, 2020). Indeed, the greater number of fragmentation incidents occur when disgruntled commanders lead loyal combatants out of the rebel group (Ives, 2019; Nagel & Doctor, 2020; Stern, 2022a). Finally, combatants or footsoldiers constitute the lowest strata of the group. Combatants, however, are not actors who reflexively respond to external stimuli. Alternatively, combatants are autonomous actors who have the agency (Shapiro & Siegel, 2012; Shapiro, 2013; Abrahms & Potter, 2015) to decide on their present and future affiliation with the rebel group (Shapiro, 2013; Stern, 2022a). In fact, combatants are often those new recruits whose preferences the rebel command try to align with group objectives (Christia, 2012; Shapiro, 2013; Hoover Green, 2016, 2018) in order to foster cohesion (Wood, 2009; Gates, 2017). 83 2.3.2 How Military Training Builds Group Cohesion? Insights from Military Sociology Broadly speaking, studies in military sociology introduce two competing bodies of literature to explain how MT fosters group cohesion. The first body of literature argues that MT generates collective sentiment between soldiers and between soldiers and their officers through high levels of interpersonal interactions. Frequent interactions during MT produce deep feelings of trust, solidarity and comradeship, which motivate members to cohere and ?stick together,? even under extreme conditions (Shils & Janowitz, 1948; Henderson, 1985; Griffith, 1989; Wong et al., 2003; Kolditz, 2006; Wong, 2006; Siebold, 2007, 2011). Decisively, these emotive bonds can alter individual preferences and motivate the combatant to behave in ways in which he would otherwise not (Lott & Lott, 1965; King, 2013, p. 29). Indeed, scholars who associate between MT, personal bonds, and cohesion attest to a large compilation of anecdotal evidence, in which soldiers personally relate that loyalty to their ?buddy? was integral to their actions (Shils & Janowitz, 1948; Strachan, 2006; King, 2013, p. 30). This view, therefore, emphasizes small group dynamics between soldiers and between soldiers and their immediate officers. Accordingly, military organizations are conceptualized as systems of small groups ? commonly termed ?primary groups.? Each primary group is integrated into the broader organization through collective experience of shared procedures, practices, and norms (Shils & Janowitz, 1948; Griffith, 1989; Wong et al., 2003; Kolditz, 2006; Wong, 2006; Siebold, 2007, 2011). Small 84 groups generate pride, discipline, and internal cohesion and allow for relatively easier monitoring of members? behavior and outcome (P. Herbst, 1962). The second body of literature does not dismiss the importance of interpersonal relations in fostering cohesion (Moskos, 1975; King, 2013, pp. 32, 35). However, it argues that individuals can cohere even if they do not share emotional bonds (Moskos, 1975). The basis of cohesion is located in the very formal practices of MT, such as collective drills (Cockerham, 1978; King, 2006, 2013; MacCoun et al., 2006). Specifically, cohesion emerges through unified MT which the armed forces utilize to cultivate cohesion and lessen the ubiquitous possibility of deviance under the pressures of combat (King, 2007; King, 2013, p. 36). Training cannot be reduced to the technical skills transferred from officers to individual soldiers. In fact, the goal of training is to suppress the individual and prioritize the group. During MT, combatants synchronize their conduct more meticulously (King, 2006, 2007, 2013; Ben-Shalom et al., 2005; MacCoun et al., 2006). Improved coordination between soldiers reduces the likelihood of individual misinterpretation and deviance and increases cohesion (King, 2006). By repeatedly training together, fighters realize their responsibility to one another (King, 2007). In addition, the detailed orders and protocols that characterize MT and anticipate battle are purposely engineered to reduce the consequences of any potential discord on the battlefield (King, 2007). Drills enable combatants to respond to threats that come their way in unison, thus lessening the possibility of fragmentation (King, 2013, pp. 272-273). Training aims at generating and enhancing cohesion from the level of the combat unit to the larger military organization (King, 2013, pp. 272-273). The routinization of 85 MT, which includes multiple small and large decision-making, helps align members? preferences with those of the group (Festinger & Carlsmith, 1959; King, 2013). Despite their differences, important commonalities exist between the two arguments presented above. While proponents of the first argument emphasize interpersonal relationships, they also acknowledge the central role that collective MT plays in fostering comradeship and cohesion (cf. Siebold, 1999; Siebold & Lindsay, 1999; Siebold, 2007). Similarly, proponents of the second argument do not dismiss the role emotional bonds play in group cohesion, but posit that MT is essential for those bonds to emerge. Crucially, both mechanisms reach an identical conclusion: increased levels of MT foster higher group cohesion (Kenny, 2011). Rebel groups use MT as a way to foster long-term identification with group goals and improve combatant retention rates (Wood, 2009; Gates, 2017). In Sierra Leone, the RUF?s MT enhanced group identity and solidarity, helping the group to sustain its war against the government for over a decade (Gates, 2017; Checkel, 2017). The LRA organized itself in small group operations of about 10 to 15 soldiers, the size of a squad in the US army (Gates, 2017). The small unit structure reflects the organizational advantages seen in modern militaries, serving to create an effective bond between soldiers (cf. Shils & Janowitz, 1948; P. Herbst, 1962; Kolditz, 2006; Wong, 2006; Siebold, 2007, 2011; King, 2013). Conversely, rebel groups that did not invest in MT faced a higher likelihood of fragmentation. For instance, LURD was often described as a poorly trained and incapable military force (Hazen, 2013, p. 116). In 2003, LURD members splintered from the group and formed MODEL (J. Braithwaite & Cunningham, 2020). 86 I argue that compared to other forms of rebel socialization ? specifically, selective recruitment and PE ? MT is uniquely positioned to foster cohesion. Specifically, MT generates bonding on all three levels of the rebel group: horizontal bonding, that is, interpersonal bonds among combatants; vertical bonding, that is, interpersonal bonds between combatants and commanders at the combat unit level; and institutional bonding, that is, members? bonds to the overall organizational mission and its esprit de corps, as generated by group leaders. Combined, these bonds make splintering more costly. When combatants are socially and professionally tied to one another, the costs of exiting the group increases, and fragmentation is less likely. Horizontal, vertical, and institutional bonds are critical components of military unit cohesion (Siebold, 2007). Existing research on civil war emphasizes the role that horizontal and vertical ties play in promoting rebel group cohesion and hindering fragmentation (Staniland, 2014; Salverda & Otto, 2014; Ives, 2019). As I explain below, MT not only works to cultivate bonds within each dimension, but also facilitates their interaction. The outcome is greater rebel group cohesion overall. I therefore expect that periods of MT are associated with increased levels of internal cohesion. 2.3.3 MT and Horizontal Bonding Studies on military cohesion define horizontal bonding as social bonds among soldiers (Siebold, 2007). That is, bonding between members of the same authority level. I apply this definition to rebel groups and describe bonding among combatants as horizontal bonding. I argue that rebel groups that provide their members with MT are more likely to increase interpersonal and professional bonds in the combatant-combatant relationship and are therefore less likely to fragment. 87 Specifically, I argue that MT is more likely to increase horizontal bonds between combatants for two main reasons. First, MT opens opportunities for interpersonal bonds to develop. Strong personal commitments to their fellow combatants promote unity under the extremis of combat. Scholars have taken note that the reason soldiers remain steadfast under the extremis of combat is not because of some grandiose idea of ideology or patriotism, but because of their attachment to their ?primary group? of armed associates (Stouffer et al., 1949; Cockerham, 1978; N. Stewart, 1991; Wong et al., 2003; Strachan, 2006; Kenny, 2011). Second, and relatedly, cohesion is highly vulnerable during periods under fire. Under these strenuous and disorienting conditions, combatants are most likely to turn to deviant courses of action, which culminate in the group?s breakdown (Little, 1964; King, 2006; Strachan, 2006; Kenny, 2011). As battlefield losses and casualties soar, rebel fighters may start to question their reason for fighting (Christia, 2012; Salverda & Otto, 2014; Ives, 2019; Nagel & Doctor, 2020). Among rebel groups, splits may occur when the combatants? survival is at stake (Christia, 2012, p. 35). A breakdown in group cohesion will follow as protection-seeking rebels (Staniland, 2012) collectively split from the rebel group in order to survive (Christia, 2012, pp. 43-44). Stronger horizontal bonds increase the costs of fragmentation, which would require combatants to turn their back on a close comrade. In sum, when a combatant develops close personal relations with his co-combatants and when he has confidence in his fellows? ability to ?watch his back? in the battlefield, the costs of exiting the group increase, and fragmentation is less likely. 88 Conversely, the absence of MT reduces the combatants? opportunities to foster interpersonal relations and, subsequently, horizontal bonds. Relatedly, without MT, a combatant cannot develop confidence in his fellows? capability to ?watch his back? on the battlefield. Finally, without MT, the combatants? coordination with one another is hindered, and the likelihood of individual deviance increases, adversely affecting the ability of combatants to remain united in the battlefield. In sum, decreased horizontal bonds tarnish both interpersonal and professional relationship between combatants. Consequently, splintering becomes less costly. If combatants are neither emotionally nor professionally tied to one another, the costs of exiting the group diminish, and fragmentation is more likely. 2.3.4 MT and Vertical Bonding Studies on military cohesion define vertical bonding as social bonds between soldiers and their immediate officers ? that is, bonding across authority levels of authority (Shils & Janowitz, 1948; Siebold, 2007). I apply this definition to rebel groups and describe bonding between combatants and their commanders as vertical bonding. I argue that rebel groups that provide their members with MT are more likely to increase interpersonal and professional bonds in the commander-combatant relationship and are therefore less likely to fragment. Specifically, I argue that MT is more likely to increase vertical bonds between commanders and combatants. Similar to the mechanism that creates strong horizontal bonds, frequent and continued interactions between combatants and their immediate commander during MT foster higher levels of coordination, trust, teamwork, loyalty, and willingness to commit themselves to their co-fighters in the combat unit (Strachan, 89 2006; King, 2006, 2007, 2013; Siebold, 2007, 2011; Kenny, 2011). The maintenance of vertical bonds through MT discourages commanders from motivating their combatants by using less sustainable methods of group cohesion, such as coercion (Gates, 2017; McLauchlin, 2020) and financial gain (Weinstein, 2007). During MT, commanders emerge not only as instructors, but also as co-fighters that share the onerous process of MT and contribute to overcoming shared threats (Kenny, 2011). When combatants develop close personal relations with their commander, and when they have confidence in their commander?s fighting skills and commitment to his subordinates, the costs of exiting the group increase, and fragmentation is less likely. Critically, MT helps narrow preference divergence between combatants, commanders, and the leadership. Because commanders are in charge of training combatants, they can directly observe combatants? behavior and gain crucial insights into their subordinates? behavior. These insights can then be communicated upwards to the leadership which, in turn, depends on their commanders for information regarding the group?s fighting forces (Nagel & Doctor, 2020). In other socialization processes, for example PE, both leaders and commanders have little way of knowing whether combatants have internalized the material instructed them, and the possibility of deviance remains relatively large. Even when individuals are selectively recruited, prospective recruits have an incentive to hold their true motivations for joining the rebel group private (Weinstein, 2007; Humphreys & Weinstein, 2008). During MT, however, combatants demonstrate their commitment to each other and to their commanders. In a study of the Spanish Civil War, McLauchlin 90 (2020, p. 3) argues that difficult training, submission to military discipline, and braving harsh conditions are costly signals that build trust and show the sacrifices rebels are willing to make to fight. Furthermore, MT not only simulates the dynamics of war, but also provides combatants the opportunity to apply what they have learned. Importantly, commanders personally monitor a combatant?s behavior, correct him on the spot, continually coordinate the activities of individual combatants toward a unifying end, and assess a combatant?s overall performance (King, 2007, 2013). A combatant?s performance in MT is a strong signal about his skills, understanding of the instructions, and commitment to the group?s goals, as well as to other members of the combat unit. This strengthens the commander-combatants bond and provides an important source of information concerning combatants? commitment to the group?s objectives, as formulated by the leadership. However, it is precisely close commander-combatant bonds that can undermine overall cohesion and exacerbate fragmentation (King, 2013, p. 32) Strong bonds between commanders and their combatants can become a source of opposition to the greater organization (Wesbrook, 1980; King, 2013, pp. 32, 73). Increased levels of vertical bonds give rebel commanders greater confidence in their ability to mobilize the combatants needed to build a functional rebel splinter faction (Nagel & Doctor, 2020). Indeed, fragmentation often occurs when disgruntled commanders lead a group of loyal combatants out of the rebel group (Christia, 2012; McLauchlin, 2015; Ives, 2019; Nagel & Doctor, 2020; Stern, 2022a). Hence, an additional layer of social bonding ? that of the larger institution ? is necessary to foster overall organizational cohesion. 91 2.3.5 MT and Institutional Bonding Institutional bonding refers to members? bonds to the overall organization (Siebold, 2007). Institutional bonding is where members? trust in the organization lies. Institutional bonding helps glue horizontal and vertical bonds by providing a general sense of purpose and meaningfulness that is connected to the cause of the conflict (Siebold, 2007). Accordingly, more recent models of military cohesion incorporate horizontal and vertical bonding into the wider organization. The outcome is a structure of relations that support organizational cohesion altogether (Siebold, 2007). I apply the logic of institutional bonding to rebel groups and explain how MT fosters institutional bonding. I argue that rebel groups that provide their members with MT are more likely to increase members? attachment to the rebel group and are therefore less likely to fragment. To capture institutional bonding, I speculate that rebel leaders are those who most prominently drive such bonding. How, then, does MT facilitate institutional bonding? To begin, MT assists rebel leaders in narrowing the divergence in preferences between members by stripping away pre-military standards, mindsets, and identities and replacing them with those of the group (Lott & Lott, 1965; King, 2007, 2013, pp. 29, 272-273; Wood, 2009; Hoover Green, 2016). The routinization of MT encourages combatants to internalize and to conform to the group?s norms (Stouffer et al., 1949; Shils, 1957; Henderson, 1985; N. Stewart, 1991; Wong et al., 2003; Strachan, 2006; King, 2006, 2007, 2013; Siebold, 2007, 2011; Kenny, 2011). Conversely, the absence of MT restricts the ability of rebel leaders to align members? preferences with those of 92 the group and, consequently, the likelihood of combatant deviance from group goals increases. Second, MT is a costly socialization process. When leaders allocate resources to MT, both commanders and combatants see their peers as following a leadership that invests in their skills and personal security. From the fighter?s perspective? both the combatant?s and the commander?s ? investment in MT is a strong signal that the relationship between him and the rebel group will be durable, that they cannot be replaced easily, and that the leadership takes the personal safety of its combatants seriously. After all, MT instructs combatants with practical skills that make the difference between life and death on the battlefield. Hence, when rebel groups appropriate time and resources to MT, it demonstrates to the combatants that their labor is not expendable, otherwise the rebel group would have less incentive to invest so heavily in their training (Weinstein, 2007, p. 137; Haer et al., 2011). Finally, MT strengthens institutional bonds by imprinting the significance of the power structure (Fritz, 1995; King, 2006; Kenny, 2011). In doing so, MT captures a central role in the preservation of institutional links both vertically up and down the chain of command and horizontally across service hierarchies (Siebold, 2007). When fighters feel like part of an organization where leaders invest in their skills and safety, the costs of exiting the organization and forming a new rebel group increase, and fragmentation is less likely. In sum, the uniqueness of MT as a socialization process which cultivates rebel group cohesion is derived from its capacity to cultivate bonds on all three levels of the rebel group. Horizontally, MT fosters cohesion among combatants. Vertically, MT 93 fosters cohesion between combatants and their commanders. Institutionally, MT fosters cohesion between leaders and their combat units, resulting in overall organizational cohesion. 2.4 Hypothesis All else equal, periods of MT will be followed by a decreased likelihood of fragmentation. 2.5 Data and Methods I analyze the effect of rebel MT on the probability of rebel fragmentation in armed intrastate conflict. The hypothesis above is composed of two primary variables: MT and fragmentation. The dataset I use to test my hypothesis relies mainly on variables and cases from Nagel and Doctor (2020). In addition, I draw on other existing datasets for relevant variables, as described below. The cross-national data includes a sample of 83 rebel groups active in 47 armed intrastate conflicts against their governments, resulting in 25+ yearly battle deaths between 1989 and 2010. In total, my data contains 588 observations recorded at the rebel-year level. To test my hypothesis, I estimate a series of binary logistic regression models. 94 Table 2.1: Descriptive statistics. Statistic N Mean St. Dev. Min Max Fragmentation 588 0.05 0.2 0 1 Military training (lagged) 587 0.2 0.4 0.0 1.0 Leftist rebels 588 0.1 0.3 0 1 Forced recruitment 410 0.5 0.5 0.0 1.0 Rebel sexual violence 488 0.2 0.4 0.0 1.0 Rebel group size (logged) 551 8.4 1.6 3.4 11.7 Territorial control 569 0.4 0.5 0.0 1.0 Resource extortion 410 0.4 0.5 0.0 1.0 Conflict duration 588 13.6 12.3 0 60 Multiparty conflict 588 0.6 0.5 0 1 External support 588 0.1 0.3 0 1 Conflict intensity 399 1.3 0.5 1.0 2.0 Peace negotiations 578 0.4 0.5 0.0 1.0 2.5.1 The Dependent Variable The dependent variable, rebel fragmentation, captures instances in which a segment of a rebel group formally and collectively exits that rebel group and establishes a new, independent rebel organization (Nagel & Doctor, 2020; Doctor, 2020).22 According to Nagel and Doctor (2020), for an event to qualify as an episode of fragmentation, the new rebel group must be recorded by the UCDP Armed Conflict Dataset (Pettersson & Eck, 2018) or Nonstate Actors Dataset (NSA) (D. Cunningham et al., 2013). The dependent variable is coded 1 if a rebel group experiences fragmentation during a given year and 0 otherwise. Given the rebel-year level unit of analysis, fragmentation is a relatively rare event. It occurs 27 times, or in 4.5 percent of the observations. 22 Whereas the puzzle and main question in this article address rebel group cohesion, what I ultimately observe are instances of fragmentation. 95 Alternatively, we can see that 26.5 percent of the sampled rebel groups fragment at least once, and fragmentation is present in 18 of the 47 armed intrastate conflicts in the data. Figure 2.1 displays the proportion of active rebel groups that have experienced fragmentation (solid line) and the total number of active rebel groups (dashed line), aggregated by year. Figure 2.1: Distribution of rebel fragmentation, 1989 to 2010. 96 2.5.2 The Independent Variable The primary independent variable is military training. MT describes whether a rebel group has been conducting MT in a given year. The covert nature of rebel groups, organizational variation, and scholarly focus on a relatively limited number of groups make data collection on MT particularly challenging. Indeed, no dataset on rebel group MT exists. To measure MT, I use the Non-State Armed Groups (NAGs ver. 2015) Dataset (San-Akca, 2016). The NAGs dataset codes 9 different forms of international support of rebel groups that are engaged in violent conflict between 1946 and 2010: safe havens for members, safe havens for leadership, the existence of a headquarters in a country, training camps, weapons and logistics aid, financial aid, transport of the military equipment and advice, troops, and training. For this study, I use the NAGs?s training variable as a proxy for MT. I aggregate the continuous variable training provided in the NAGs data to create a dummy variable, coded 1 if a rebel group was reported to conduct MT in a given conflict year and 0 otherwise. Like other forms of rebel socialization, the effect of MT on cohesion is not instantaneous. I therefore lagged MT one year. By utilizing the NAGs?s training variable as a proxy, I do not measure MT directly. Rather, the variable captures instances where a rebel group received MT from an external actor or actors.23 Using a proxy is not without qualifications. 23 In this article, MT indicates whether a rebel group was trained by an external actor, not whether they performed training autonomously. Collecting group-year data on whether they conducted MT is beyond the scope of this project. That being said, the mechanism of MT should not be different based on whom conducted the training. 97 The dataset includes cases of rebel groups that did conduct MT; for example, RENAMO, LTTE, and Hezbollah. However, because the training variable is a proxy, this information is not reflected in the dataset. Nevertheless, by testing my hypothesis on cases of rebel groups that most certainly engaged in MT and cases where the group conducted MT but the information is not reflected in the dataset indicates a potential for an even greater effect of MT on cohesion. In other words, a fuller picture through a direct measure of rebels? MT may suggest an even greater impact of MT on cohesion. I decided to focus on the NAGs?s training variable for three main reasons. First, within the NAGs data, training is the closest variable for capturing MT. The existence of, for example, safe havens, is in and of itself insufficient to conclude that MT has been performed. Second, unlike other potential proxies of MT, in particular training camps that are on foreign soil, training?according to the NAGs?s definition?refers specifically to MT that is potentially conducted in the rebels? territory. This implies greater rebel autonomy in deciding to conduct MT. Finally, the argument presented here seeks to explore the effect of MT on fragmentation and not why some rebel groups receive external support in the form of MT while others do not. In other words, it is MT, as a socialization method, that I theorize explains cohesion, not variation in rebel MT, as a form of international support. Figure 2.2 displays the proportion of active rebel groups that have conducted MT (solid line) and the total number of active rebel groups (dashed line), aggregated by year. In the data, 26 of the 83 rebel groups sampled conduct MT. 98 Figure 2.2: Distribution of rebel MT, 1989 to 2010. A cross-tabulation of my two main variables is supportive of my thesis (see Appendix Table 2.2). Of the 27 groups that fragmented, 25 did not conduct MT. Of the 134 group-year cases with MT, two groups or 1.5 percent experience fragmentation. Of the 454 group-year cases with no MT, twenty-five groups or 5.5 percent experience fragmentation. In other words, the rate of fragmentation among rebel groups that conducted MT is nearly four times lower compared with rebel groups that did not conduct MT. 2.5.3 Control Variables and Alternative Explanations I begin by controlling for the main alternative methods of rebel socialization: selective recruitment, PE, and sexual violence. First, differences in recruitment methods are 99 decisive for the internal cohesion of rebel groups (Cohen, 2016). Rebel groups that vet their recruits are more likely to benefit from fighters who are committed to the cause of the organization and are less likely to defect (Weinstein, 2007). Screening for dedicated recruits also helps narrow preference divergence within the group and reduces the likelihood of splits (Shapiro, 2013). Rebel groups that screen new recruits are better positioned to facilitate control over and policing of noncooperation (Kalyvas, 2006; Fjelde & Nilsson, 2018). Conversely, rebel groups that rely on coerced recruitment exhibit lower levels of social cohesion (Cohen, 2017) and may be more prone to fragmentation due to their soldiers? detachment from the group and its objective (Eck, 2014). I use Nagel and Doctor (2020)?s dichotomous indicator to account for rebel groups that abduct fighters to fill their ranks.24 The authors? indicator is based on data from Cohen (2016). Second, I control for leftist rebel groups. Existing research shows that communist PE generates greater control over combatants? behavior (Hoover Green, 2016, 2018) and that leftist rebel groups tend to be more cohesive (Oppenheim et al., 2015; Fjelde & Nilsson, 2018; Walther & Pedersen, 2020). I employ Nagel and Doctor (2020)?s communist binary variable to account for whether a rebel group is a leftist rebel group or not.25 24 This variable measures coerced participation and not whether recruits were screened. Admittedly, a variable that measures selective recruitment directly could have yielded different results. As part of the robustness checks, I replaced the abduction indicator with an indicator that measures whether a rebel group has ever forced recruitment. Results are robust to this test (see Appendix Table 2.8). 25 The underpinning assumption which has guided me to use the leftist rebel group variable is based on studies which indicate that leftist rebel groups are more likely to carry out PE programs. However, the variable does not measure whether a group conducted PE programs or not. The fact that leftist rebel groups are not more or less cohesive, does not mean that PE does not contribute to cohesion. Moreover, there can be cases of religious, nationalist or other 100 Third, I control for rebel groups that use sexual violence as a cohesion-building mechanism (Nagel & Doctor, 2020; Nagel, 2021). Multiple studies argue that disunity between fighters increases the likelihood of rape because it facilitates the creation of social bonds between combatants (Cohen, 2013, 2016, 2017; Wood, 2009, 2018). To control for sexual violence, I use Nagel and Doctor (2020)?s binary variable, coded 1 if a rebel group was reported to have perpetrated any sexual violence in a given conflict year. It is coded 0 otherwise. The authors? indicator for sexual violence is originally based on the Sexual Violence in Armed Conflict (SVAC) Dataset (Cohen & Nord?s, 2014). I also include controls for a number of group and conflict-level factors that are likely to influence the relationships between MT and fragmentation. At the group level, I first control for rebel group size. Even if it is an imperfect metric, group size is frequently used in both qualitative and large-N statistical analyses (cf. Kalyvas & Kocher, 2007; Akcinaroglu, 2012; Christia, 2012; Krause, 2013; Warren & Troy, 2015; Gade et al., 2019; Mahoney, 2020). I speculate that a larger number of members makes a rebel group more susceptible to fragmentation. Larger groups hinder collective action, so the larger the group is, the more likely it is to fragment (Posner, 2004; Seymour et al., 2016). Moreover, as group size increases, so does the potential for a larger number of preference divergences between group members. Greater preference divergence impedes group cohesion (Shapiro, 2013; Burch & Ochreiter, ideologically-motivated rebel groups that carry out PE programs. Admittedly, a variable that measures PE directly could have yielded different results (e.g., PE as a significant or insignificant driver of cohesion). As part of the robustness checks, I replaced the leftist rebel group indicator with an indicator for whether a rebel group was founded with a specific political ideology or not. Results remain robust to this test as well (see Appendix Table 2.9). 101 2020). Similarly, a larger number of group members is associated with increased opportunities for fragmentation (Christia, 2012). As the number of opportunities for splits increases, the relative cost and associated risks of splintering decrease. However, it is precisely because larger groups can be more difficult to control that larger rebel groups are more likely to conduct MT. MT enables larger groups to control the behavior of their recruits and socialize them into the broader organizational norms and practices better. To control for both possibilities, I take the natural log of rebel armed forces based on the NSA (D. Cunningham et al., 2013). Second, a rebel group?s command structure could influence fragmentation (Sinno, 2011; Asal et al., 2012; Burch & Ochreiter, 2020; Joo & Mukherjee, 2021; Stern, 2022a). Specifically, rebel groups with more centralized command structures are less likely to fragment (Asal et al., 2012). It is theoretically reasonable to assume that MT requires sufficient control from its leadership over its commanders and fighters. In the absence of such control, rebel leaders would find it difficult to amass resources necessary for MT or to convince recruits to go through strenuous MT in the first place. I therefore propose that, compared to rebel groups with decentralized command structures, groups with a centralized command are better structured to decide on and pull the resources required to conduct MT. To control for the level of rebel command structure, I employ D. Cunningham et al. (2013)?s central control variable. This variable is dichotomous. It indicates whether the rebels have a clear central command or not. Third, I control for territorial control. Territorial gains may impact the calculus for different actors within the group (Burch & Ochreiter, 2020), alter incentives for continued cooperation (Woldemariam, 2016), increase the probability of conflict (Lilja, 102 2009; Weidmann, 2009; Fjelde & Nilsson, 2012; Schulhofer-Wohl, 2013; Pischedda, 2018), and serve as the catalyst for fragmentation within rebel groups (Woldemariam, 2018). Alternatively, other research has found that stationary rebel groups are less likely to fragment (Mosinger, 2018). Territorial control enables rebel groups to identify and neutralize potential opposition, thereby deterring defection (Kalyvas, 2006; Seymour, 2014). The logistics of MT require some level of territorial control. Territorial control provides the rebels the necessary training space for military drills and may protect the group from both government counterinsurgency and disturbances from rival rebels (Sinno, 2011). To account for whether a rebel group controls a territory, I incorporate D. Cunningham et al. (2013)?s territorial control variable. This is a dummy variable that indicates whether the rebel group controls territory or not. Fourth, I control for the rebel group?s source of funding. Rebel groups that rely on material endowment to attract recruits are more likely to attract opportunistic individuals who expect fast payoffs for their fighting (Weinstein, 2007). Among such groups, loyalty is, to a large degree, contingent on the material rewards leaders can deliver their followers (Weinstein, 2007). When rebel leaders are unable to offer pecuniary rewards to their combatants, the likelihood of fragmentation increases (Lidow, 2016). I also suggest that groups that extort natural resources will be less likely to invest in socializing their recruits. The availability of lootable resources allows rebel groups to replace their combatants at a relatively low cost. This often reduces the incentive to invest time and resources in combatant training (Weinstein, 2007, p. 137; Haer et al., 2011). From the combatants? side of the equation, opportunistic recruits who join the rebel group for material gain will be less interested in highly demanding 103 activities such as MT. In addition, commanders may use revenues extorted from natural resources to launch a new group (Nagel & Doctor, 2020). I use Nagel and Doctor (2020)?s data to code for rebel groups that subsist on the extortion of national resources. The authors? indicator is based on Walsh et al. (2018)?s Rebel Contraband Dataset (RCD). This is a binary variable which indicates whether a rebel group extorted any of the thirty-one natural resources recorded in the RCD in a given year. I include a number of relevant controls at the conflict level. First, as conflict duration increases, leaders consolidate power and are better positioned to thwart internal opposition. Simultaneously, however, this may induce alienated members to split (Joo & Mukherjee, 2021; Stern, 2022a). Moreover, the probability of fragmentation is relatively high when a rebel group initially emerges, but diminishes after a group survives for several years (Carter, 2012). From the perspective of MT, as conflict duration increases, so are the opportunities for rebel groups to design and implement MT programs. To account for those possibilities, I use Nagel and Doctor (2020)?s data to control for conflict duration. The authors? continuous indicator is based on the conflict start date specified in the UCDP Armed Conflict Dataset (Pettersson & Eck, 2018). Second, I control for conflict-years in which multiple rebel groups are battling the government. All else equal, as the number of rebel groups that occupy the conflict space increases, so does the perceived cost of splitting and, in turn, the likelihood of fragmentation decreases (Fjelde & Nilsson, 2018). Moreover, fighting against the government and rival rebel groups reduces the opportunities, space, time, and resources available for groups to give their combatants MT. To account for the effect of conflicts 104 with multiple rebel groups on fragmentation, I follow Nagel and Doctor (2020)?s coding for multi-dyadic conflicts. This is a binary variable that indicates whether or not the conflict included more than a single rebel group for every conflict-year observed. Third, existing studies on rebel fragmentation find that increased levels of conflict intensity affect a group?s probability of splitting (Kriesberg, 1973, p. 249; Staniland, 2014; Woldemariam, 2016). For example, rebel groups descend into internecine conflict when they incur significant military losses (Christia, 2012; Staniland, 2012). Lapses in conflict intensity, however, can contribute to group cohesion (Stein, 1976). For instance, when rebels are relatively satisfied with their group?s prewar institutional arrangements, an increase in conflict intensity will serve as a stimulus to unify the group (McLauchlin & Pearlman, 2012). From the standpoint of MT, increased levels of conflict intensity not only limit opportunities for MT, but are also likely to shift the priorities of rebels from training to fighting or hiding. Information for the level of conflict intensity is taken from the UCDP Dyadic Dataset (Pettersson & ?berg, 2020). This is a binary variable, which takes the value of 1 if the intensity level is minor?that is between 25 and 999 battle-related deaths in a given year?and the value 2 if the intensity level is qualified as war?that is at least 1,000 battle-related deaths in a given year. Fourth, studies have found that whether a rebel group receives external support shapes the propensity of groups to fragment (D. Cunningham, 2010; Seymour et al., 2016; K. Cunningham, 2017; Tamm, 2016, 2019; Ives, 2019). For example, rebel groups which receive support from foreign sponsors are less likely to fragment because they are better able to compensate their members with selective benefits, which 105 alleviates collective action problems and aids with group cohesion (Carter, 2012). The provision of weapons, camps, military advisers, and safe havens by external supporters (Bapat, 2007; Salehyan, 2007, 2008) enable groups to conduct MT more easily and securely. I follow Nagel and Doctor (2020)?s coding for external support. This is a binary variable that indicates whether a rebel group received external support from other government or not. Finally, fragmentation often occurs when extremist and moderate factions reach an internal impasse about whether to continue fighting or to negotiate a peace agreement with the government (Stedman, 1997; Kydd & Walter, 2002; Nilsson & S?derberg Kovacs, 2011; Pearlman, 2011; Findley & Rudloff, 2012). Specifically, peace negotiations can exacerbate divisions within the rebel group, thus increasing the likelihood of breakaway factions during the peace process (Duursma & Fliervoet, 2021). Theoretically, the occurrence of peace negotiations also reduces the incentive for MT, all else equal. By halting MT, rebels signal their commitment to the negotiation process to the government. This affects government perceptions of the viability of a rebel group as a bargaining partner in peace negotiations. Accordingly, I control for the occurrence of negotiations between the government and the rebel group. Information for the occurrence of peace negotiations is taken from Duursma & Fliervoet (2021). This is a binary variable coded as 1 in the presence of peace negotiations at the rebel-year level and 0 in the absence of such negotiations. 2.6 Results Table 2.2 reports the regression coefficients and their standard errors from the logistic regression models. The coefficients exhibit the expected change in the log-odds of 106 fragmentation in a rebel-year for a unit change in MT, all else equal. To account for possible sources of within-unit heterogeneity, the coefficient standard errors are clustered by rebel group. I find robust evidence that rebel groups that have conducted MT are substantially less likely to fragment compared to those that have not. Table 2.2: Logit regression of MT on rebel fragmentation, 1989 to 2010. Rebel Fragmentation (1) (2) (3) (4) Military training (lagged) -1.349* -1.321* -1.592** -1.417* (0.741) (0.758) (0.780) (0.830) Leftist rebels -0.611 -0.352 0.006 (0.774) (0.795) (0.912) Forced recruitment 0.362 -0.296 -0.325 (0.456) (0.545) (0.687) Rebel sexual violence 1.649*** 0.979 (0.534) (0.686) Rebel group size (logged) -0.104 (0.234) Territorial control 0.960 (0.754) Resource extortion -0.491 (0.659) Conflict duration -0.087** (0.041) Multiparty conflict 1.795** (0.705) External support -0.385 (0.819) Conflict intensity 0.276 (0.698) Central command -1.403 (0.971) Peace negotiations 0.866 (0.730) Constant -2.840*** -2.766*** -2.845*** -1.627 (0.206) (0.351) (0.398) (1.734) Observations 587 410 334 309 Log Likelihood -107.117 -83.126 -68.776 -56.683 Akaike Inf. Crit. 218.234 174.252 147.552 141.367 Note: *p<0.1; **p<0.05; ***p<0.01 107 Model 1 displays the bivariate relationship between my measure of MT and the likelihood of fragmentation in the following year without any controls. In support of my hypothesis, this basic regression indicates that these variables correlate with one another. Other reported regression models further support this correlation. Model 2 includes controls for the main alternative methods of formal rebel socialization: selective recruitment and PE. Accounting for these variables, I find support for my hypothesis: rebel groups that socialize their recruits through MT are less likely to fragment than rebel groups that either abstain from MT or rely on other formal methods of socialization. Model 3 adds to model 2 the informal method of rebel socialization: sexual violence. I find that rebel groups that conduct MT are less likely to fragment compared to groups that engage in other, both formal and informal, methods of rebel socialization, all else equal. Finally, model 4 combines models 1 through 3 to control for potential confounding variables with regard to group-level factors and conflict-level factors. On average, MT decreases the probability of rebel fragmentation by 7 percentage points, all else equal. Percentage-wise, MT decreases the probability of fragmentation by 75 percent. Some of the controls produce interesting results as well. According to model 3, rebel groups that do not screen for prospective recruits or those that instruct PE programs are less likely to fragment, albeit insignificantly. However, groups that perpetrate sexual violence are significantly more likely to experience splits. Indeed, a recent study has found that rebel groups that perpetrate acts of sexual violence are more likely to fragment than rebel groups that abstain from sexual violence (Nagel & Doctor, 2020). 108 2.7 Predicted Probabilities Figure 2.3 displays the predicted probabilities of fragmentation associated with variation in MT. To calculate these values, I use the coefficients and standard errors from model 4 to predict changes in the probability of fragmentation based on changes in the main explanatory variable while holding all control variables at their mean value. Figure 2.3 shows the separate predicted values with 0.95 confidence intervals. Figure 2.3: Predicted probabilities of rebel fragmentation, 1989 to 2010. I find that the predicted probability of fragmentation is 0.038, suggesting that nearly 4 percent of rebel groups without MT are predicted to fragment, holding all other variables at their sample means. For rebel groups that conduct MT, the predicted probability of fragmentation drops to just under 0.01, suggesting that about 1 percent of 109 rebel groups with MT are predicted to fragment. In sum, MT is associated with a 3.0 percentage point decrease in fragmentation. In percentage change terms, MT reduces the probability of fragmentation by about 75 percent. This predicted change in the displayed values is statistically significant at the 0.95 confidence level. Following a year in which a rebel group conducts MT, the probability of fragmentation in that group shifts from 0.04 to 0.01, all else equal. Given the unit of analysis and the relative rarity of this event?fragmentation only occurs in 4.5 percent of my observations? this is a meaningful effect. In sum, my results indicate that MT is highly correlated with internal cohesion. Rebel groups that conduct MT are significantly and substantively less likely to experience fragmentation, all else equal. The repercussion of abstaining from MT is an increased risk of fragmentation, all else equal. 2.8 Robustness Checks and Further Analysis As a robustness check, I replicate my analysis using annual fixed effects model (shown in Appendix Table 2.5). My decision to run an annual fixed effects model is because fragmentation does not occur in some years in my sample. Annual fixed effects analysis (i.e., year as a dummy variable) controls for factors that change each year and are likely to be common to all rebel groups for a given year. Results remain robust to this test. In addition, I perform robustness checks using different lag structures for my independent variable, military training, lagged by one year. I begin by capturing instances in which a rebel group is recorded as conducting MT in either of the first or second lagged years (i.e., either at time t-1 or t-2) (shown in Appendix Table 2.6). As expected, the coefficient?s sign indicates that MT is negatively associated with fragmentation. However, the statistically insignificant results across the four models 110 suggest that the cohesion-building effects of MT may deteriorate after two years. In order to check whether the effect of MT on fragmentation diminishes over time or whether it is associated with a particular lagged year, I captured instances in which a rebel group is recorded as conducting MT in either of the first, second or third lagged years (either at time t-1 or t-2 or t-3) (shown in Appendix Table 2.7). Again, the coefficient?s sign is in the expected direction, however results remain statistically insignificant across the four models. These models? results suggest that the effect of MT on fragmentation are strongest in the short-term but this effect does not last. Indeed, as we look further in time, more groups with MT fragment. For instance, with a one-year lag, two rebel groups that conducted MT fragmented. With a three-year lag, however, eight groups that conducted MT fragmented. The six additional groups that fragmented after having MT in the three-year lag model reduce the coefficient size considerably. Thus, even though the size of the standard error decreases across the different lag structures models, the estimated coefficient increases, hence the loss of significance. Because the conditions under which MT matter are in the short-term, it is reasonable to posit that as a method of rebel socialization MT alone might be insufficient to inhibit fragmentation. Future research may consider addressing this point. Finally, I replicate my models, replacing my original indicators of selective recruitment and PE. In case abduction should be too narrow, I perform robustness checks with a control that indicates a broader conceptualization of forced recruitment. Specifically, I replace the indicator that accounts for rebel groups that abduct fighters to fill their ranks with an indicator that measures whether a rebel group has ever forced recruitment. Results are robust to this test (see Appendix Table 2.8). In addition to leftist 111 rebel groups, other ideologically-motivated rebel groups are also likely to carry out PE. To address this possibility, I run robustness checks with a control that indicates whether the rebel group was founded with a specific political ideology. I use the Foundations of Rebel Group Emergence (FORGE)?s (J. Braithwaite & Cunningham, 2020) dichotomous variable ideology to account for whether a rebel group was founded with a specific political ideology or not. Results remain robust to this test too (see Appendix Table 2.9). 2.9 Conclusion This article argues that MT hinders rebel fragmentation and increases cohesion. As a form of rebel socialization MT is uniquely positioned to foster interpersonal bonds among combatants, between combatants and commanders at the combat unit level, and between the rebel group members? and the overall organization. Combined, these bonds make splintering more costly. When rebels are socially and professionally tied to one another, the costs of exiting the group increase, and fragmentation is less likely. Testing my argument on a global sample of 83 rebel organizations active between 1989 and 2010, I find that rebel groups that conduct MT are less prone to fragmentation, all else equal. This study contributes to research on rebel fragmentation in a number of ways. First, it brings forth cohesion - as opposed to fragmentation - as the puzzle in need of explanation. Second, to address the puzzle of cohesion, the article emphasizes the role of rebel socialization in fostering group cohesion. Third, and more specifically, this study connects group structure with MT, a method of rebel socialization, in order to convey a more nuanced explanation of the processes by which rebel groups inhibit 112 fragmentation and build cohesion. It highlights the particular attributes of MT for the promotion of group cohesion and illustrates how MT influences the organizational stability of an insurgency. This study demonstrates the importance of intragroup factors and rebel socialization in influencing a rebel group?s disposition toward fragmentation. I find that rebel groups that have recently conducted MT are less likely to fragment by about 75 percent. This implies that rebel leaders intend on building group cohesion should provide their subordinates with MT. From a policy perspective, fragmentation results in more prolonged and more destructive conflicts. Fragmented groups are less likely to achieve their goals, whereas cohesive groups tend to be more disciplined and are better able to credibly commit to their program of action. A government that seeks to support a rebel group may therefore want to consider assistance in the form of MT. Conversely, a government that seeks to weaken a rebel group should target the group?s MT, if such training indeed takes place. But research remains incomplete. First, while I find support for my argument, suggesting that the MT socialization argument has broad applicability, the data were not collected for this particular analysis and as such do not provide a decisive test. Future data collection efforts should aim at collecting more data on MT at the group level. Closer attention should be given to the special characteristics of MT and their respective impacts for group cohesion. Second, because the conditions under which MT matter are in the short-term, it is reasonable to posit that MT alone may be insufficient to inhibit fragmentation. Future research may want to investigate how rebel groups supplement MT with other forms of socialization in order to build greater internal cohesion. 113 2.10 Appendix Appendix Figure 2.4: Alternative presentation of rebel fragmentation over time. Appendix Figure 2.4 presents the distribution of the dependent variable, rebel fragmentation over time. In contrast to Figure 2.3 from the main manuscript, this figure illustrates the proportion of active rebel groups which are observed to fragment in a given year. This figure also indicates that fragmentation varies not just within groups but across them as well. 114 Appendix Table 2.3: Cases of rebel fragmentation in the data sample. Group Year Military training Rebel (lagged) Fragmentation AFRC 1998 0 1 ALiR 2000 0 1 Ansar al-Islam 2007 0 1 ASG 1998 0 1 ASG 2007 0 1 Chechen Republic of 2007 0 1 Ichkeria CPP 1992 0 1 ELN 1989 0 1 FDLR 2010 0 1 Forces of the Caucasus 2010 0 1 Emirate GAM 1999 0 1 GIA 1998 1 1 Hizb-i Islami-yi 1990 0 1 Afghanistan - Khalis faction LRA 2000 1 1 LURD 2003 0 1 MFDC 1992 0 1 MFDC 2000 0 1 NLFT 2000 0 1 NPFL 1990 0 1 RCD 1999 0 1 RCD 2000 0 1 Sendero Luminoso 1992 0 1 SLM/A 2005 0 1 SPLM/A 1991 0 1 SPLM/A 1992 0 1 SPLM/A 1998 0 1 UIFSA 1997 0 1 115 Appendix Table 2.4: Cross tabulations of the main variables. Military training (lagged) Military training (lagged) = 0 = 1 Rebel Fragmentation = 0 428 132 Rebel Fragmentation = 1 25 2 Appendix Table 2.5: Annual fixed effects, 1989 to 2010. Rebel Fragmentation (1) (2) (3) (4) Military training (lagged) -1.371* -1.411* -1.671* -1.820* (0.754) (0.782) (0.871) (1.001) Leftist rebels -0.792 -0.572 -0.280 (0.803) (0.880) (1.105) Abducted recruits 0.323 -0.420 0.066 (0.492) (0.639) (0.875) Rebel sexual violence 2.176*** 1.580* (0.666) (0.860) Rebel group size (logged) -0.187 (0.312) Territorial control 1.322 (0.908) Resource extortion -0.672 (0.833) Conflict duration -0.115** (0.051) Multiparty conflict 2.149** (0.896) External support -0.809 (1.054) Conflict intensity -0.291 (0.930) Central command -1.071 (1.292) Peace negotiations 0.754 (1.005) Constant -2.676*** -2.102*** -1.757* 0.251 (1.033) (0.815) (0.933) (2.613) Observations 587 410 334 309 Log Likelihood -85.370 -65.328 -49.807 -39.580 Akaike Inf. Crit. 216.740 178.657 149.614 147.160 Note: *p<0.1; **p<0.05; ***p<0.01 116 Appendix Table 2.6: Alternative MT lag structures (t-1| t-2), 1989 to 2010. Rebel Fragmentation (1) (2) (3) (4) Military training (t-1|t-2) -0.259 -0.403 -0.691 -0.347 (0.473) (0.528) (0.563) (0.630) Leftist rebels -0.578 -0.352 -0.043 (0.775) (0.794) (0.901) Abducted recruits 0.269 -0.423 -0.539 (0.457) (0.553) (0.690) Rebel sexual violence 1.631*** 0.944 (0.548) (0.675) Rebel group size (logged) -0.087 (0.249) Territorial control 1.075 (0.780) Resource extortion -0.568 (0.676) Conflict duration -0.089** (0.040) Multiparty conflict 1.750** (0.709) External support -0.546 (0.825) Conflict intensity 0.441 (0.677) Central command -1.551 (0.966) Peace negotiations 0.867 (0.722) Constant -2.967*** -2.832*** -2.867*** -1.888 (0.224) (0.359) (0.400) (1.859) Observations 586 410 334 309 Log Likelihood -109.304 -84.904 -70.863 -58.342 Akaike Inf. Crit. 222.608 177.807 151.727 144.683 Note: *p<0.1; **p<0.05; ***p<0.01 117 Appendix Table 2.7: Alternative MT lag structures (t-1| t-2| t-3), 1989 to 2010. Rebel Fragmentation (1) (2) (3) (4) Military training (t-1|t-2|t-3) -0.048 -0.081 -0.402 -0.223 (0.431) (0.477) (0.519) (0.590) Leftist rebels -0.563 -0.294 -0.008 (0.775) (0.792) (0.902) Abducted recruits 0.226 -0.469 -0.583 (0.458) (0.555) (0.684) Rebel sexual violence 1.608*** 0.952 (0.560) (0.684) Rebel group size (logged) -0.086 (0.251) Territorial control 1.087 (0.787) Resource extortion -0.576 (0.687) Conflict duration -0.090** (0.040) Multiparty conflict 1.746** (0.711) External support -0.575 (0.827) Conflict intensity 0.495 (0.664) Central command -1.579 (0.965) Peace negotiations 0.887 (0.720) Constant -3.014*** -2.889*** -2.895*** -1.951 (0.235) (0.367) (0.405) (1.877) Observations 585 410 334 309 Log Likelihood -109.407 -85.198 -71.366 -58.424 Akaike Inf. Crit. 222.813 178.395 152.733 144.849 Note: *p<0.1; **p<0.05; ***p<0.01 118 Appendix Table 2.8: Forced recruitment, 1989 to 2010. Rebel Fragmentation (1) (2) (3) (4) Military training (lagged) -1.349* -1.238* -1.631** -1.504* (0.741) (0.752) (0.776) (0.823) Leftist rebels -0.800 -0.239 0.033 (0.765) (0.798) (0.912) Forced recruitment 0.298 -0.119 0.204 (0.575) (0.614) (0.739) Rebel sexual violence 1.551*** 0.903 (0.495) (0.667) Rebel group size (logged) -0.102 (0.227) Territorial control 0.998 (0.739) Resource extortion -0.531 (0.651) Conflict duration -0.090** (0.041) Multiparty conflict 1.752** (0.699) External support -0.317 (0.833) Conflict intensity 0.127 (0.682) Central command -1.541 (0.973) Peace negotiations 0.761 (0.713) Constant -2.840*** -2.812*** -2.878*** -1.501 (0.206) (0.520) (0.545) (1.638) Observations 587 410 334 309 Log Likelihood -107.117 -83.302 -68.905 -56.757 Akaike Inf. Crit. 218.234 174.605 147.810 141.514 Note: *p<0.1; **p<0.05; ***p<0.01 119 Appendix Table 2.9: Ideologically-motivated rebel groups, 1989 to 2010. Rebel Fragmentation (1) (2) (3) (4) Military training (lagged) -1.349* -1.406* -1.703** -1.489* (0.741) (0.771) (0.791) (0.844) Leftist rebels -1.125** -0.875 -0.943 (0.556) (0.595) (1.019) Abducted recruits 0.311 -0.292 -0.294 (0.461) (0.547) (0.683) Rebel sexual violence 1.529*** 0.945 (0.550) (0.690) Rebel group size (logged) -0.125 (0.237) Territorial control 1.126 (0.776) Resource extortion -0.665 (0.679) Conflict duration -0.073* (0.043) Multiparty conflict 1.683** (0.715) External support -0.787 (0.932) Conflict intensity 0.313 (0.695) Central command -1.589* (0.960) Peace negotiations 0.611 (0.762) Constant -2.840*** -1.888*** -2.098*** -0.450 (0.206) (0.575) (0.651) (2.155) Observations 587 404 330 309 Log Likelihood -107.117 -78.841 -67.762 -56.249 Akaike Inf. Crit. 218.234 165.682 145.525 140.498 Note: *p<0.1; **p<0.05; ***p<0.01 120 3 Foreign Fighters and Rebel Fragmentation Abstract This article investigates the effect of foreign fighters (FFs) on rebel fragmentation. First, I explore how reduced group dependency on local fighters, preference divergence, strategic disagreements, and member segregation increase the likelihood of fragmentation for rebel groups that recruit FFs. Second, I posit that rebel groups that recruit coethnic FFs will be less likely to fragment. I test these arguments on a sample of 227 rebel groups active between 1989 and 2011. I find that rebel groups that recruit FFs are significantly more likely to split. However, the recruitment of coethnic FFs does not diminish the probability of fragmentation. 121 3.1 Introduction In what way do foreign fighters (FFs) shape the trajectories of the rebel groups they join? Though the academic and policy fields have become increasingly apprehensive concerning the danger of FFs enlisting in civil wars around the world, there is still a dearth of research pertaining to how FFs influence group dynamics (Bakke, 2014; Chu & Braithwaite, 2017; A. Braithwaite & Chu, 2018). For the most part, academic research and policy have centered on the high- lethality attacks associated with FFs (Cordesman, 2005; Rich & Conduit, 2015) or the danger FFs pose when they return to their home countries, especially in the West (Bergen & Reynolds, 2005; Noonan & Khalil, 2014; Lindekilde et al., 2016; Barrett, 2017; A. Braithwaite & Chu, 2018; Malet & Hayes, 2020). Far less attention, however, has been given to questions about FFs? influence within the insurgency. Given that insurgent recruitment of FFs has increased over recent decades, both in absolute measures and in relation to the entire fighting forces, it is fitting to examine the impact of FFs on insurgencies (Malet, 2013). Existing work presupposes that FFs present a tradeoff for the rebel group. On the one hand, FFs boost the capabilities of domestic insurgents through their additional manpower, resources, and know-how (Bakke, 2014; Donnelly et al., 2017; Doctor, 2021). On the other hand, the assets that FFs bring with them come at a high cost. Specifically, FFs disrupt the internal dynamic, precipitate deep divisions, and pose a threat to the group?s internal cohesion (Bakke, 2014; Rich & Conduit, 2015; Doctor, 2021). The assertion that the presence of FFs increases the probability of group 122 fragmentation has not been tested directly. As of date, no cross-case study has attempted to determine whether FFs bear a unique effect on fragmentation. The goal of this article is to assess the effect of FFs on rebel group fragmentation, ?an event in which a segment of a rebel organization formally and collectively exits the existing rebel organization and establishes a new, independent rebel organization? (Doctor, 2020, p. 599). Because rebel fragmentation increases the likelihood of civil war incidence (K. Cunningham, 2013), shapes the intensity and duration of fighting (D. Cunningham, 2006; K. Cunningham et al., 2012; Findley & Rudloff, 2012), and the likelihood and durability of peace (Nilsson, 2010; Driscoll, 2012; Rudloff & Findley, 2016; K. Cunningham, 2017), the implications of fragmentation are highly relevant for governments fighting rebels domestically and for international actors interested in conflict resolution. By explicating the theoretical and empirical connection between FFs and the probability of rebel fragmentation, this article contributes to a growing body of literature on the transnational dimensions of violent conflicts (cf. Salehyan, 2009, 2010; Checkel, 2013; Salehyan et al., 2014). I investigate the effect of FFs on group cohesion, examining a number of mechanisms derived from previous research. I point to reduced group dependency on local fighters, preference divergence, strategic disagreements, and member segregation as mechanisms that increase the likelihood of splits amongst rebel groups that recruit FFs. Next, I posit that if the foreignness of FFs to local group members makes fragmentation more likely, then rebel groups that recruit coethnic FFs will be less likely to undergo fragmentation. 123 To test my arguments, I estimate a series of binary logistic regression models with a sample of 227 rebel groups active in 111 armed intrastate conflicts between 1989 and 2011. My main finding supports my argument that rebel groups that recruit FFs are more likely to split into multiple groups. However, against my expectations, I find that rebel groups that recruit coethnic FFs remain susceptible to fragmentation, suggesting that the gravity of coethnicity in cultivating group cohesion might be overestimated. In sum, the article?s findings help to make the link between FFs and rebel group fragmentation more explicit. It highlights the advantage of disaggregating FFs while pointing to the possible limitations of shared traits on hindering rebel fragmentation. This article is structured as follows. First, I define what FFs are and, importantly, how they are different from other foreigners who participate in an intrastate conflict. Second, I review existing studies on FFs in civil wars and explain the tradeoff associated with their enlistment. Third, I introduce an argument connecting FFs and fragmentation and highlight mechanisms that link FFs and rebel fragmentation. Fourth, I explain how rebel groups that recruit coethnic FFs are more likely to mitigate the adverse effects of FFs on group cohesion. Fifth, I present my research design, the results of my analysis, and address possible selection problems associated with FFs and fragmentation. Finally, I conclude with a discussion of the results and their implications for future research and policy. 3.2 Defining Foreign Fighters (FFs) FFs are noncitizens of conflict states who, motivated by either ideological or economic incentives, join local rebels during civil conflicts (Malet, 2013, p. 9; Hegghammer, 2010). 124 A key feature that defines FFs is that their decision to join domestic insurgents is inherently voluntary; FFs travel without direct state support to fight in a local conflict. They are not part of any government force, nor any military or paramilitary contingency, and are often unpaid (Hegghammer, 2010, pp. 58-59).26 In fact, becoming a FF often requires individuals to violate their own state?s laws; for example, by traveling to a foreign conflict zone or by joining a foreign armed force (Malet, 2013, p. 9). FFs are distinguished from salaried soldiers or mercenaries. Soldiers who enter a conflict as part of a foreign military intervention ? such as Cuban military forces in Angola during the 1970s or Russian and Iranian forces in the Syrian Civil War ? or mercenaries and other professional soldiers hired to serve in a foreign army ? such as French Foreign Legionnaires or Russian-backed Chechen combatants in Ukraine ? expect that the government that has deployed them or hired their services will provide them with legal protection, pecuniary compensation, and the necessary military gear to win the conflict. As such, they might well expect to survive the war to enjoy their compensation. Because FFs are neither deployed nor hired by their home government, they cannot expect such protections and benefits, lowering the likelihood of their surviving the conflict than conventional soldiers from the very beginning (Malet, 2013, p. 9). 26 For example, Malet (2013) argues that economic opportunities played a marginal role in the decision of foreigners to join local insurgencies. He writes that evidence indicates that FFs in multiple civil conflicts (e.g., the Israeli War of Independence and the Iraq War) were told explicitly during the recruitment process that they could expect only meager payments (p. 9). More recent evidence suggests otherwise. According to Mironova (2019, p. 148), numerous FFs in Iraq and Syria offered their services to the highest bidder. 125 Finally, FFs join and operate within the confines of the insurgency (Hegghammer, 2010, pp. 57-58) and should not be automatically confounded with international insurgents. In other words, whereas all FFs are insurgents, not all are international insurgents. Conceptually, this point is particularly important. Because organizations that have drawn heavily on FFs have engaged in international terrorism, most notably al-Qaeda and ISIS, terms such as ?Jihadists? or ?Salafi Jihadists? have become synonymous with any transnational violent Islamist, regardless of whether the perpetrator attacked a Western capital or operated inside the war zone. In reality, however, most FFs stayed in one conflict zone at a time (Hegghammer, 2010, p. 58), making the distinction between FFs and international terrorists important. 3.3 FFs in Civil Wars The phenomenon of FFs is hardly new, but it continues to shape civil conflicts in contemporary history (Mendelsohn, 2011; Malet, 2015; Donnelly et al., 2017). From the nineteenth century Romantic English poet, Lord George Byron, who fought and died alongside Greek rebel forces in their struggle for independence from the Ottoman Empire (Thomson, 1990, pp. 142, 158-59) to the Saudi national Osama Bin Laden, who joined local Afghans during the 1980s in their fight to expel the Soviet invaders, to ?the American Taliban? John Walker Lindh, a Muslim convert who fought for the Taliban in Afghanistan in 2002 until his capture by U.S. forces, the number of FFs has grown overtime, both in absolute numbers and as a proportion of total insurgencies (Malet, 2013). FFs are often drawn to conflicts that begin with humanitarian crises or civil wars that then result in atrocities perpetrated against local populations. These volunteer fighters join the conflict because they feel obligated to fight for the victimized 126 population (Donnelly et al., 2017, p. 30). FFs have participated in civil wars across all regions (Figure 3.1), including in 40 civil wars, or 36 percent of the total civil conflicts from 1989 to 2011 (Figure 3.2). Figure 3.1: Civil wars with foreign fighters by region, 1989 to 2011. 127 Figure 3.2: Civil wars with foreign fighters vs. total civil wars, 1989 to 2011. Hence, a more thorough analysis of the trajectory of FFs is a useful way to better understand contemporary political violence and insurgency, as demonstrated most recently in Iraq and Syria, where Jihadi FFs, though a relatively small percentage27 of the entire insurgency, have utilized highly effective, internet-centered information operations to disproportionately influence the battle space. These operations promote a specific narrative to the group?s audience, particularly potential recruits, of the resistance to occupation, and cause considerable problems for those who counter the insurgency. 27 Over the course of the Iraq War, as many as 4,000-5,000 FFs joined up, forming as much as 5 percent of the total insurgent forces (Hegghammer, 2010, p. 60). 128 In terms of recruitment, most of the major shifts have occurred in the latter half of the twentieth century, in tandem with major historical changes ? first, the rise of national liberation movements in the wake of the collapse of colonialism and Communism, then the attempt by international Islamic fundamentalist groups to export their struggle in 1980s Afghanistan to other areas of conflict, such as Eurasia, the Balkans, and the Middle East. Almost all fighters in 1980s Afghanistan came from the Gulf states, North and East Africa, and Central Asia, whereas ISIS drew many of its fighters from Europe and North America, in addition to recruits from the Arab world and Central Asia. Globally, ISIS has mobilized tens of thousands of individuals from more than 120 countries (Donnelly et al., 2017). Globalization, particularly technological developments in communications and transportation, have enabled a growing number of recruiters and foreign recruitees to connect and join forces on battlefields (Malet, 2013, p. 54) stretching from the Horn of Africa to Ukraine. Existing work presupposes that FFs present a tradeoff for the rebel group. On the one hand, FFs boost the capabilities and resources of domestic rebel groups (Bakke, 2014; Doctor, 2021). FFs are associated with a decreased likelihood of government victory and increased probability of reaching a negotiated settlement (Chu & Braithwaite, 2017). On the other hand, the assets that FFs bring with them come at a high cost. FFs disrupt a group?s internal dynamics, precipitate divisions, and pose a threat to the group?s internal cohesion (Bakke, 2014; Rich & Conduit, 2015; Doctor, 2021). What are the advantages and disadvantages of FFs? How do FFs enhance or hinder a group?s dynamics and capacity? 129 3.3.1 The Advantages of FFs FFs offer several major benefits for the rebel group. First, because rebels begin as the weaker party in civil wars, they try to bolster their forces by acquiring external support, including fighters. Without recruiting a sufficient number of individuals, a rebel group cannot overcome the collective action problem, survive, evolve, defeat, or extract concessions from the government (Weinstein, 2007; Mosinger, 2019; Lewis, 2020). FFs help a rebel group inflate its ranks rapidly and, thus, its capacity. Second, FFs often bring with them knowledge, experience, and skills which help insurgents improve their operational range (cf. Bacon & Muibu, 2019) and compensate for the material asymmetries they usually face in civil wars (Chu & Braithwaite, 2017). Moreover, some FFs bring with them military experience from either conventional state militaries or as veterans of previous civil wars. Arab veterans of the Afghan war against the Soviets, commonly known as the Afghan-Arabs, 28 joined forces with local combatants in Bosnia (Malet, 2013, p. 159), Tajikistan, and Chechnya (C. Moore & Tumelty, 2008), and later joined or established an array of other Jihadist groups, such as Al-Qaeda or the Algerian GIA (Bergen & Reynolds, 2005). More than half of the Arab FFs who fought in Chechnya were veterans from Afghanistan, Tajikistan, or Bosnia (Donnelly et al., 2017, p. 11). During the First Chechen War (1994-1996), Emir Khattab, a Saudi national and a prominent veteran of the Afghan war, led a contingent of Arabs from the Middle East to help create a formal military structure for the Chechen rebels and established training 28 The term ?Afghan Arabs? refers to those primarily Arab mujahedeen who came to Afghanistan to assist their fellow Muslims against the Soviets. 130 camps in mountainous parts of Chechnya, where hundreds of young men passed through to become fighters (C. Moore & Tumelty, 2008, p. 416; Bakke, 2014, p. 167). By 1999, the use of suicide bombings, a tactic foreign to Chechen tradition, has gradually become more common (Nivat, 2005; Speckhard & Ahkmedova, 2006, as cited in Bakke, 2014, p. 179). Some FF veterans of Chechnya have gone on to carry out other terrorist attacks or become members of other terrorist groups, including one leader of AQAP, two of the 9/11 hijackers, and those involved in the string of Istanbul bombings in 2003 (C. Moore & Tumelty, 2008, p. 423). FFs who joined the Croat side during the Yugoslav War (1991-1995) carried out most of the special operations (Arielli, 2012, pp. 9, 13). In the Philippines, Indonesians who fought in Afghanistan boosted the ranks of the Abu Sayyaf group. In late 1997, mujahedeen from Egypt and Saudi Arabia led suicide attacks against the Filipino army (Malet, 2013, p. 189). In Afghanistan, FFs provided local fighters with military training. In Pakistan?s tribal regions, Taliban FFs set up suicide bombing training camps and established roadside-bomb factories (Rohde, 2007). In Somalia, FFs from the Gulf served as trainers and advisors, and local insurgents eagerly adopted and adapted many of the technological skills the FFs brought with them. Between 2007 and 2008, FFs brought specialized and tactical skills to al-Shabaab, such as improvised explosive devices, suicide bombings, sniping, and anti-tank techniques, and these advancements enhanced the recruits? basic skillset (Bacon & Muibu, 2019, pp. 293-294, 298). Numerous members of ISIS were veterans of both Chechen wars against Russia (Mironova, 2019, p. 136). They commanded insurgent forces, instructed other 131 combatants in shooting, driving military vehicles, and navigation, and illustrated a level of ability not seen in that battle zone before (Mironova, 2019; Sly, 2013). Third, FFs tend to be more devout than their local peers. To begin, in order to join the insurgency, FFs must overcome a greater number of hurdles compared to local combatants who are already present in the battlefield (Mironova, 2019, p. 37). In Chechnya, for instance, the number of FFs was regulated partly by the difficulties of traversing the region?s harsh terrain (C. Moore & Tumelty, 2008, p. 422). Moreover, FFs are often considered more extreme and violent than local forces. Their dedication to their goals drives FFs to participate in more dangerous operations, particularly suicide missions (Cordesman, 2005). The new tactics that FFs in Afghanistan have introduced have increased both Afghan and Western casualty rates (Rohde, 2007). In Iraq and Syria, the number of FFs who joined different Jihadi groups often stood in a stark contrast to their lethality. During the early years of the war in Iraq, FFs comprised less than 10 percent of the Iraqi insurgency. However, they accounted for more than 90 percent of suicide bombings and high lethality attacks (Quinn & Shrader, 2005, as cited in Malet, 2013, p. 6; Cordesman, 2005). According to a local member of Jabhat al- Nusra, a Jihadist rebel group fighting against Syrian government forces, FFs arrived to Syria ??to die, so having them is actually good because they make the group stronger on the frontline? (Mironova, 2019, p. 139). Fourth, FFs provide rebel groups with an additional source of income. FFs tend to be wealthier than local fighters and often bring their own money and even small arms (Malet, 2013, p. 3; Bakke, 2014; Mironova, 2019, pp. 37, 137-138). This reduces FFs? dependency on both the group and local communities? resources for operational support 132 (Doctor, 2021). In addition, having contacts and networks in their home communities make FFs better fundraisers for the group (Mironova, 2019, pp. 37, 137-138). Al- Shabaab actively recruited FFs and used their participation as a way to muster external resources (Bacon & Muibu, 2019, p. 281). Relatedly, some foreigners possess crucial contacts within the international war industry ? to acquire weapons, for example ? that locals either completely lack or need time to cultivate (Mironova, 2019, p. 37). Finally, FFs can serve as effective channels for propaganda. By leaving their lives behind and joining local rebels abroad, FFs vouch for and legitimize the cause of local fighters while proving their own dedication to the enemy (Mironova, 2019, p. 37). FFs who joined the Croat side during the Yugoslav War helped boost morale (Arielli, 2012). A local fighter in the Croatian forces claimed that he felt as if the involvement of FFs meant that some still paid attention to what was going on in Croatia (Arielli, 2012, p. 10). Rebel groups, too, utilize FFs to promote their cause and recruit additional fighters. Different technical skills ? for example those related to language, information technology, and social media ? help disseminate the group?s message and recruit additional FFs. For instance, Taliban FFs in Pakistan enhanced their internet presence, doubling the amount of well-crafted fundraising and recruiting videos online (Rohde, 2007). In Somalia, al-Shabaab?s FFs feature prominently in the group?s recruitment videos, calling on other non-Somalis to join the fight (Bacon & Muibu, 2019, p. 293). 3.3.2 The Disadvantages of FFs Despite the advantages that FFs confer on a rebel group, they can also weaken it and become a considerable burden (Bakke, 2014; Rich & Conduit, 2015; Mironova, 2019). In Chechnya, for example, the arrival of FFs have strengthened the insurgency against 133 Russia in terms of resource mobilization. However, they have also catalyzed the division of the Chechen insurgency into two camps ? one Islamist branch and one nationalist ? in addition to increasing defections to the Russian side under Chechen President Ramzan Kadyrov (Bakke, 2014). There are several ways by which FFs disrupt organizational dynamics. First, FFs are often ignorant of the local context, language, customs, and terrain (Mironova, 2019, p. 38; P. Moore, 2019). Dissonance between FFs? ideas regarding the conflict?s objectives and tactics and the local context of the war hinders locals from supporting the rebels (Bakke, 2014) and may prompt resistance from local rebels and the general population, whose support is central to organizational cohesion (Bakke, 2014; Mosinger, 2018). During the Afghan War in the 1980s, FFs and local Afghans ?disagreed on almost everything, including politics and religious rituals? (Kohlmann, 2004, p. 8, as cited in Malet, 2013, p. 179). FFs considered local Afghans ?dumb? and illiterate in the Quran (Hafez, 2009, p. 78, as cited in Malet, 2013, p. 180). They viewed the Afghans as sacrilegious and as trying to uproot local cultural customs (Malet, 2013, p. 179). Furthermore, many of the FFs came from upper-middle class families. Native Afghan and local rebels were particularly wary of their new Arab comrades, or ?Gucci? soldiers, who were disconnected from the social and religious reality of Afghan life (Kohlmann, 2004, p. 8, as cited in Malet, 2013). In Chechnya, some locals found that the Islamist Arab fighters enforced customs that were unfamiliar, such as forcing women to wear veils or confining them to the home. Local Chechens did not perceive such a form of Islamism to be the ?true? religion (Bakke, 2014, p. 176). 134 In the years that followed the American invasion of Afghanistan in late 2001, Afghan villagers in the southern provinces of Oruzgan, Kandahar, and Helmand described two distinct groups of Taliban fighters. The ?local Taliban? allowed some development projects. But the ?foreign Taliban? threatened to kill anyone who cooperated with the Afghan government, foreign aid groups, or anyone who refused to provide food and support (Rohde, 2007). FFs who joined al-Shabaab have had to navigate numerous cultural obstacles and overcome widespread mistrust from the Somali population (Bacon & Muibu, 2019, p. 295). Because of local skepticism and suspicion of foreigners, FFs were kept apart from the local populace, resulting in heavier costs than benefits for the insurgency (Bacon & Muibu, 2019, p. 296). Moreover, multiple studies find that the presence of FFs is associated with higher levels of civilian victimization (P. Moore, 2019; Doctor, 2021). Because FFs reduce the dependency of the rebel group on the local population, rebel leaders feel less incentive to punish FFs for civilian victimization. Concurrently, leaders need to socialize the new FFs into the group in order to build internal cohesion. To boost cohesion, rebel leaders will often allow or, at the very least, tolerate sexual violence against civilians (Doctor, 2021). In the case of ISIS, a surge in the number of FFs led to more prevalent levels of sexual violence against civilians (Doctor, 2021). Second, FFs can tilt the balance within a rebel group. Often, FFs take the lead in a variety of roles which, before their arrival, were occupied by local rebels. In the case of the Afghan Taliban, FFs have increasingly captured the midlevel command positions and replaced dozens of midlevel and senior local Taliban officials who have been killed by NATO and American forces (Rohde, 2007). In trying to expand their 135 influence within the insurgent organization, FFs may use their financial resources to support more extreme local fighters. In Pakistan, FFs have financed younger Taliban leaders in the country?s tribal areas who have closer ties to Al-Qaeda. Consequently, the influence of older, more traditional Taliban leaders has diminished (Rohde, 2007). In Syria, FFs who joined ISIS were incorporated into all levels of leadership (Mironova, 2019, p. 163), which is historically atypical for FFs (Daymon et al., 2020). Their ubiquitous presence at checkpoints or as battlefield commanders have given FFs an observably formidable stature across rebel territory and earned them de facto rule of towns and cities under rebel control (Sly, 2013). In laying the theoretical groundwork for this article, the following section develops an argument of FFs and fragmentation and highlights mechanisms that connect FFs and rebel fragmentation. I then turn to explain how rebel groups that recruit coethnic FFs are more likely to offset the adverse effects of FFs on group cohesion. 3.4 The Argument: Foreign Fighters and Rebel Fragmentation I argue that FFs upset the bonds between members of an insurgent group as well as the rebels? sense of commitment to the group in general, thereby constituting a potential source of internal divisions. Building on existing studies which emphasize the ramifications of FFs on the internal dynamics of insurgent organizations, I point to four key mechanisms by which FFs exacerbate internal divisions and, ultimately, precipitate fragmentation. 136 3.4.1 Reduced Dependence on Local Fighters The recruitment of FFs reduces the group?s dependence on local fighters for material and operational support (Bakke, 2014; Doctor, 2021), thus shifting the strategic calculus of fragmentation. From the local fighter?s perspective, the recruitment of FFs signals that his relationship with the rebel group is not durable, his labor is expendable, and that the leadership views him as a fungible asset which can be replaced easily. As a result, fighters have weaker incentives to remain in the group, and exiting it is less costly. From the leadership?s perspective, access to FFs ? and the resources they provide ? reduces the incentive of the leadership to resist the split of discontented fighters. 3.4.2 Preference Divergence Existing research emphasizes preference divergence among group members as a threat to cohesion and a major cause of rebel fragmentation (Bakke et al., 2012; Shapiro, 2013; Warren & Troy, 2015; Woldemariam, 2018; Burch & Ochreiter, 2020). As outsiders joining a domestic conflict, FFs are often associated with preference divergence concerning the conflict?s objectives and tactics, which causes divisions with local forces (Bakke, 2014; P. Moore, 2019). FFs may introduce a different rationale for fighting compared to local fighters (Malet, 2013, p. 42). Specifically, FFs introduce new approaches on how to frame the conflict ? that is, what the war is about ? and tactics ? that is, how the war should be fought (Bakke, 2014). If these ideas are unwelcomed by local fighters, the result could threaten group cohesion (Bakke, 2014). In Chechnya, tactical and ideological disputes between foreign and local fighters adversely affected group cohesion (Bakke, 2014; Rich & Conduit, 2015) and resulted in a split between local and transnational Islamists (Bakke, 2014). 137 A consistent theme across Malet?s (2013) rich historical analysis of FFs in civil wars is the comparative discord between locals and FFs (Malet, 2013, p. 207), which makes schisms between them inevitable (Malet, 2013, pp. 200-201). Evidence indicates that foreign and local fighters mobilize for different reasons and these dissimilarities resulted in major strategic and political schisms in rebel groups that drew heavily on FFs (Malet, 2013, p. 2; Mironova, 2019, p. 144). For instance, during the 1980s, Afghan domestic rebels clashed with newly arrived FFs over the conflict?s objective (see also Donnelly et al., 2017). Whereas local Afghan rebels insisted that the goal of the struggle was regime change (Malet, 2013, p. 173) and adopted a more pragmatic fighting style ? toggling between the battlefield and their homes (Bergen, 2006, p. 50, as cited in Malet, 2013, p. 179) ? the international mujahedeen saw the Afghan conflict as the inauguration of a global war in the name of Jihad (Bergen & Reynolds, 2005), which later led to the creation of Al- Qaeda in 1988 (Malet, 2013, p. 184). Similarly, many locals joined ISIS out of a rational desire for good governance (Byman, 2019, p. 195). FFs, by contrast, were driven by a supranational ideology and employed radical tactics in order to pursue their goal of an Islamic Caliphate (Rich & Conduit, 2015). This alienated local combatants and disrupted the group?s internal cohesion (Rich & Conduit, 2015; Mironova, 2019, p. 209). Likewise, in Iraq, FFs brought transnational ideologies to the conflict, whereas many local fighters fought for tribal, local, or national aims, and these differences resulted in crucial doctrinal disputes among the insurgent groups (Donnelly et al., 2017, p. 17). In Syria, thousands of FFs flocked to the country to join the war against Bashar al-Assad?s regime. Some were 138 motivated by romantic ideas of revolution, some by contempt for the Syrian regime, and others by a commitment to establish an Islamic Caliphate (Abdul-Ahad, 2012). Others were veterans from other Jihadist conflicts, such as Iraq, Yemen, and Afghanistan (Abdul-Ahad, 2012). Tensions between local rebels, whose goal was to overthrow Bashar al-Assad?s regime, and Islamist FFs, who were fighting for the establishment of a global Islamic Caliphate, resulted in fierce inter-rebel wars (Gade et al., 2019). In both Iraq and Syria, most of the FFs had only a short history with any organized Islamist insurgency. Many of them were seeking martyrdom rather than pursuing any explicitly delineated political goals (Cordesman, 2005). 3.4.3 Strategic Disagreements Strategic disagreements among rebels emerge as a result of discrepancies concerning motivations for fighting, preferable tactics to use, and different expectations about the war?s outcomes. Within the Chechen insurgency, nationalists opposed the targeting of Russian civilians out of concern that it would delegitimize the movement?s cause and validate Russian president Vladimir Putin?s claims that the insurgency was part of an Islamist international terrorism network (C. Moore & Tumelty, 2008; Bakke, 2014, pp. 181-183). Al-Shabaab?s leaders considered FFs useful as long as they were cooperative. The group was receptive of the financial, material, and professional assistance the FFs brought, but refused to submit to their orders or strategic suggestions (Bacon & Muibu, 2019, p. 295). Even when local rebel leaders accept the strategies that FFs promote, local commanders may resist the change, fostering intragroup divisions that make it harder to maintain organizational cohesion (Bakke, 2014). Indeed, fragmentation often 139 occurs when disgruntled commanders lead a group of loyal combatants out of the rebel group (Ives, 2019; Nagel & Doctor, 2020; Stern, 2022a). Splits are more likely to occur when an opportunity to reach a negotiated settlement with the government emerges. An internal impasse between extremist and moderate factions about whether to continue fighting or to negotiate a peace agreement (Stedman, 1997; Kydd & Walter, 2002; Nilsson & Kovacs, 2011; Findley & Rudloff, 2012) can exacerbate divisions within the rebel group, thus increasing the likelihood of breakaway factions (Duursma & Fliervoet, 2021). FFs are found to hinder attempts to achieve a negotiated settlement, instead opting to continue their struggle against the government (Chu & Braithwaite, 2017). For example, FFs in Afghanistan were more averse to accepting ceasefires as a result of their different and more extreme ideological motivations (Gall, 2014, pp. 16-17, as cited in Donnelly et al., 2017, p. 14). Across Jihadi groups fighting in Iraq and Syria, FFs were more likely to reject negotiating with the government, even if it would have been favorable for the group (Mironova, 2019, p. 144). According to a local Jabhat al-Nusra fighter, FFs always called for ??new military operations and were absolutely against any deals with the regime? (Mironova, 2019, p. 144). 3.4.4 Segregation As a function of their different national, linguistic and cultural backgrounds, FFs tend to self-segregate and create their own subgroups within the insurgent organization. Segregated units can precipitate a source of opposition to the greater organization, undermine its overall cohesion, and exacerbate fragmentation (Wesbrook, 1980; King, 2013, pp. 32, 73). Strong bonds between fighters at the combat unit level can give rebel 140 commanders greater confidence in their ability to mobilize the combatants needed to build a functional rebel splinter faction (Ives, 2019; Nagel & Doctor, 2020; Stern, 2022a). In Afghanistan, despite declaring themselves members of a transnational Islamic community, wealthy Arabs volunteers from the Gulf coalesced into Saudi-dominated units and established all-Saudi training camps. Muslim FFs from Thailand, the Philippines, and other Southeast Asian countries trained separately and formed their own ethnic-based combat unit (Bergen, 2006, pp. 41-42, 53, as cited in Malet, 2013, p. 176; Conboy, 2006, pp. 50-51, as cited in Malet, 2013, pp. 177-178). FFs who joined the Croatian forces in the 1990s tended to stick together. They would usually coalesce into units exclusively comprised of FFs led by English-speaking commanders. If the number of non-Croatian volunteers was high enough, FFs even formed their own international units to circumvent language complications. Unlike local Croats that fought for independence, many volunteers integrated internationalist beliefs, originating from their own foreignness and, in extension, their units? esprit de corps (Hutt, 2010, p. 104, as cited in Arielli, 2012, p. 9). In both Jabhat al-Nusra and ISIS, FFs often segregated into subgroups according to ethnicity, nationality, language, and even according to hometowns. This divided the fighters even further, strengthened the bonds among fighters at the combat unit level, and undermined overall organizational cohesion (Mironova, 2019, p. 142). According to a local Jabhat al-Nusra fighter, ?Those ethnic groups were sometimes annoying because it is hard for anyone to get close to them? (Mironova, 2019, p. 162). Another local fighter added, ?My problem with foreign fighters was their lack of trust in us. We 141 did not interact a lot and when we are all fighting, they stick to their ethnic groups in fear of betrayal from us? (Mironova, 2019, p. 162). On the other hand, FFs expressed similar concerns that the local fighters would abandon them in battle (Mironova, 2019, p. 162). In ISIS, close ethnic subgroups sometimes circumvented official channels ? resolving issues, making major decisions, and even setting up unofficial chains of command on their own. This separation along ethnic and linguistic lines further undermined local leadership?s authority over those foreign subgroups (Mironova, 2019, p. 143). ISIS?s Chechen constituency was particularly known for banding together and forming their own units (Abdul-Ahad, 2012; Mironova, 2019, p. 154). Whenever one Chechen fighter was promoted, he would press for the promotion of his fellow Chechens as well (Mironova, 2019, p. 154). A local Arab referred to them as the muhajiroun, or ?immigrants? (Abdul-Ahad, 2012), further illustrating their estrangement from local combatants. The problem of the self-segregation of FFs ?resulted in internal conflicts between different ethnic subgroups? and decreased the overall cohesion of the rebel group (Mironova, 2019, p. 143). Matters worsened to such a point that military leader Umar Shishani had to especially address commanders saying, ?Do not chose people close to you based on where they came from, but only based on their fear of God and their professional qualities? (Mironova, 2019, p. 143). In sum, considering the theoretical mechanisms and rich empirical evidence presented above, I expect that the presence of FFs would lead to a higher probability of fragmentation at the rebel group level. 142 3.5 Hypothesis 1 All else equal, insurgent groups that recruit FFs will be more likely to fragment than groups with a predominantly local membership base. However, not all FFs exhibit equal levels of foreignness to local conflicts (P. Moore, 2019), with potential implications for group cohesion. Existing studies find that rebel groups that recruit from excluded ethnic groups have a stronger sense of collective solidarity (Wucherpfennig et al., 2012) compared to more ethnically diverse rebel groups (Humphreys & Weinstein, 2006). Shared traits ? such as ethnicity, language, culture, and norms ? improve communication between members, thereby helping to narrow preference divergence, an important source of fragmentation (Bakke et al., 2012; Shapiro, 2013; Warren & Troy, 2015; Woldemariam, 2018; Burch & Ochreiter, 2020). Consequently, ethnically homogenous rebel groups are less likely to experience splits (Gates, 2002; Fjelde & Nilsson, 2018). Rebel groups that recruit on the global stage are more likely to attract ethnically, nationally, linguistically, and culturally diverse individuals than groups that recruit domestically (Malet, 2013). Ethnic differences between foreign and local fighters ? as well as among a diverse body of FFs ? pose significant challenges to group cohesion. For example, FFs that lack the linguistic skills necessary to interact with local fighters reduce the ability of group members to communicate with one another (P. Moore, 2019) or develop strong interpersonal bonds, which are critical for group cohesion (Stern, 2022b). Local fighters may be less forthcoming about working with FFs who do not fully understand local customs (Bakke, 2014; P. Moore, 2019). Poor inter-combatant 143 communication also leads to weaker coordination, increases the likelihood of individual deviance, and decreases cohesion (King, 2006). FFs who flocked to Afghanistan during the 1980s came from various backgrounds (Gerges, 2009, pp. 80-81, as cited in Malet, 2013, p. 173). FFs leaders, most notably Abdullah Azzam, who attempted to foster rebels? unity based upon an Islamist perspective that transcended ethnicity and regionalism, succumbed to the triple threat of ethnic rivalry, envy, and distrust (Magnus & Naby, 2002, p. 142, as cited in Malet, 2013, p. 183; Donnelly et al., 2017, p. 5). Even though ISIS?s members shared the same goal of establishing an Islamic Caliphate, deep linguistic and cultural divisions made accomplishing that goal especially challenging (Ahram, 2019). Ethnic-based prejudices among ISIS?s FFs were highly prevalent, creating deep inter-combatant ruptures. Saudi FFs disdained other Arabs, Central Asians considered themselves superior to Uzbeks, and French-speaking members of ISIS were unwilling to cooperate with Russian-speaking fighters. As time passed, these sentiments and behaviors sparked conflicts both between ethnic subgroups and the insurgent organization as a whole (Mironova, 2019, pp. 154-156). Notably, in the early phases of the war in Iraq, Seifullah al-Shishani and a group of Uzbek fighters split from the leadership of the Georgian-Chechen commander Umar Shishani to form their own group under al-Shishani?s leadership (Donnelly et al., 2017, p. 23; Mironova, 2019, pp. 155-156). A local Jabhat al-Nusra fighter reported that a major pitfall of FFs was that they ?had their own communities and way of living. They weren?t highly integrated into our society? (Mironova, 2019, p. 143). 144 Conversely, sharing ethnic and linguistic traits with local fighters not only enables FFs to better integrate into the group, but also to behave more like locals and, as a result, gain the trust of the local population (P. Moore, 2019). This leads to the observable implication that variation in the type of FFs will be associated with variation in the probability of fragmentation. Specifically, I expect that rebel groups that recruit coethnic FFs will be associated with a decreased likelihood of fragmentation. 3.6 Hypothesis 2 All else equal, insurgent groups that recruit coethnic FFs will be less likely to fragment than groups that recruit non-coethnic FFs. 3.7 Data and Methods I analyze the effect of FFs on the probability of rebel fragmentation in armed intrastate conflict. The first hypothesis is composed of two primary variables: FFs and fragmentation. The main explanatory variable in the second hypothesis is coethnic FFs. I use the variable coethnic FFs to assess the impact of varying degrees of FF foreignness on fragmentation. The dataset I use to test my hypotheses relies mainly on variables and cases drawn from P. Moore (2019) and Duursma & Fliervoet (2021). In addition, I draw on other existing datasets for relevant variables, as described below. The cross-national data includes a sample of 227 rebel groups active in 111 armed intrastate conflicts against their governments between 1989 and 2011 (Melander et al., 2016).29 In total, my data contains 839 observations recorded at the rebel-year level. To test my hypothesis, I estimate a series of binary logistic regression models. 29 Rebel group and conflict data are based on data from the Uppsala Conflict Data Program (UCDP) and The Peace Research Institute Oslo (PRIO) Dataset. 145 Table 3.1: Descriptive statistics. Statistic N Mean St. Dev. Min Max Fragmentation 839 0.02 0.2 0 1 Foreign fighters (lagged) 838 0.4 0.5 0.0 1.0 Coethnic foreign fighters (lagged) 838 0.1 0.2 0.0 1.0 Rebel strength 838 1.7 0.7 1.0 5.0 Islamist rebels 832 0.3 0.4 0.0 1.0 Ethnicity 832 0.6 0.5 0.0 1.0 Social ties 839 0.8 0.4 0 1 Resource extortion 814 0.3 0.5 0.0 1.0 Conflict intensity 839 0.1 0.3 0 1 Conflict duration 839 3.9 5.6 0 34 Multiparty conflict 839 0.4 0.5 0 1 International conflict 839 0.1 0.3 0 1 3.7.1 The Dependent Variable The dependent variable, rebel fragmentation, captures instances in which a segment of a rebel group formally and collectively exits that rebel group and establishes a new, independent rebel organization (Nagel & Doctor, 2020; Doctor, 2020). I rely on Duursma and Fliervoet?s (2021) dataset to code the year in which a rebel split occurs. The dependent variable is coded 1 if a rebel group experiences fragmentation during a given year and 0 otherwise. Given the rebel-year level unit of analysis, fragmentation is a relatively rare event. It occurs 20 times, or in 2.3 percent of the observations. Alternatively, 8.3 percent of the sampled rebel groups fragment at least once, and fragmentation is present in 14 of the 111 armed intrastate conflicts in the data. Figure 3.3 displays the number of active rebel groups that have experienced fragmentation and the total number of active rebel groups, aggregated by year. 146 Figure 3.3: Distribution of rebel fragmentation, 1989 to 2011. 3.7.2 The Independent Variable The primary independent variable is foreign fighters. The variable describes whether FFs joined a rebel group during civil war. Data on FFs come from P. Moore (2019). FFs is coded 1 for any year in which a rebel group recruited FFs and 0 for any year in which the rebel group did not recruit FFs. Because it is reasonable to assume that the effect of FFs on cohesion is not instantaneous, I lagged FFs one year. To assess the impact of varying degrees of FF foreignness, I rely on P. Moore?s (2019) coethnic foreign fighters. The variable accounts for ethnic similarities between FFs and local members of the rebel group (P. Moore, 2019). Coethnic FFs is a binary variable coded 1 if the FFs recruited were coethnics with the majority of the rest of the 147 rebels and 0 otherwise.30 Because it is reasonable to assume that the effect of coethnic FFs on cohesion is not instantaneous, I lagged coethnic FFs one year. P. Moore (2019) reports two drawbacks of Malet?s (2016) data on FFs. First, the data fail to denote the number of FFs in relation to the number of local fighters. Some groups ? such as the AFDL in Congo (Campbell, 1997, as cited in P. Moore, 2019, p. 285) ? are comprised mainly of FFs, whereas other groups are not ? only 3% of the Croatian irregulars? fighting forces in the 1991 conflict against Yugoslav forces, for example, consisted of FFs (Arielli, 2012, as cited in P. Moore, 2019, p. 285). Second, the data does not account for the phase during which the FFs entered into the conflicts. Having such a timeline is important, as FFs may have joined the group after internal divisions were set in motion. However, the diverse anecdotal data provided in this article proposes that FFs enlisted before instances of fragmentation and were often catalysts for rebel fragmentation. Therefore, I do not believe the omission of information on the timing of FFs enlistment skew the results of this analysis. In addition, the covert nature of rebel groups, organizational variation, and extensive scholarly and policy focus on Islamist insurgency groups make it hard to acquire reliable, within-group data on FFs flows (Doctor, 2021). Moreover, the dataset does not include cases of more recent rebel groups, most notably those engaged in civil conflicts in the Middle East that did recruit FFs ? for example, Jabhat al-Nusra. Nevertheless, by testing my hypotheses on cases of rebel groups that most certainly recruited FFs, and considering that the absolute numbers and proportion of FFs who 30 P. Moore (2019) draws information about these two variables from existing data on FFs in civil conflicts collected by Malet (2016). 148 join domestic conflicts has grown overtime (Malet, 2013), the dataset indicates a potential for an even greater effect of FFs on fragmentation. In other words, a fuller picture of FFs joining local insurgents may suggest an even greater impact of FFs on fragmentation. Figure 3.4 displays the total number of active rebel groups and the number of rebel groups that have recruited FFs, aggregated by year. In the data, 70 of the 227 rebel groups sampled recruited FFs. FFs were present in 40, or 36 percent, of the 111 civil conflicts contained in the dataset. Insightfully, Figure 3.4 suggests that in the final years covered in the data, the recruitment of FFs has been on the rise again, probably driven by the recent civil conflicts in the Middle East. Figure 3.4: Distribution of FFs, 1989 to 2011. 149 A cross-tabulation of my two main variables supports my thesis (see Appendix Table 3.6). Of the 20 groups that fragmented, 14 recruited FFs. Of the 294 group-year cases with FFs, fourteen groups or 4.7 percent experienced fragmentation. Of the 544 group-year cases with no FFs, six groups, or 1.1, percent experienced fragmentation. In other words, the rate of fragmentation among rebel groups that recruited FFs is more than four times higher compared with rebel groups that did not recruit FFs. 3.7.3 Control Variables and Alternative Explanations I include a set of control variables across each model to account for alternative explanations of fragmentation and a number of group and conflict-level factors that are likely to influence the relationships between FFs and fragmentation. At the group level, I first control for rebel strength. Stronger rebel groups are more likely to avoid fragmentation (Mahoney, 2020). Because stronger rebel groups tend to be better organized, more effective, and more likely to win the conflict, they are also more attractive to prospective FFs (Malet, 2013, pp. 52-53). On the other hand, weaker groups might frame their struggle as existential in order to attract FFs (Malet, 2013). The rebel strength variable is a compound measure of rebel strength relative to the government from the Nonstate Actors Dataset (NSA) and ranges from 1 to 5 (D. Cunningham et al., 2013). Second, recent evidence shows that Islamist groups are particularly successful in fostering cohesion (Mironova, 2019, p. 98). At the same time, however, extreme ideology is more difficult to control. Ideological differences between Abu Bakr al- Baghdadi?s al-Qaeda in Iraq (AQI) and al-Qaeda?s core under Ayman al-Zawahiri resulted in a split in April 2013, when Baghdadi announced the establishment of ISIS 150 (Donnelly et al., 2017, pp. 22-23). Later, in ISIS and other Islamists groups in Syria, radical Islamist ideology had the opposite effect of what the armed group uses its ideology for, decreasing group cohesion rather than strengthening it (Mironova, 2019, p. 209). Historically, Islamism has not been the sole reason for FFs, nor has it been the only way to explain FFs (Malet, 2013, pp. 2-3; Malet, 2015). However, because of the international community?s recent focus on Islamist insurgency and the fact that Islamist groups experienced high participation from FFs (Walter, 2017; Hegghammer, 2010; Malet, 2013; P. Moore, 2019), I include a binary indicator that accounts for whether an insurgent group is an Islamist rebel group. I use the Foundations of Rebel Group Emergence (FORGE) dataset (J. Braithwaite & Cunningham, 2020) to indicate whether Islam motivated the group?s foundation. Third, compared to more ethnically diverse rebel groups, rebel groups formed around a shared ethnicity have a stronger sense of collective solidarity (Wucherpfennig et al., 2012; Humphreys & Weinstein, 2006) and better mechanisms to inhibit fragmentation (Gates, 2002; Fjelde & Nilsson, 2018). 31 Insurgent organizations endeavor to recruit FFs by equating the fate of the transnational community, of which the FFs and local insurgents are members, with that of distant civil conflicts (Malet, 2010, 2013). I use FORGE (J. Braithwaite & Cunningham, 2020) data to indicate whether the rebel group was explicitly founded around a particular ethnic identity. Fourth, it is generally argued that the cohesion of rebel groups is dependent on the social institutions that support them. More specifically, strong social ties are key to the development of cohesive insurgent organizations (Staniland, 2014; Parkinson, 2013; 31 I separate Islam from ethnic ethnicity because these two types of identities do not always overlap. For example, Arab Islamists in Bosnia, Chechnya and Afghanistan. 151 Fjelde & Nilsson, 2018). Indeed, developing and maintaining organizational cohesion becomes especially challenging if the civilian population the insurgents claim to represent is fractured by differing cleavages (Bakke, 2014; Staniland, 2014). Conversely, strong social ties enable the group to better mobilize locals, thereby reducing the need to recruit FFs. Because FFs are not as connected with the local population, they are less likely to share social ties with major players in local conflicts (P. Moore, 2019). I use Duursma and Fliervoet?s (2021) data to code for social ties. The authors? indicator for social ties is based on FORGE data (J. Braithwaite & Cunningham, 2020). Finally, I control for the rebel group?s source of funding. Rebel groups that rely on material endowment to attract recruits are more likely to attract opportunistic individuals who expect fast payoffs for their fighting (Weinstein, 2007). Among such groups, loyalty is, to a large degree, contingent on the material rewards leaders can deliver to their followers (Weinstein, 2007). When rebel leaders are unable to offer pecuniary rewards to their combatants, the likelihood of fragmentation increases (Lidow, 2016). In addition, commanders may use revenues extorted from natural resources to launch a new group (Nagel & Doctor, 2020).32 I also suggest that groups that extort natural resources will be more likely to attract FFs. The availability of 32 While it is logically intuitive to associate the availability of material opportunities for personal gain with the arrival of FFs, some studies empathize that FFs are often more interested in the nonmaterial goal of the war (cf. Malet, 2013, pp. 8-12; Mironova, 2019, pp. 146-148). For example, Malet (2013, p. 9) finds that in some conflicts, such as the Israeli War of Independence, and the Iraq War, FFs were notified during the recruitment process that they would receive insignificant payment. According to Mironova (2019, p. 146), for many insurgents in ISIS and Jabhat al-Nusra, material benefits were not as appealing as the intangible objectives of the war; their reasons for being in Syria were not because of material incentives, and to focus on financial compensation was seen as unethical. 152 lootable resources allows rebel groups to replace their combatants at a relatively low cost. This often reduces the incentive to invest time and resources in mobilizing local rebels (Weinstein, 2007; Haer et al., 2011; Doctor, 2021). From the combatants? side of the equation, opportunistic foreigners who join the rebel group for material gain will be less likely to undertake certain combat risks and will be more likely to shift loyalties to the highest bidder (Seymour, 2014; Mironova, 2019, p. 148). According to an ISIS fighter, many people who joined for money ?would have converted to Christianity if it paid well? (Mironova, 2019, p. 148). I use the Rebel Contraband Dataset (RCD) (Walsh et al., 2018) data to code for rebel groups that subsist on the extortion of national resources. This is a binary variable which indicates whether a rebel group extorted any of the thirty-one natural resources recorded in the RCD in a given year. I include a number of relevant controls at the conflict level. First, existing studies on rebel fragmentation find that increased levels of conflict intensity affect a group?s probability of splitting (Kriesberg, 1973, p. 249; Staniland, 2014; Woldemariam, 2016). For example, rebel groups descend into internecine conflict when they incur significant military losses (Christia, 2012; Staniland, 2012). Lapses in conflict intensity, however, can contribute to group cohesion (Stein, 1976). For instance, when rebels are relatively satisfied with their group?s prewar institutional arrangements, an increase in conflict intensity will serve as a stimulus to unify the group (McLauchlin & Pearlman, 2012). Moreover, an increase in conflict intensity helps recruiters validate claims that their common group is facing an existential threat and that foreigners? participation is necessary for the survival of their people and, ultimately, themselves (Malet, 2010, p. 100). Relatedly, desperate conflict conditions, a result of intensified conflict, are found 153 to cause insurgents to recruit beyond their local borders (Asal & Malet, 2021). Conflict intensity is a binary variable, which takes the value of 1 if the number of battle-related deaths exceeds 1,000 in a given calendar year and the value 0 if the number of battle- related deaths in a given calendar year is lower. Second, the probability of fragmentation is relatively high when a rebel group initially emerges, but diminishes after a group survives for several years (Carter, 2012). Others argue that as conflict duration increases, alienated members are more likely to split (Joo & Mukherjee, 2021). Regarding FFs, as conflict duration increases, so do the opportunities for rebel recruiters to attract FFs and for prospective FFs to overcome the obstacles involved with their arrival to the battlefield. Additionally, the new ideas that FFs bring with them regarding conflict objectives and tactics are likely to prolong the conflict because it makes the achievement of a negotiated settlement more difficult (Chu & Braithwaite, 2017). To account for those possibilities, I use Duursma and Fliervoet?s (2021) continuous indicator, conflict duration. The variable counts the years since the start of the conflict. Third, I control for conflict-years in which multiple rebel groups are battling the government. All else equal, as the number of rebel groups that occupy the conflict space increases, so does the perceived cost of splitting and, in turn, the likelihood of fragmentation decreases (Fjelde & Nilsson, 2018). Conversely, a larger number of rebel groups increases the opportunities for FFs to ?shop? for the group that aligns with their preferences best. To account for those possibilities, I use Duursma and Fliervoet?s (2021) data to control for multiple rebel groups. This is a binary variable that indicates 154 whether or not the conflict included more than a single rebel group for every conflict- year observed. Finally, studies have found that external intervention by governments shapes the propensity of groups to fragment (D. Cunningham, 2010; Seymour et al., 2016; K. Cunningham, 2017; Tamm, 2016, 2019; Ives, 2019). For example, rebel groups which receive support from foreign sponsors are less likely to fragment because they are better able to compensate their members with selective benefits, which alleviates collective action problems and aids with group cohesion (Carter, 2012). It is also possible that external support for the government against which the rebels fight further conveys a collective sense of injustice which, in turn, stimulates prospective FFs to join local rebels in order to help them in their struggle. Evidence shows that internationalized conflicts generate a particular appeal for FFs (Malet, 2010, 2013). I employ Duursma and Fliervoet?s (2021) binary variable for international conflict. The variable indicates whether the conflict was an international one or not. 3.8 Results Table 3.2 reports the regression coefficients and their standard errors from logistic regression models 1-4. Across all models, the coefficient for FFs is large, positive, and statistically significant, consistent with my first hypothesis. To account for possible sources of within-unit heterogeneity, the coefficient standard errors are clustered by rebel group. I find robust evidence that rebel groups that have recruited FFs are substantially more likely to fragment compared to those that have not. 155 Table 3.2: Logit regression of FFs on rebel fragmentation, 1989 to 2011. Rebel Fragmentation (1) (2) (3) (4) Foreign fighters (lagged) 1.500*** 1.483*** 1.593*** 1.627*** (0.493) (0.542) (0.508) (0.564) Rebel strength 0.087 0.112 (0.327) (0.327) Islamist rebels -0.797 -1.113 (0.830) (0.779) Ethnicity -0.253 -0.244 (0.865) (0.792) Social ties 0.670 1.022 (1.066) (1.013) Resource extortion -1.190 -0.951 (0.755) (0.773) Conflict intensity 0.032 -0.211 (0.669) (0.688) Conflict duration -0.047 -0.051 (0.052) (0.051) Multiparty conflict 0.645 0.930* (0.471) (0.492) International conflict -0.959 -0.545 (0.787) (0.793) Constant -4.496*** -4.559*** -4.587*** -5.119*** (0.411) (0.860) (0.499) (0.940) Observations 838 806 838 806 Log Likelihood -89.294 -85.884 -86.787 -82.861 Akaike Inf. Crit. 182.589 185.768 185.574 187.723 Note: *p**p***p<0.01 Model 1 displays the bivariate relationship between my measure of FFs and the likelihood of fragmentation in the following year without any controls. In support of hypothesis 1, these variables correlate with one another. Other reported regression 156 models further support this correlation. Model 2 incorporates group-level controls. Accounting for these variables, I find support for my first hypothesis: insurgent groups that recruit FFs are more likely to fragment in the following year than groups that abstained from FF recruitment in a given year. Model 3 incorporates conflict-level controls. Again, the FFs variable retains its statistical significance and positive coefficient. Finally, model 4 combines models 1 through 3 to control for potential confounding variables with regard to group and conflict-level factors. On average, FFs increase the probability of rebel fragmentation by nearly 4 percentage points, all else equal. Percentage-wise, FFs increase the probability of fragmentation by 392 percent. Table 3.3 reports the regression coefficients and their standard errors from logistic regression models 5-8. Against the expectations of hypothesis 2, across all models, the coefficient for coethnic FFs is positive, indicating that coethnic FFs are associated with an increased likelihood of rebel fragmentation. This finding contradicts my second hypothesis which predicts that coethnic FFs will decrease the probability of rebel fragmentation. Models 5 and 6 suggest that even when FFs and local insurgents are coethnic, the probability of fragmentation may actually significantly increase. 157 Table 3.3: Logit regression of coethnic FFs on rebel fragmentation, 1989 to 2011. Rebel Fragmentation (5) (6) (7) (8) Coethnic foreign fighters (lagged) 1.157* 0.774 1.160* 0.703 (0.646) (0.686) (0.652) (0.688) Rebel strength 0.349 0.431 (0.295) (0.300) Islamist rebels -0.211 -0.513 (0.830) (0.776) Ethnicity -0.308 -0.470 (0.880) (0.815) Social ties 0.787 1.154 (1.076) (1.025) Resource extortion -1.286* -1.131 (0.754) (0.766) Conflict intensity 0.255 -0.069 (0.673) (0.683) Conflict duration -0.027 -0.026 (0.055) (0.054) Multiparty conflict 0.848* 0.982** (0.467) (0.485) International conflict -0.656 -0.530 (0.795) (0.794) Constant -3.820*** -4.519*** -4.105*** -5.128*** (0.245) (0.819) (0.426) (0.913) Observations 838 806 838 806 Log Likelihood -93.209 -89.436 -90.964 -86.890 Akaike Inf. Crit. 190.418 192.871 193.928 195.780 Note: *p**p***p<0.01 What might explain why coethnic FFs are associated with an increased probability of rebel fragmentation? First, the majority of FFs do not share the same 158 ethnicity as local insurgents (Malet, 2013, p. 54). Indeed, according to my data, only 16 percent of FFs are also coethnic to local rebels (Figure 3.5). This finding raises questions about the value of ethnic homogeneity to group cohesion in general and within the context of FFs in particular. Figure 3.5: Distribution of FFs vs. coethnic FFs, 1989 to 2011. For instance, the rationale the Cuban government offered to many of the 50,000 Cuban FFs who participated in the Angolan Civil War (1975-1978) emphasized ethnic ties between Cuba and Africa. Nevertheless, ethnic division between local and FFs remained salient (Malet, 2020), suggesting that nationality, as opposed to ethnicity, might have played a greater role in promoting tensions between locals and foreigners. 159 Second, even within one seemingly homogeneous ethnic group, there is often a myriad of political objectives and organizational configurations which affect group cohesion (Pearlman & Cunningham, 2012). For example, in 1968 two groups split off from the PFLP ? the PFLP-GC, led by former leaders of the PLF, and the DFLP, which consisted of other former members of the PFLP (Sayigh, 1997; Pearlman, 2011). The IRA in 1969 underwent similar fragmentation, splitting between the OIRA and the PIRA, which itself later broke into the CIRA and the RIRA (Moloney, 2002). In Bangladesh from 1975 to 1992, a number of different groups emerged to represent the Chittagong Hill peoples in their fight against the state (K. Cunningham, 2011). Third, civil war scholars have depicted transnational ethnic diaspora communities as more politically uncompromising and hawkish compared with home constituencies who are often more pragmatic because they bear the direct costs of the conflict (Shain, 2002; Collier, 2000; Roth, 2015; Piazza, 2018). Decisively, allyship between local insurgents and foreigners is often predicated on the most hardline, intransigent members of diaspora communities who supply auxiliary materials and may push for more extreme forms of force than local individuals (Malet, 2013, p. 23). Relatedly, more recently there has been an increase in primarily sectarian transnational identities that perceive themselves more as cosmopolitans than as nationals, such as the Muslim diaspora and other global civil society movements (Sheffer, 2003, pp. 65-68, as cited in Malet, 2013, pp. 22-23). These communities corroborate the assertion that political identities supersede shared ethnic or cultural history and that imagined communities that imbue a sense of identity in its constituents can be constructed on a variety of ideological foundations (Malet, 2013, pp. 22-23). 160 Finally, even groups that recruit FFs along shared traits face major problems in the screening process. In ISIS, although all FFs had to have a local fighter vouch for their abilities, it was difficult to corroborate these accounts (Mironova, 2019, p. 38). Furthermore, while FFs claimed to fight for the caliphate and shared the same identity as their Muslim brothers, many still held to their ethnic roots (Mironova, 2019, p. 151). Many coethnic FFs in Syria and Iraq were unfamiliar with those countries or lacked a wide network of contacts; in some cases, they did not even speak the local language (Mironova, 2019, p. 37), leading to armed disputes between local Arab fighters and Arab FFs who spoke different languages (Mironova, 2019, p. 151). In sum, the result that coethnic FFs are associated with an increased likelihood of rebel fragmentation implies that fighters? ?foreignness? remains a decisive factor in trying to predict fragmentation. Within the scope of this research, coethnicity seems to be insufficient in offsetting the effect of fighters? ?foreignness? on rebel fragmentation. 3.9 Predicted Probabilities Figure 3.6 displays the predicted probabilities of fragmentation associated with variation in FFs. To calculate these values, I use the coefficients and standard errors from model 4 to predict changes in the probability of fragmentation based on changes in the main explanatory variable while holding all control variables at their mean value. Figure 3.6 shows the separate predicted values with 0.95 confidence intervals. 161 Figure 3.6: Predicted probabilities of rebel fragmentation, 1989 to 2011. I find that the predicted probability of fragmentation is 0.008, suggesting that less than 1 percent of rebel groups without FFs are predicted to fragment, holding all other variables at their sample means. For rebel groups that recruit FFs, the predicted probability of fragmentation increases to over 0.04, suggesting that about 4 percent of rebel groups with FFs are predicted to fragment. In sum, FFs are associated with a 3.7 percentage point increase in fragmentation. In percentage change terms, FFs increase the probability of fragmentation by about 392 percent. This predicted change in the displayed values is statistically significant at the 0.95 confidence level. Following a year in which a rebel group recruits FFs, the probability of fragmentation in that group shifts from less than 0.01 to nearly 0.04, all else equal. 162 Given the unit of analysis and the relative rarity of this event, this is a meaningful effect. In sum, my results indicate that FFs are highly correlated with rebel fragmentation. Rebel groups that recruit FFs are significantly and substantively more likely to experience fragmentation, all else equal. The repercussion of not recruiting FFs is a decreased risk of fragmentation, all else equal. 3.10 Selection Effects It is plausible to assume that rebel groups that attract FFs have particular attributes that make them especially appealing to FFs. This implies that FFs are not randomly chosen, but are rather ?selected? into specific types of rebel groups fighting in distinct conflicts. That is, there are certain group and conflict factors which make rebel groups more likely to recruit FFs than others. Concurrently, such factors might also be related to fragmentation. As discussed previously, stronger rebel groups might be more attractive to prospective FFs (Malet, 2013, pp. 52-53). Rebel groups with weaker ties to the local population will find it harder to mobilize local support (Staniland, 2014) and are therefore more likely to fragment (Fjelde & Nilsson, 2018). Islamist insurgency groups may capitalize on their universal message in order to attract FFs (Hegghammer, 2010; Walter, 2017) and are particularly successful in fostering cohesion (Mironova, 2019). If that is indeed the case, selection effects are likely to confound accurate analysis. I employ a bivariate probit selection model to account for the endogenous choice of rebel groups to recruit FFs. The model involves two processes with dichotomous outcomes. In the first equation (the ?selection? stage), the outcome is whether FFs are present or not. I include multiple group and conflict level variables which might explain 163 the probability of FFs joining local civil conflicts. In the second equation (the ?outcome? stage), fragmentation is estimated simultaneously while taking into account the correlation in the equations? error processes. Table 3.4 reports the bivariate probit results. Accounting for self-selection in the recruitment of FFs, the bivariate probit results show that the estimate for FFs is statistically significant. The statistically significant estimate captures the correlation between the two equations, decreases concern for endogeneity bias, and indicates that regular probit should be unbiased. In simpler terms, the results in Table 3.4 show that the effect of FFs on fragmentation is unbiased and, indeed, insurgent groups that recruit FFs will be more likely to fragment, all else equal. 164 Table 3.4: Bivariate probit of FFs on rebel fragmentation, 1989 to 2011. Rebel Fragmentation Foreign fighters (lagged) 2.448*** (0.692) Rebel strength 0.024 (0.095) Islamist rebels -0.983*** (0.278) Ethnicity -0.240 (0.286) Social ties 0.447 (0.397) Resource extortion -0.301 (0.231) Conflict intensity -0.379 (0.250) Conflict duration -0.012 (0.015) Multiparty conflict 0.293 (0.198) International conflict -0.524* (0.241) Constant -2.479*** (0.346) Observations 806 Theta -0.886(-0.997,0.338) Tau -0.693(-0.953,0.219) Total edf 19 Note: *p**p***p<0.01 165 3.11 Robustness Checks and Further Analysis I perform a number of additional tests to assess the robustness of my primary findings. On the whole, these checks continue to provide strong support for hypothesis 1. Full regression results for all robustness tests appear in the appendix. I replicate my analysis using annual fixed effects model (shown in Appendix Table 3.7). My decision to run an annual fixed effects model is because fragmentation does not occur in some years in my sample. Annual fixed effects analysis controls for factors that change each year and are likely to be common to all rebel groups for a given year. For instance, events such as the collapse of the Soviet Union eased travel restrictions and enabled a growing number of individuals to join local forces in battlefield. Results remain robust to this test. In addition, I perform robustness checks using different lag structures for my independent variable, foreign fighters, lagged by one year. Theoretically, the influence of FFs might diminish as FFs become ?domesticated? over the long time they spend in the civil war and with the local population (cf. Bacon & Muibu, 2019). I begin by capturing instances in which a rebel group is recorded as including FFs in either of the first or second lagged years (shown in Appendix Table 3.8). The coefficient?s sign indicates that FFs are positively and significantly associated with fragmentation across all models. These results suggest that the divisive effects of FFs on group cohesion remain profound over time. In order to further check whether the effect of FFs on fragmentation remains robust over time or whether it is associated with a particular lagged year, I captured instances in which a rebel group is recorded as including FFs in either of the first, second, or third lagged years (shown in Appendix Table 3.9). Again, 166 the coefficient?s sign is in the expected direction and results remain statistically significant across the four models. In sum, the results of these models suggest that the effect of FFs on fragmentation remain robust over time. Finally, in addition to Islamist rebel groups, other ideologically-motivated rebel groups are also likely to attract FFs (Malet, 2013, p. 24; Malet, 2020). To address this possibility, I use the FORGE (J. Braithwaite & Cunningham, 2020) dataset?s dichotomous variable ideology to control for whether a rebel group was founded with a specific political ideology or not. Results remain robust to this test too (see Appendix Table 3.10). Indeed, existing studies find that rebel groups formed around a shared identity have advantages which promote cohesion, whereas ideological differences are often associated with fragmentation (Oppenheim et al., 2015; Seymour et al., 2016; Gade et al., 2019; Perkoski, 2019). Nevertheless, when I disaggregated ideology into different categories (e.g., Islamist, leftist) the results are insignificant. Hence, the results demonstrate the importance of ideology for cohesion building, but also imply that certain types of ideologies are likely to impact fragmentation. 3.12 Conclusion In this article, I test hypotheses about the relationship between FFs and rebel fragmentation generally, as well as about the effect of coethnic FFs, a subcategory of FFs, on fragmentation. I find that rebels that recruit FFs are more likely to fragment compared to groups that draw mainly from local fighters. This finding is robust across selection effects, time-variants, and cross-sectional analyses of rebel fragmentation. Results show that the rate of fragmentation among rebel groups that recruited FFs is 167 more than four times higher than rebel groups that did not recruit FFs. Substantively, rebel groups that have recently recruited FFs are more likely to fragment by 392 percent. In addition, I contend that if the foreignness of FFs to local group members makes fragmentation more likely, than rebel groups that recruit coethnic FFs will be less likely to split. However, against my expectation I find that groups that recruit coethnic FFs remain susceptible to fragmentation. In other words, coethnicity seems to be insufficient in offsetting the effect of fighters? ?foreignness? on rebel fragmentation. Fighters? ?foreignness? remains a decisive factor in trying to predict fragmentation. This study contributes to research on rebel fragmentation in a number of ways. First, this article tests directly the scholarly assertion that the presence of FFs increases the probability of group fragmentation. In doing so, this article moves beyond specific qualitative case studies and employs a cross-case study approach to determine the effect of FFs on fragmentation. Second, to better account for variation in the effect of FFs on rebel fragmentation, the article disaggregates FFs according to coethnicity. However, I find that coethnic FFs are associated with an increased likelihood of rebel fragmentation. This finding points to the possible limitations of shared traits for hindering rebel fragmentation. From a policy perspective, fragmentation results in more prolonged and more destructive conflicts. Fragmented insurgent groups are less likely to achieve their goals, whereas cohesive ones tend to be more disciplined and are better able to credibly commit to their program of action. Because of the effect of FFs on fragmentation, governments that seek to end conflicts should limit the flow of FFs into domestic conflicts (Bergen & Reynolds, 2005; P. Moore, 2019). It is also critical to restrict the 168 flow of FFs because of their tendency to transfer their skills, resources, and abilities to new violent actors (Donnelly et al., 2017; Daymon et al., 2020). Relatedly, security and punitive measures to limit the movement of FFs into civil wars should be supplemented by programs set up by the home states to reintegrate them into society (Malet, 2015; Malet & Hayes, 2020). The arrival of FFs may also affect long-term trends, particularly with regards to post-conflict stability and governance. The lack of FFs? commitment to cultivating channels of mutual exchange with local communities should also warrant concern for efforts related to post-conflict democratization (Huang, 2016). For instance, because FFs in Somalia lacked the local knowledge needed to engage in governance activities, their involvement provoked serious backlash with the local population (Bacon & Muibu, 2019, p. 299). But research remains incomplete. First, the specific mechanisms by which FFs influence group fragmentation cannot be confirmed solely through statistical methods. Future qualitative research might elucidate specific processes and pathways by which FFs exacerbate fragmentation. Second, the ratio of FFs to local rebels itself would likely impact some of the group dynamics mentioned above. Thus, future data collection efforts should focus on gathering more disaggregated data on FFs. In particular, there is a need for more information on the numbers of FFs at the group and conflict level as well as their numbers relative to local insurgents (see Monaco, 2016). Additional information should record the timing of the arrival of FFs in civil wars. This information is particularly important for the study of rebel fragmentation because it would allow researchers to 169 better understand whether FFs have joined the group after internal divisions were set in motion or whether FFs have actually caused such divisions (cf. Bacon & Muibu, 2019). Future data collection should also include cases of more recent rebel groups, most notably civil conflicts in the Middle East, that recruited FFs but are not covered in existing data. Finally, future data should aim to better capture whether members who splintered from the organization are local or foreign combatants. While shown to disrupt group dynamics, existing studies suggest that because FFs lack the same protection network local fighters with familial and tribal connections do, they are more likely to obey even the most undesirable commands and remain with the group until the end of the conflict (Mironova, 2019, p. 139). Third, research would benefit from exploring the positions that FFs occupy within the organizational hierarchy. Is fragmentation more or less likely when FFs serve under the command of domestic insurgent leaders? Is fragmentation more or less likely when FFs hold leadership positions or form independent units? Evidently, it seems that FFs are more likely to diminish group cohesion when organized into foreign-controlled units, such as in Chechnya and within ISIS. Conversely, FFs that fought under the command of local leaders, for example in Jabhat al-Nusra and al-Shabaab, had a lesser impact (Rich & Conduit, 2015; Donnelly et al., 2017, p. 29; Bacon & Muibu, 2019). Lastly, variation in the relationship between FFs and fragmentation suggests that different rebel groups manage FFs differently in order to maximize their benefits and reduce negative side effects (cf. Mironova, 2019). Indeed, when FFs are aligned with the organizations they join, they can better serve as a resource to local insurgencies (Chu & Braithwaite, 2017, p. 6). In particular, future research should investigate how rebels 170 socialize FFs into their organization (Stern, 2022b); align a combatant?s preferences (San?n & Wood, 2014); facilitate the creation of social bonds; strengthen individuals? confidence in each other; and generate a sense of collective duty that diminishes attempts to desert or mutiny (Cohen, 2017). 171 3.13 Appendix Appendix Figure 3.7: Proportion of rebel fragmentation in a given year over time. Appendix Figure 3.7 presents the distribution of the dependent variable, rebel fragmentation, over time. In contrast to Figure 3.4 from the main manuscript, this figure illustrates the proportion of rebel groups which are observed to fragment in a given year. This figure also indicates that fragmentation varies not just within groups but across them as well. 172 Appendix Table 3.5: Cases of rebel fragmentation in the data sample. Group Year Foreign fighters Rebel (lagged) Fragmentation AQIM 2011 1 1 ARS/UIC 2008 1 1 CNDD 1998 0 1 CRA 1994 0 1 Fatah 2000 0 1 FRUD 1994 0 1 GIA 1998 1 1 Hizb-i Islami-yi 1990 1 1 Afghanistan JEM 2006 1 1 LURD 2003 1 1 MILF 2011 1 1 MPA 1990 1 1 NLFT 2003 0 1 NPLF 1990 1 1 Palipehutu 1991 1 1 RUF 1997 1 1 SLM/A 2005 1 1 SLM/A 2006 1 1 USC/SSA 1991 0 1 WNBF 1996 1 1 Appendix Table 3.6: Cross tabulations of the main variables. Foreign fighters (lagged) Foreign fighters (lagged) = 0 = 1 Rebel Fragmentation = 0 538 280 Rebel Fragmentation = 1 6 14 173 Appendix Table 3.7: Annual fixed effects, 1989 to 2011. Rebel Fragmentation (1) (2) (3) (4) Foreign fighters (lagged) 1.511*** 1.552*** 1.558*** 1.661*** (0.508) (0.564) (0.522) (0.591) Rebel strength -0.009 0.051 (0.348) (0.357) Islamist rebels -0.940 -1.311 (0.906) (0.912) Ethnicity -0.416 -0.545 (0.955) (0.931) Social ties 1.088 1.560 (1.160) (1.184) Resource extortion -1.152 -0.987 (0.770) (0.795) Conflict intensity 0.087 -0.225 (0.683) (0.736) Conflict duration -0.039 -0.036 (0.052) (0.050) Multiparty conflict 0.598 0.912* (0.495) (0.530) International conflict -0.865 -0.467 (0.813) (0.840) Constant -22.233 -3.267*** -22.512 -4.073*** (5,905.375) (1.157) (5,846.367) (1.263) Observations 838 806 838 806 Log Likelihood -75.460 -72.830 -73.683 -70.629 Akaike Inf. Crit. 198.919 201.660 203.366 205.259 Note: *p**p***p<0.01 174 Appendix Table 3.8: Alternative FFs lag structures (t-1| t-2), 1989 to 2011. Rebel Fragmentation (1) (2) (3) (4) Foreign fighters (t-1| t-2) 1.331*** 1.265** 1.397*** 1.350** (0.493) (0.543) (0.506) (0.561) Rebel strength 0.119 0.164 (0.327) (0.330) Islamist rebels -0.710 -1.009 (0.814) (0.771) Ethnicity -0.241 -0.271 (0.851) (0.786) Social ties 0.710 1.029 (1.052) (1.003) Resource extortion -1.187 -0.962 (0.755) (0.774) Conflict intensity 0.079 -0.143 (0.668) (0.683) Conflict duration -0.039 -0.040 (0.051) (0.050) Multiparty conflict 0.688 0.948* (0.470) (0.492) International conflict -0.951 -0.549 (0.788) (0.794) Constant -4.435*** -4.592*** -4.573*** -5.163*** (0.411) (0.860) (0.503) (0.936) Observations 838 806 838 806 Log Likelihood -90.409 -87.043 -88.018 -84.236 Akaike Inf. Crit. 184.818 188.086 188.036 190.472 Note: *p**p***p<0.01 175 Appendix Table 3.9: Alternative FFs lag structures (t-1| t-2| t-3), 1989 to 2011. Rebel Fragmentation (1) (2) (3) (4) Foreign fighters (t-1|t-2|t-3) 1.193** 1.086** 1.255** 1.162** (0.493) (0.540) (0.505) (0.557) Rebel strength 0.161 0.214 (0.322) (0.325) Islamist rebels -0.617 -0.925 (0.802) (0.762) Ethnicity -0.229 -0.287 (0.844) (0.782) Social ties 0.722 1.041 (1.045) (0.998) Resource extortion -1.201 -0.980 (0.755) (0.774) Conflict intensity 0.124 -0.121 (0.668) (0.685) Conflict duration -0.035 -0.036 (0.051) (0.050) Multiparty conflict 0.727 0.972** (0.469) (0.492) International conflict -0.955 -0.599 (0.790) (0.797) Constant -4.380*** -4.626*** -4.557*** -5.211*** (0.411) (0.854) (0.506) (0.931) Observations 838 806 838 806 Log Likelihood -91.219 -87.808 -88.825 -85.023 Akaike Inf. Crit. 186.439 189.617 189.651 192.045 Note: *p**p***p<0.01 176 Appendix Table 3.10: Ideologically-motivated rebel groups, 1989 to 2011. Rebel Fragmentation (1) (2) Foreign fighters (lagged) 1.467*** 1.497*** (0.526) (0.538) Rebel strength 0.038 0.082 (0.330) (0.327) Ideological rebels -1.522*** -1.511** (0.558) (0.594) Ethnicity -0.097 -0.021 (0.663) (0.676) Social ties 1.074 1.228 (0.928) (0.955) Resource extortion -1.265* -1.052 (0.757) (0.784) Conflict intensity -0.055 (0.686) Conflict duration -0.025 (0.054) Multiparty conflict 0.839* (0.489) International conflict -0.628 (0.796) Constant -4.090*** -4.654*** (0.889) (0.977) Observations 806 806 Log Likelihood -82.752 -80.740 Akaike Inf. Crit. 179.503 183.481 Note: *p**p***p<0.01 177 Conclusion Fragmentation presents challenges for rebel groups, the governments they are battling, and international actors seeking to promote compromise between these actors. While the academic literature on rebel group fragmentation is growing, there remain major gaps in our knowledge as to how and why rebel groups split, and what rebel groups can do to promote cohesion. This dissertation makes theoretical and empirical contributions to our understanding of rebel group fractionalization. In this concluding section, I summarize the theories developed, discuss my central findings, consider some key lessons and implications from this dissertation, and note potential directions for future work. Chapter 1 focuses on an under-theorized social stratum that drives faction emergence: the junior cadres. Combining political opportunity structure with motivation to explain factionalization, Chapter 1 shows how in rebel groups with a centralized leadership, factions are driven by disenfranchised junior cadres and not by disagreements at the leadership level as current literature argues (Pearlman, 2011; Asal et al., 2012; Shapiro, 2013). Original data and personal interviews with Fatah junior cadres reveal that in a centralized rebel group, like Fatah, factions will emerge when the ability of junior cadres to advance in the organization through peaceful mechanisms is systematically blocked by the senior leadership. On the whole, Chapter 1 elucidates the processes by which rebel groups splinter and contributes to the broader discussion on 178 the roles of motivation, structure, and agency in shaping the dynamics of rebel group fragmentation. There are a number of directions to pursue in future research that build on the theory and findings of Chapter 1. Future research should incorporate junior cadres and the rebel groups? decision-making structure into large-N studies in order to assess the theoretical arguments presented here and elsewhere on a larger number of rebel group cases. For example, does the fact that those who factionalized were the junior cadres and not the senior leaders carry any implications for conflict resolution? Furthermore, future researchers might consider how leaders respond to junior cadres? grievances as a way to prevent further factionalization. Why would the same rebel leadership hold on to their seats for such a long time, refusing to share power (or at least the spoils of power) better with junior cadres, especially if doing so will lead to factionalization? Do senior leaders design democratic procedures and form institutions to contain grievances short of factionalization? And if so, what determines their success or failure over time? The examination of Fatah through its different developmental phases ? from a revolutionary movement to a ruling political party in the 1990s ? also affords an opportunity to generalize organizational characteristics such as decision-making structure and junior cadres? access to leadership positions in order to improve our understanding of how these organizational features galvanize or inhibit factionalization among nonviolent actors too. For instance, to what degree are instances of factionalization, such as the Republican Tea Party or the Democratic Squad, representative of organizational structure and members? grievances? 179 In Chapter 2, I argue that the solution to the question of fragmentation lies in rebel socialization. Chapter 2 demonstrates that, as a form of rebel socialization, military training (MT) is uniquely positioned to foster internal cohesion in rebel groups, and makes the theoretical and empirical connection between MT and fragmentation explicit. Frequent interactions during MT produce deep emotive and professional bonds which can alter individual preferences and motivate the combatant to behave in ways he would otherwise not (Lott & Lott, 1965; King, 2013, p. 29). In line with the theoretical expectations of Chapter 2, I find that rebel groups that conduct MT are significantly less likely to split into multiple groups. In addition, Chapter 2 further develops the leader-junior cadre model in Chapter 1. To fully grasp the structure of rebel groups in the study of fragmentation, the theoretical framework in Chapter 2 accounts for more elaborated dynamics between leaders, commanders, and combatants. Future data collection efforts should aim at collecting more data on MT at the group level.33 Closer attention should be given to the special characteristics of MT and their respective impacts for group cohesion. Moreover, the finding that the effects of MT on fragmentation are strongest in the short-term but diminish overtime implies that MT alone may be insufficient to inhibit fragmentation. Future research may want to investigate how rebel groups supplement MT with other forms of socialization in order to build greater internal cohesion. Beyond fragmentation, future studies will explore the relationship between rebel solicitation and rebel group behavior. For example, are 33 The data on MT were not collected for the particular analysis in Chapter 2 and as such do not provide a decisive test. 180 groups that build cohesion through MT ? as opposed to political education ? more likely to victimize civilians? Or win concessions from the government? From Angola to Afghanistan to Croatia, Chapter 3 utilizes insights from an array of case studies in order to examine the scholarly assertion that the presence of FFs increases the probability of group fragmentation (Bakke, 2014; Rich & Conduit, 2015; Doctor, 2021). To that end, Chapter 3 is the first cross-national analysis to explicate the theoretical and empirical connection between FFs and the probability of rebel fragmentation. I find that insurgent organizations that recruit FFs are significantly more likely to split. Insightfully, however, I find that rebel groups that recruit coethnic FFs are not less likely to splinter. This finding raises questions about the value of ethnic homogeneity in the context of FFs and implies that fighters? ?foreignness? remains a decisive factor in trying to predict fragmentation. However, the mechanisms by which FFs influence group fragmentation cannot be confirmed solely through statistical methods. Future qualitative research might elucidate specific processes and pathways by which FFs exacerbate fragmentation. Future data collection efforts should also focus on gathering more disaggregated data on FFs. For example, more information on the numbers of FFs at the group and conflict level, as well as their numbers relative to local insurgents, the timing of the arrival of FFs in civil wars, data on cases of more recent rebel groups primarily in the Middle East, and data on whether members who splintered from the organization are local or foreign combatants. In addition, future research would benefit from exploring the positions that FFs occupy within the organizational hierarchy. Is fragmentation more or 181 less likely when FFs serve under the command of domestic insurgent leaders? Is fragmentation more or less likely when FFs hold leadership positions or form independent units? Lastly, variation in the relationship between FFs and fragmentation suggests that when FFs are aligned with the organizations they join, they can better serve as a resource to local insurgencies (Chu & Braithwaite, 2017, p. 6). Future research should investigate how rebels socialize FFs into their organization (Stern, 2022b) in order to create a deeper change in one?s sense of self (Checkel, 2017; H?ppner, 2017); align a combatant?s preferences; generate a sense of collective duty that diminishes attempts to desert or mutiny (Cohen, 2017); and so on. Taken together, this dissertation carries several potential implications for other theories of political violence and civil war. First, fragmentation is one way for a new rebel group to emerge (Fjelde & Nilsson, 2018). Investigating the microfoundations of fragmentation would deepen research on how and why rebel groups form to begin with (Staniland, 2014; Lewis, 2017; Larson & Lewis, 2018; Walter, 2019; Lewis, 2020; J. Braithwaite & Cunningham, 2020). Second, this project promotes the notion that cohesion should be perplexing for scholars of civil conflicts. In doing so, it signifies a positive development in the way we approach studying the inner workings of rebel organizations. Third, by explicating the theoretical and empirical connection between FFs and the probability of rebel fragmentation, this dissertation contributes to a growing body of literature on the transnational dimensions of violent conflicts (cf. Salehyan, 2009, 2010; Checkel, 2013; Salehyan et al., 2014). Several policy implications can be derived from each chapter. First, understanding how rebel groups are able to retain members effectively is indispensable 182 for governments that are fighting rebels because it can facilitate the development of effective tools for detaching members from the rebel group (Haer et al., 2011). Second, existing research finds that states are more likely to negotiate with rebel leaders that come to power through local selection processes such as elections (K. Cunningham & Sawyer, 2019). If the exclusion or inclusion of junior cadres from senior leadership positions tell us something about rebel groups cohesion, governments interested in negotiating with the rebels should not only monitor how leadership has been reshuffled but also whether junior cadres were incorporated in the process in a meaningful way and whether they gained access to senior leadership positions. Third, junior cadres? expressions of their grievances are an invaluable testimony to the level of their satisfaction with existing institutional arrangements. Because factionalization is mostly driven by aggrieved junior cadres, identifying the roots of their grievances can help policy analysts to assess in-group differences and forecast organizational splits. Depending on the government?s objectives, such insights can help it to identify which individuals constitute an opportunity for dialogue, isolate extremists, and exacerbate intragroup divisions in order to create fissures. Fourth, cohesive groups tend to be more disciplined and are better able to credibly commit to their program of action. A government that seeks to support a rebel group may therefore want to consider assistance in the form of MT. Conversely, a government that seeks to weaken a rebel group should target the group?s MT, if such training indeed takes place. Fifth, because of the effect of FFs on fragmentation, governments that seek to end conflicts should limit the flow of FFs into domestic conflicts. It is also critical to 183 restrict the flow of FFs because of their tendency to transfer their skills, resources, and abilities to new violent actors (Bergen & Reynolds, 2005; Donnelly et al., 2017; Daymon et al., 2020). In addition, security and punitive measures to limit the movement of FFs into civil wars should be supplemented by programs set up by the home states to reintegrate them into society (Malet, 2015; Malet & Hayes, 2020). Finally, the arrival of FFs may also affect long-term trends, particularly with regards to post-conflict stability and governance. The lack of FFs? commitment to cultivating channels of mutual exchange with local communities should also warrant concern for efforts related to post-conflict democratization (Huang, 2016). 184 Bibliography Abdul-Ahad, G. (2012, September 23). Syria: The foreign fighters joining the war against Bashar al-Assad. The Guardian. https://www.theguardian.com/world/2012/sep/23/syria-foreign-fighters- joining-war Abrahms, M., & Potter, P. B. K. (2015). Explaining Terrorism: Leadership Deficits and Militant Group Tactics. International Organization, 69(2), 311?342. https://doi.org/10.1017/S0020818314000411 Abu-Amr, Z. (1993). Hamas: A Historical and Political Background. Journal of Palestine Studies, 22(4), 5?19. https://doi.org/10.2307/2538077 Abu Toameh, K. (2008). Freed Fatah member wants reforms. Jerusalem Post. Available from: https://www.jpost.com/middle-east/freed-fatah-member- wants-reforms. Ahram, A. I. (2019). Sexual violence, competitive state building, and Islamic State in Iraq and Syria. Journal of Intervention and Statebuilding, 13(2), 180?196. https://doi.org/10.1080/17502977.2018.1541577 Akcinaroglu, S. (2012). Rebel interdependencies and civil war outcomes. Journal of Conflict Resolution, 56(5), 879?903. https://doi.org/10.1177/0022002712445741 185 al-Ghani al-Shami, A. (2016). Conferences of the ?Fatah? Movement Since its Inception [online], Quds Press International News Agency. Available from: http://www.qudspress.com/index.php?page=show&id=25753. al-Omari, G. (2016). Fatah?s General Conference: Clarifying Succession and Managing Dissent, The Washington Institute for Near Eastern Policy. Available from: https://www.washingtoninstitute.org/policy-analysis/fatahs- general-conference-clarifying-succession-and-managing-dissent. al-Tawheed, L. (2021). Database: Identifiers of Designated Islamic Terrorist Organizations [online], Jihad Intel. Available from: https://jihadintel.meforum.org/group/174/liwa-al-tawheed. Aly, A.M.S., Feldman, S., & Shikaki, K. (2013). Arabs and Israelis: Conflict and peacemaking in the Middle East. New York: Palgrave Macmillan. Arielli, N. (2012). In search of meaning: Foreign volunteers in the Croatian Armed Forces, 1991?95. Contemporary European History, 21(1), 1?17. Arquilla, J., & Karasik, T. (1999). Chechnya: A glimpse of future conflict? Studies in Conflict & Terrorism, 22(3), 207?229. https://doi.org/10.1080/105761099265739 Asal, V., Brown, M., & Dalton, A. (2012). Why Split? Organizational Splits Among Ethnopolitical Organizations in the Middle East. Journal of Conflict Resolution, 56(1), 94?117. https://doi.org/10.1177/0022002711429680 Asal, V., & Malet, D. (2021). Nobody more terrible than the desperate: Conflict conditions and rebel demand for foreign fighters. Studies in Conflict & Terrorism, 1?19. https://doi.org/10.1080/1057610X.2021.1961715 186 B2. Election Results for the Fatah Central Committee, Bethlehem (2009). Journal of Palestine Studies, 39 (1), 170-171. Bacon, T., & Muibu, D. (2019). The domestication of Al-Shabaab. The Journal of the Middle East and Africa, 10(3), 279?305. https://doi.org/10.1080/21520844.2019.1658986 Baines, E. (2014). Forced marriage as a political project: Sexual rules and relations in the Lord?s Resistance Army. Journal of Peace Research, 51(3), 405?417. https://doi.org/10.1177/0022343313519666 Bakke, K. M. (2014). Help wanted? The mixed record of foreign fighters in domestic insurgencies. International Security, 38(4), 150?187. https://doi.org/10.1162/ISEC_a_00156 Bakke, K. M., Cunningham, K. G., & Seymour, L. J. M. (2012). A Plague of Initials: Fragmentation, Cohesion, and Infighting in Civil Wars. Perspectives on Politics, 10(2), 265?283. https://doi.org/10.1017/S1537592712000667 Bapat, N. A. (2007). The internationalization of terrorist campaigns. Conflict Management and Peace Science, 24(4), 265?280. https://doi.org/10.1080/07388940701643607 Bara, C. (2014). Incentives and opportunities: A complexity-oriented explanation of violent ethnic conflict. Journal of Peace Research, 51 (6), 696-710. Baroud, R. (2019). As Abbas Ages, Fatah Moves to Consolidate Power. The Washington Report on Middle East Affairs, 38 (2), 49-50. Barrett, R. (2017). BEYOND THE CALIPHATE: Foreign Fighters and the Threat of Returnees. The Soufan Center. https://thesoufancenter.org/wp- 187 content/uploads/2017/11/Beyond-the-Caliphate-Foreign-Fighters-and-the- Threat-of-Returnees-TSC-Report-October-2017-v3.pdf Bartov, O. (1992). Hitler?s Army: Soldiers, Nazis, and War in the Third Reich. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780195079036.001.0001 BBC News (2004). Al-Aqsa Sets Out Palestinian Reform Plan, Disengagement Deal Conditions, 28 July BBC News, 28 July. Ben-Shalom, U., Lehrer, Z., & Ben-Ari, E. (2005). Cohesion During Military Operations: A Field Study on Combat Units in the Al-Aqsa Intifada. Armed Forces & Society, 32(1), 63?79. https://doi.org/10.1177/0095327X05277888 Berman, E. (2009). Radical, Religious, and Violent: The New Economics of Terrorism. The MIT Press. https://doi.org/https://doi.org/10.7551/mitpress/7881.001.0001 Berman, P. (1974). Revolutionary Organization: Institution-Building within the People?s Liberation Armed Forces. Heath: Lexington Books. Bergen, P. L. (2006). The Osama bin Laden I know: An oral history of al-Qaeda?s leader. Simon and Schuster. Bergen, P., & Reynolds, A. (2005). Blowback revisited: Today?s insurgents in Iraq are tomorrow?s terrorists. Foreign Affairs, 84(6), 2?6. https://doi.org/10.2307/20031771 Blattman, C., & Miguel, E. (2010). Civil war. Journal of Economic Literature, 48(1), 3-57. 188 Bloom, M. M. (2004). Palestinian suicide bombing: Public support, market share, and outbidding. Political Science Quarterly, 119(1), 61?88. https://doi.org/10.2307/20202305 Bloom, M. (2005). Dying to Kill: The Allure of Suicide Terror, New York: Columbia University Press. Braithwaite, A., & Chu, T. S. (2018). Civil conflicts abroad, foreign fighters, and terrorism at home. Journal of Conflict Resolution, 62(8), 1636?1660. https://doi.org/10.1177/0022002717707304 Braithwaite, J. M., & Cunningham, K. G. (2020). When Organizations Rebel: Introducing the Foundations of Rebel Group Emergence (FORGE) Dataset. International Studies Quarterly, 64(1), 183?193. https://doi.org/10.1093/isq/sqz085 Breslawski, J. (2018). In the Spotlight: How International Attention Affects Militant Behavior. Terrorism and Political Violence, 33 (1), 1-23. Breslawski, J., & Ives, B. (2018). Killing for God? Factional Violence on the Transnational State. Journal of Conflict Resolution, 20 (10), 1-27. Br?ning, M. (2013). Political Parties in Palestine: Leadership and Thought. New York: Palgrave Macmillan. Brown, N. J. (2010). Are Palestinians Building a State? Carnegie Commentary [online]. Available from: http://www.carnegieendowment.org/files/palestinian_state1.pdf. Brynen, R. (1995). The Neopatrimonial Dimension of Palestinian Politics. Journal of Palestine Studies, 25 (1), 23-36. 189 Buhaug, H., Cederman, L., & R?d, J. K. (2008). Disaggregating Ethno-Nationalist Civil Wars: A Dyadic Test of Exclusion Theory. International Organization 62 (3), 531?551. Burch, M., & Ochreiter, L. (2020). The emergence of splinter factions in intrastate conflict. Dynamics of Asymmetric Conflict, 13(1), 47?66. https://doi.org/10.1080/17467586.2019.1650385 Byman, D. (2019). Road warriors: Foreign fighters in the armies of jihad. Oxford University Press. Byman, D., Chalk, P., Hoffman, B., Rosenau, W., & Brannan, D. (2001). Trends in Outside Support for Insurgent Movements. Santa Monica, CA: RAND Corporation. Campbell, S. (1997). What Kabila is hiding: Civilian killings and impunity in Congo. Human Rights Watch. https://www.hrw.org/report/1997/10/01/what-kabila- hiding/civilian-killings-and-impunity-congo Cantin, M. O. (2021). Pathways to Violence in Civil Wars: Combatant Socialization and the Drivers of Participation in Civilian Targeting. International Studies Review, 23(4), 1566?1594. Carter, D. B. (2012). A Blessing or a Curse? State Support for Terrorist Groups. International Organisation, 66 (1), 129-151. https://doi.org/10.1017/S0020818311000312 Cederman, L. E., Weidmann, N. B., & Gleditsch, K. S. (2011). Horizontal Inequalities and Ethnonationalist Civil War: A Global Comparison. American Political Science Review, 105 (3), 478-495. 190 Checkel, J. T. (2017). Socialization and violence: Introduction and framework. Journal of Peace Research, 54(5), 592?605. https://doi.org/10.1177/0022343317721813 Checkel, J. T. (Ed.). (2013). Transnational dynamics of civil war. Cambridge University Press. Cheong, D. D. (2012). Fatah?s al-Aqsa Martyrs? Brigades: Armed struggle as a response to a crisis of legitimacy. Melbourne Journal of Politics, 35, 39-59. Christia, F. (2012). Alliance formation in civil wars. New York: Cambridge University Press. https://doi.org/10.1017/CBO9781139149426 Chu, T. S., & Braithwaite, A. (2017). The impact of foreign fighters on civil conflict outcomes. Research & Politics, 4(3), 205316801772205. https://doi.org/10.1177/2053168017722059 Clarke, C. P. (2015). Terrorism, Inc.:The Financing of Terrorism, Insurgency, and Irregular Warfare. United States: ABC-CLIO/PRAEGER. Cobban, H. (1984). The Palestinian Liberation Organization: People, Power, and Politics. Cambridge: Cambridge University Press. Cockerham, W. C. (1978). Attitudes Toward Combat Among U.S. Army Paratroopers. Journal of Political & Military Sociology, 6(1), 1?15. Cohen, D. K. (2013). Explaining Rape during Civil War: Cross-National Evidence (1980?2009). American Political Science Review, 107(3), 461?477. https://doi.org/10.1017/S0003055413000221 Cohen, D. K. (2016). Rape During Civil War. Cornell University Press. https://doi.org/https://doi.org/10.7591/9781501705984 191 Cohen, D. K. (2017). The ties that bind: How armed groups use violence to socialize fighters. Journal of Peace Research, 54(5), 701?714. https://doi.org/10.1177/0022343317713559 Cohen, D. K., & Nord?s, R. (2014). Sexual violence in armed conflict: Introducing the SVAC dataset, 1989?2009. Journal of Peace Research, 51(3), 418?428. https://doi.org/https://doi.org/10.1177/0022343314523028 Collier, P. (2000). Economic causes of civil conflict and their implications for policy. World Bank. https://web.worldbank.org/archive/website01241/WEB/IMAGES/ECONONM I.PDF Collier, P., & Hoeffler, A. (2004). Greed and grievance in civil war. Oxford Economic Papers, 56(4), 563-595. Collier, P., & Hoeffler, A. E. (1999). Justice-seeking and loot-seeking in civil war, (Report No. 28151), The World Bank. Available from: http://documents.worldbank.org/curated/en/631151468782127692/Justice- seeking-and-loot-seeking-in-civil-war. Collier, P., & Hoeffler, A. E. (2002). Greed and grievance in civil war, (Report No. 28126), The World Bank. Available from: http://documents.worldbank.org/curated/en/998891468762911498/Greed-and- grievance-in-civil-war. Conboy, K. J. (2006). The second front: Inside Asia?s most dangerous terrorist network. Equinox Publishing. 192 Cordesman, A. H. (2005). Iraq?s Evolving Insurgency. Center for Strategic and International Studies. https://csis-website-prod.s3.amazonaws.com/s3fs- public/legacy_files/files/media/csis/pubs/050805_iraqi_insurgency.pdf [Unpublished Manuscript] Coser, L. A. (1957). Social Conflict and the Theory of Social Change. The British Journal of Sociology, 8 (3), 204. Cowan, S. (2010). Inside the People?s Liberation Army: A military perspective. European Bulletin of Himalayan Research, 37, 82?116. Cunningham, D. E. (2006). Veto Players and Civil War Duration. American Journal of Political Science, 50(4), 875?892. https://doi.org/https://doi.org/10.1111/j.1540-5907.2006.00221.x Cunningham, D. E. (2010). Blocking resolution: How external states can prolong civil wars. Journal of Peace Research, 47(2), 115?127. https://doi.org/10.1177/0022343309353488 Cunningham, D. E., Gleditsch, K. S., & Salehyan, I. (2013). Non-state actors in civil wars: A new dataset. Conflict Management and Peace Science, 30(5), 516? 531. https://doi.org/10.1177/0738894213499673 Cunningham, D. E., Gleditsch, K. S., Gonz?lez, B., Vidovi?, D., & White, P. B. (2017). Words and Deeds: From Incompatibilities to Outcomes in Antigovernment Disputes. Journal of Peace Research, 54 (4), 468-83. Cunningham, D. E., & Lemke, D. (2013). Combining Civil and Interstate Wars. International Organization, 67(3), 609?627. https://doi.org/10.1017/S0020818313000167 193 Cunningham, K. G. (2011). Divide and conquer or divide and concede: How do states respond to internally divided separatists? American Political Science Review, 105(2), 275?297. https://doi.org/10.1017/S0003055411000013 Cunningham, K. G. (2013). Actor Fragmentation and Civil War Bargaining: How Internal Divisions Generate Civil Conflict. American Journal of Political Science, 57 (3), 659?672. https://doi.org/10.1111/ajps.12003 Cunningham, K. G. (2014). Inside the Politics of Self-Determination. New York: Oxford University Press. Cunningham, K. G. (2017). The Oslo Forum Papers: Understanding fragmentation in conflict and its impact on prospects for peace. Centre for Humanitarian Dialogue, (6), 2-17. Cunningham, K. G., Bakke, K. M., & Seymour, L. J. (2012). Shirts Today, Skins Tomorrow: Dual Contests and the Effects of Fragmentation in Self- Determination Disputes. Journal of Conflict Resolution, 56 (1), 67-93. https://doi.org/10.1177/0022002711429697 Cunningham, K. G., Huang, R., & Sawyer, K.M. (2021). Voting for militants: Rebel elections in civil war. Journal of Conflict Resolution, 65(1), pp.81-107. Cunningham, K. G., & Sawyer, K. (2019). Conflict negotiations and rebel leader selection. Journal of Peace Research, 56(5), 619-634. Dawson, R. E., Prewitt, K., & Dawson, K. S. (1977). Political Socialization: An Analytical Study. Little, Brown. 194 Daymon, C., de Roy van Zuijdewijn, J. H., & Malet, D. (2020). Career foreign fighters: Expertise transmission across insurgencies. Career Foreign Fighters: Expertise Transmission Across Insurgencies. Doctor, A. C. (2020). A Motion of No Confidence: Leadership and Rebel Fragmentation. Journal of Global Security Studies, 5(4), 598?616. https://doi.org/https://doi.org/10.1093/jogss/ogz060 Doctor, A. C. (2021). Foreign fighters and conflict-related sexual violence. International Studies Quarterly, 65(1), 69?81. https://doi.org/10.1093/isq/sqaa087 Donnelly, M. G., Sanderson, T. M., & Fellman, Z. (2017). Foreign Fighters in History. Center for Strategic & International Studies. http://csis-website- prod.s3.amazonaws.com/s3fs- public/publication/171220_history_foreign_fighter_project.pdf Doom, R., & Vlassenroot, K. (1999). Kony?s message: A new koine? The Lord?s Resistance Army in northern Uganda. African Affairs, 98(390), 5?36. https://doi.org/https://doi.org/10.1093/oxfordjournals.afraf.a008002 Driscoll, J. (2012). Commitment Problems or Bidding Wars? Rebel Fragmentation as Peace Building. Journal of Conflict Resolution, 56(1), 118?149. https://doi.org/10.1177/0022002711429696 Duursma, A., & Fliervoet, F. (2021). Fueling Factionalism? The Impact of Peace Processes on Rebel Group Fragmentation in Civil Wars. Journal of Conflict Resolution, 65(4), 788?812. https://doi.org/10.1177/0022002720958062 195 Dyrstad, K. and Hillesund, S. (2020). Explaining support for political violence: grievance and perceived opportunity. Journal of conflict resolution, 64(9), 1724-1753. Eck, K. (2009). From Armed Conflict to War: Ethnic Mobilization and Conflict Intensification. International Studies Quarterly, 53(2), 369?388. https://doi.org/10.1111/j.1468-2478.2009.00538.x Eck, K. (2010). Recruiting rebels: Indoctrination and political education in Nepal. In M. Lawoti & A. K. Pahari (Eds.), The Maoist Insurgency in Nepal: Revolution in the twenty-first century (pp. 33?51). Routledge. Eck, K. (2014). Coercion in Rebel Recruitment. Security Studies, 23(2), 364?398. https://doi.org/10.1080/09636412.2014.905368 Edwards, M. R. (2005). Organizational Identification: A Conceptual and Operational Review. International Journal of Management Reviews, 7(4), 207?230. https://doi.org/10.1111/j.1468-2370.2005.00114.x Emirbayer, M., & Goldberg, C. A. (2005). Pragmatism, Bourdieu, and Collective Emotions in Contentious Politics. Theory and Society, 34 (5-6), 469-518. Faulkner, C. M., & Doctor, A. C. (2021). Rebel Fragmentation and the Recruitment of Child Soldiers. International Studies Quarterly, 65(3), 647?659. https://doi.org/10.1093/isq/sqab031 Fearon, J.D., & Laitin, D.D. (2003). Ethnicity, insurgency, and civil war. American political science review, 97(1), 75-90. 196 Festinger, L., & Carlsmith, J. M. (1959). Cognitive consequences of forced compliance. The Journal of Abnormal and Social Psychology, 58(2), 203?210. https://doi.org/10.1037/h0041593 Findley, M., & Rudloff, P. (2012). Combatant Fragmentation and the Dynamics of Civil Wars. British Journal of Political Science, 42 (4), 879-901. https://doi.org/10.1017/S0007123412000099 Fishman, A. (2015). Fatah on the Verge of Eruption. Ynetnews.com, 23 Oct. Available from: https://www.ynetnews.com/Ext/Comp/ArticleLayout/CdaArticlePrintPreview/ 0,2506,L-4715319,00.html. Fjelde, H., & Nilsson, D. (2012). Rebels against rebels: Explaining violence between rebel groups. Journal of Conflict Resolution, 56(4), 604?628. https://doi.org/10.1177/0022002712439496 Fjelde, H., & Nilsson, D. (2018). The rise of rebel contenders: Barriers to entry and fragmentation in civil wars. Journal of Peace Research, 55(5), 551?565. https://doi.org/10.1177/0022343318767497 Forney, J. F. (2015). Who can we trust with a gun? Information networks and adverse selection in militia recruitment. Journal of Conflict Resolution, 59(5), 824? 849. https://doi.org/10.1177/0022002715576752 Franklin, K. (2004). Enacting masculinity: Antigay violence and group rape as participatory theater. Sexuality Research and Social Policy, 1(2), 25?40. https://doi.org/10.1525/srsp.2004.1.2.25 197 Freeman, M. (2012). Introduction to Financing Terrorism: Case Studies, In: M. Freeman, ed. Financing Terrorism. London: Routledge, 3-6. Friedman, T. L. (1986). Abu Nidal: Life of a P.L.O. Renegade, The New York Times, 1 Jan, p. 4. Frisch, H. (1993). The Palestinian Movement in the Territories: The Middle Command. Middle Eastern Studies, 29 (2), 254-255. Fritz, S. G. (1995). Frontsoldaten: The German Soldier in World War II. University Press of Kentucky. Fuchs, S. (2001). Beyond Agency. Sociological Theory, 19 (1), 24-40. Gade, E. K., Hafez, M. M., & Gabbay, M. (2019). Fratricide in Rebel Movements: A Network Analysis of Syrian Militant Infighting. Journal of Peace Research, 56 (3), 321-335. https://doi.org/10.1177/0022343318806940 Gall, C. (2014). The wrong enemy: America in Afghanistan, 2001-2014. Houghton Mifflin Harcourt. Galletta, A. (2013). Mastering the Semi-Structured Interview and Beyond: From Research Design to Analysis and Publication. New York: New York University Press. Gates, S. (2002). Recruitment and Allegiance: The Microfoundations of Rebellion. Journal of Conflict Resolution, 46(1), 111?130. https://doi.org/10.1177/0022002702046001007 Gates, S. (2017). Membership matters: Coerced recruits and rebel allegiance. Journal of Peace Research, 54(5), 674?686. https://doi.org/10.1177/0022343317722700 198 Gates, S., Hegre, H., Jones, M.P., & Strand, H. (2006). Institutional inconsistency and political instability: Polity duration, 1800?2000. American Journal of Political Science, 50(4), 893-908. Gerard, H. B., & Mathewson, G. C. (1966). The effects of severity of initiation on liking for a group: A replication. Journal of Experimental Social Psychology, 2(3), 278?287. https://doi.org/10.1016/0022-1031(66)90084-9 Gerges, F. A. (2009). The far enemy: Why Jihad went global. Cambridge University Press. Gleditsch, K. S. and Ruggeri, A. (2010). Political opportunity structures, democracy, and civil war. Journal of Peace Research, 47(3), 299-310. Goldman, L., Giles, H., & Hogg, M. A. (2014). Going to extremes: Social identity and communication processes associated with gang membership. Group Processes & Intergroup Relations, 17(6), 813?832. https://doi.org/10.1177/1368430214524289 Griffith, J. (1989). The Army?s New Unit Personnel Replacement and Its Relationship to Unit Cohesion and Social Support. Military Psychology, 1(1), 17?34. https://doi.org/10.1207/s15327876mp0101_2 Gurr, T. R. (1970). Why Men Rebel. New Jersey: Princeton University Press. Gurr, T. R. (2000). Peoples Versus States: Minorities at Risk in The New Century. Washington, D.C.: United States Institute of Peace Press. Haer, R., & Banholzer, L. (2015). The creation of committed combatants. Small Wars & Insurgencies, 26(1), 25-48. 199 Haer, R., Banholzer, L., & Ertl, V. (2011). Create Compliance and Cohesion: How Rebel Organizations Manage to Survive. Small Wars & Insurgencies, 22 (3), 415-434. https://doi.org/10.1080/09592318.2011.581477 Hafez, M. M. (2000). Armed Islamist Movements and Political Violence in Algeria. Middle East Journal, 54(4), 572?591. Hafez, M. M. (2009). Jihad after Iraq: Lessons from the Arab Afghans. Studies in Conflict & Terrorism, 32(2), 73?94. https://doi.org/10.1080/10576100802639600 Hansen, S. J. (2013). Al-Shabaab in Somalia the history and ideology of a militant Islamist group, 2005-2012. Oxford University Press. Hays, N. A., & Bendersky, C. (2015). Not All Inequality is Created Equal: Effects of Status Versus Power Hierarchies on Competition for Upward Mobility. Journal of Personality and Social Psychology, 108 (6), 867-882. Hazen, J. M. (2013). What Rebels Want: Resources and Supply Networks in Wartime. Cornell University Press. https://doi.org/https://doi.org/10.7591/9780801467578 Hecker, T., Hermenau, K., Maedl, A., Hinkel, H., Schauer, M., & Elbert, T. (2013). Does perpetrating violence damage mental health? Differences between forcibly recruited and voluntary combatants in DR Congo. Journal of Traumatic Stress, 26(1), 142?148. https://doi.org/10.1002/jts.21770 Hegghammer, T. (2010). The rise of Muslim foreign fighters: Islam and the globalization of Jihad. International Security, 35(3), 53?94. https://doi.org/10.1162/ISEC_a_00023 200 Hegre, H. (2014). Democracy and armed conflict. Journal of Peace Research, 51(2), 159-172. Henderson, W. D. (1985). Cohesion, the human element in combat: Leadership and societal influence in the armies of the Soviet Union, the United States, North Vietnam, and Israel. National Defense University Press. Herbst, J. (2000). Economic incentives, natural resources and conflict in Africa. Journal of African Economics, 9(3), 270?294. https://doi.org/10.1093/jae/9.3.270 Herbst, P. G. (1962). Autonomous Group Functioning: An Exploration in Behavior Theory and Measurement. Tavistock Publications. Hijazi, S., & Lovatt, H. (2018, March 20). Central Committee (FCC). Mapping Palestinian Politics; European Council on Foreign Relations. https://ecfr.eu/special/mapping_palestinian_politics/central_committee/ Hilal, J. (1998). The Effect of the Oslo Political System, In: G. Giacamen and D. J. Lonning, ed. After Oslo: New Realities, Old Problems. London: Pluto Press, 121-145. Hill, J. N. C. (2009). Identity in Algerian Politics: The Legacy of Colonial Rule. Boulder: Lynne Rienner Publishers. Hoffman, B. (2017). Inside Terrorism. New York: Columbia University Press. H?igilt, J. (2016). Fatah from Below: The Clash of Generations in Palestine. British Journal of Middle Eastern Studies, 43 (4), 456-471. 201 Hoover Green, A. (2016). The commander?s dilemma: Creating and controlling armed group violence. Journal of Peace Research, 53(5), 619?632. https://doi.org/10.1177/0022343316653645 Hoover Green, A. (2017). Armed group institutions and combatant socialization: Evidence from El Salvador. Journal of Peace Research, 54(5), 687?700. https://doi.org/10.1177/0022343317715300 Hoover Green, A. (2018). The Commander?s Dilemma: Violence and Restraint in Wartime. Cornell University Press. https://doi.org/https://doi.org/10.1515/9781501726484 H?ppner, G. (2017). Rethinking socialization research through the lens of new materialism. Frontiers in Sociology, 2, 13. https://doi.org/10.3389/fsoc.2017.00013 Huang, R. (2016). The wartime origins of democratization: Civil war, rebel governance, and political regimes. Cambridge University Press. Humphreys, M., & Weinstein, J. M. (2006). Handling and manhandling civilians in civil war. American Political Science Review, 100(3), 429?447. https://doi.org/10.1017/S0003055406062289 Humphreys, M., & Weinstein, J. M. (2008). Who Fights? The Determinants of Participation in Civil War. American Journal of Political Science, 52 (2), 436- 455. https://doi.org/10.1111/j.1540-5907.2008.00322.x Huntington, S. P. (1968). Political Order in Changing Societies. New Haven: Yale University Press. Hutt, S. (2010). Paint: A boy soldier?s journey. Panic Press. 202 Ives, B. (2019). Ethnic external support and rebel group splintering. Terrorism and Political Violence, 33(7), 1546?1566. https://doi.org/10.1080/09546553.2019.1636035 Jasper, J. (2004). A Strategic Approach to Collective Action: Looking for Agency in Social-Movement Choices. Mobilization: An International Quarterly, 9 (1), 1- 16. Joo, M. M., & Mukherjee, B. (2021). Rebel command and control, time, and rebel group splits. International Interactions, 47(2), 318?345. https://doi.org/10.1080/03050629.2021.1848823 Jubran, M. (1990). Abu Nidal: Portrait of a Renegade. The Washington Report on Middle East Affairs, 22 Feb, p. 9. Kalyvas, S. N. (2006). The Logic of Violence in Civil War. Cambridge University Press. https://doi.org/https://doi.org/10.1017/CBO9780511818462 Kalyvas, S. N. (2008). Ethnic Defection in Civil War. Comparative Political Studies, 41 (8), 1043?1068. Kalyvas, S. N., & Kocher, M. A. (2007). How ?free? is free riding in civil wars? : Violence, insurgency, and the collective action problem. World Politics, 59(2), 177?216. https://doi.org/10.1353/wp.2007.0023 Kenny, P. D. (2011). Organizational Weapons: Explaining Cohesion in the Military (HiCN Working Papers No. 107; Households in Conflict Network). The Institute of Development Studies. Kepel, G. (2002). Jihad: The Trail of Political Islam (A. F. Roberts, Trans.). Harvard University Press. 203 Khalidi, R., & Stork, J. (1983). Behind the Fateh Rebellion. MERIP Reports, (119), 6- 12. King, A. (2006). The Word of Command: Communication and Cohesion in the Military. Armed Forces & Society, 32(4), 493?512. https://doi.org/10.1177/0095327X05283041 King, A. (2007). The Existence of Group Cohesion in the Armed Forces: A Response to Guy Siebold. Armed Forces & Society, 33(4), 638?645. https://doi.org/10.1177/0095327X07301445 King, A. (2013). The Combat Soldier: Infantry Tactics and Cohesion in the Twentieth and Twenty-First Centuries. OUP Oxford. https://doi.org/10.1093/acprof:oso/9780199658848.001.0001 Kohlmann, E. (2004). Al-Qaida?s jihad in Europe: The Afghan-Bosnian network. Berg Publishers. Kolditz, T. A. (2006). Research in In Extremis Settings: Expanding the Critique of ?Why They Fight.? Armed Forces & Society, 32(4), 655?658. https://doi.org/10.1177/0095327X05283853 Kosteas, V. D. (2011). Job Satisfaction and Promotions. Industrial Relations: A Journal of Economy and Society, 50 (1), 174-194. Krause, P. (2013). The political effectiveness of non-state violence: A two-level framework to transform a deceptive debate. Security Studies, 22(2), 259?294. https://doi.org/10.1080/09636412.2013.786914 Kriesberg, L. (1973). The sociology of social conflicts. Prentice-Hall. 204 Kydd, A., & Walter, B. F. (2002). Sabotaging the Peace: The Politics of Extremist Violence. International Organization, 56(2), 263?296. https://doi.org/10.1162/002081802320005487 Larson, J. M., & Lewis, J. I. (2018). Rumors, Kinship Networks, and Rebel Group Formation. International Organization, 72 (4), 871-903. Levitsky, S., & Way, L. A. (2002). Elections Without Democracy: The Rise of Competitive Authoritarianism. Journal of Democracy, 13 (2), 51-65. Levitsky, S., & Way, L. A. (2010). Competitive Authoritarianism: Hybrid Regimes after the Cold War. New York: Cambridge University Press. Levy, J. S. (2008). Case Studies: Types, Designs, and Logics of Inference. Conflict Management and Peace Science, 25 (1), 1-18. Lewis, J. I. (2017). How Does Ethnic Rebellion Start? Comparative Political Studies, 50(10), 1420?1450. https://doi.org/https://doi.org/10.1177/0010414016672235 Lewis, J. I. (2020). How Insurgency Begins: Rebel Group Formation in Uganda and Beyond. Cambridge University Press. https://doi.org/https://doi.org/10.1017/9781108855969 Lichbach, M. I. (1995). The Rebel?s Dilemma. Ann Arbor: University of Michigan Press. https://doi.org/10.3998/mpub.13970 Lidow, N. H. (2016). Violent Order: Understanding Rebel Governance Through Liberia?s Civil War. Cambridge University Press. Lilja, J. (2009). Trapping constituents or winning hearts and minds? Rebel strategies to attain constituent support in Sri Lanka. Terrorism and Political Violence, 21(2), 306?326. https://doi.org/10.1080/09546550902765615 205 Lindekilde, L., Bertelsen, P., & Stohl, M. (2016). Who goes, why, and with what effects: The problem of foreign fighters from Europe. Small Wars & Insurgencies, 27(5), 858?877. https://doi.org/10.1080/09592318.2016.1208285 Little, R. W. (1964). Buddy Relations and Combat Performance. In M. Janowitz (Ed.), The New Military: Changing Patterns of Organization (pp. 195?223). Russell Sage Foundation. Littman, R., & Paluck, E. L. (2015). The cycle of violence: Understanding individual participation in collective violence: cycle of violence. Political Psychology, 36, 79?99. https://doi.org/10.1111/pops.12239 Lott, A. J., & Lott, B. E. (1965). Group cohesiveness as interpersonal attraction: A review of relationships with antecedent and consequent variables. Psychological Bulletin, 64(4), 259?309. https://doi.org/10.1037/h0022386 MacCoun, R. J., Kier, E., & Belkin, A. (2006). Does Social Cohesion Determine Motivation in Combat?: An Old Question With an Old Answer. Armed Forces & Society, 32(4), 646?654. https://doi.org/10.1177/0095327X05279181 MacFarquhar, N. (1993). Hunted by Israel, Fatah Fighters Operate as Clandestine Police. Associated Press, Gaza Strip. Mackenzie, S. P. (1992). Politics and military morale: Current affairs and citizenship education in the British Army 1914-1950. Oxford: Clarendon Press. Magnus, R. H., & Naby, E. (2002). Afghanistan: Mullah, Marx, and mujahid. Westview Press. 206 Mahoney, C. W. (2020). Splinters and schisms: Rebel group fragmentation and the durability of insurgencies. Terrorism and Political Violence, 32(2), 345?364. https://doi.org/10.1080/09546553.2017.1374254 Maksoud, C. (Ed.). (1973). ?Palestine Lives: Interviews with Leaders of the Resistance,? Palestine Research Center and Kuwaiti Teachers Association, 43, 55. Malet, D. (2010). Why foreign fighters?: Historical perspectives and solutions. Orbis, 54(1), 97?114. Malet, D. (2013). Foreign fighters: Transnational identity in civil conflicts. Oxford University Press. Malet, D. (2015). Foreign fighter mobilization and persistence in a global context. Terrorism and Political Violence, 27(3), 454?473. https://doi.org/10.1080/09546553.2015.1032151 Malet, D. (2016). The Foreign Fighter Project: 2016 version. https://davidmalet.com/foreign-fighter Malet, D. (2020). Workers of the world, unite! Communist foreign fighters 1917?91. European Review of History: Revue Europ?enne d?histoire, 27(1?2), 33?53. Malet, D., & Hayes, R. (2020). Foreign fighter returnees: An indefinite threat? Terrorism and Political Violence, 32(8), 1617?1635. https://doi.org/10.1080/09546553.2018.1497987 McLauchlin, T. (2015). Desertion and collective action in civil wars. International Studies Quarterly, 59(4), 669?679. https://doi.org/10.1111/isqu.12205 207 McLauchlin, T. (2020). Desertion: Trust and Mistrust in Civil Wars. Cornell University Press. McLauchlin, T., & Pearlman, W. (2012). Out-Group Conflict, In-Group Unity? Exploring the Effect of Repression on Intramovement Cooperation. Journal of Conflict Resolution, 56(1), 41?66. https://doi.org/10.1177/0022002711429707 Melander, E., Pettersson, T., & Themn?r, L. (2016). Organized violence, 1989?2015. Journal of Peace Research, 53(5), 727?742. Mendelsohn, B. (2011). Foreign fighters?Recent trends. Orbis, 55(2), 189?202. https://doi.org/10.1016/j.orbis.2011.01.002 Meyer, D. S., & Minkoff, D. C. (2004). Conceptualizing Political Opportunity. Social Forces, 82 (4), 1457-1492. Miller, A. D. (1983). PLO: Politics of Survival. New York: Praeger Press. Milshtein, M. (2004). Between Revolution and Statehood: Fatah and the Palestinian Authority (Hebrew). Tel Aviv: The Moshe Dayan Center. Mironova, V. G. (2019). From Freedom Fighters to Jihadists: Human Resources of Non State Armed Groups. Oxford University Press. Moloney, E. (2002). The secret history of the IRA. W. W. Norton. Monaco, L. O. (2016, March 30). Intelligence and National Security in American Society [Remarks]. Intelligence Studies Project, The University of Texas- Austin. https://obamawhitehouse.archives.gov/the-press- office/2016/03/30/remarks-lisa-o-monaco-intelligence-studies-project- university-texas 208 Moore, C., & Tumelty, P. (2008). Foreign fighters and the case of Chechnya: A critical assessment. Studies in Conflict & Terrorism, 31(5), 412?433. https://doi.org/10.1080/10576100801993347 Moore, P. (2019). When do ties bind? Foreign fighters, social embeddedness, and violence against civilians. Journal of Peace Research, 56(2), 279?294. https://doi.org/10.1177/0022343318804594 Mosinger, E. S. (2018). Brothers or others in arms? Civilian constituencies and rebel fragmentation in civil war. Journal of Peace Research, 55(1), 62?77. https://doi.org/10.1177/0022343316675907 Mosinger, E. S. (2019). Balance of loyalties: Explaining rebel factional struggles in the Nicaraguan revolution. Security Studies, 28(5), 935?975. https://doi.org/10.1080/09636412.2019.1662481 Moskos, C. C. (1975). The American Combat Soldier in Vietnam. Journal of Social Issues, 31(4), 25?37. https://doi.org/10.1111/j.1540-4560.1975.tb01009.x Muller, E. N., & Seligson, M. A. (1987). Inequality and Insurgency. American Political Science Review, 81 (2), 425-451. Nagel, R. U. (2021). Conflict-Related Sexual Violence and the Re-escalation of Lethal Violence. International Studies Quarterly, 65(1), 56?68. https://doi.org/10.1093/isq/sqaa086 Nagel, R. U., & Doctor, A. C. (2020). Conflict-Related Sexual Violence and Rebel Group Fragmentation. Journal of Conflict Resolution, 64(7?8), 1226?1253. https://doi.org/10.1177/0022002719899443 Nasiri, O. (2006). Inside the Jihad. Basic Books. 209 Newton, C. (2015). The Rocket Men Who Start Gaza?s Wars. The Daily Beast. Available from: https://www.thedailybeast.com/with-irans-rocket-men-who- start-gazas-wars. Nilsson, D. (2010). Turning weakness into strength: Military capabilities, multiple rebel groups and negotiated settlements. Conflict Management and Peace Science, 27(3), 253?271. https://doi.org/10.1177/0738894210366512 Nilsson, D., & S?derberg Kovacs, M. (2011). Revisiting an elusive concept: A review of the debate on spoilers in peace processes. International Studies Review, 13(4), 606?626. https://doi.org/10.1111/j.1468-2486.2011.01080.x Nivat, A. (2005). The black widows: Chechen women join the fight for independence?and Allah. Studies in Conflict & Terrorism, 28(5), 413?419. https://doi.org/10.1080/10576100500180394 Noonan, M., & Khalil, P. (2014). North American foreign fighters. Journal for Deradicalization, 1, 66?96. Oppenheim, B., Steele, A., Vargas, J. F., & Weintraub, M. (2015). True Believers, Deserters, and Traitors: Who Leaves Insurgent Groups and Why. Journal of Conflict Resolution, 59(5), 794?823. https://doi.org/10.1177/0022002715576750 Ortiz, R. D. (2002). Insurgent Strategies in the Post-Cold War: The Case of the Revolutionary Armed Forces of Colombia. Studies in Conflict & Terrorism, 25(2), 127?143. https://doi.org/10.1080/105761002753502484 210 Parkinson, S. E. (2013). Organizing rebellion: Rethinking high-risk mobilization and social networks in war. American Political Science Review, 107(3), 418?432. https://doi.org/10.1017/S0003055413000208 Pateman, R. (1998). Eritrea: Even the Stones Are Burning. Red Sea Press. Pearlman, W. (2011). Violence, Nonviolence, and the Palestinian National Movements. Cambridge University Press. Pearlman, W. (2013). Rebel fragmentation in Syria and Palestine [online], Foreign Policy. Available from: https://foreignpolicy.com/2013/10/10/rebel- fragmentation-in-syria-and-palestine/. Pearlman, W., & Cunningham, K. G. (2012). Nonstate Actors, Fragmentation, and Conflict Processes. Journal of Conflict, 56 (1), 3-15. https://doi.org/10.1177/0022002711429669 Perkoski, E. (2019). Internal politics and the fragmentation of armed groups. International Studies Quarterly, 63(4), 876?889. https://doi.org/10.1093/isq/sqz076 Peterson, A. L. (1996). Martyrdom and the politics of religion: Progressive Catholicism in El Salvador?s civil war. SUNY Press. Pettersson, T., & Eck, K. (2018). Organized violence, 1989?2017. Journal of Peace Research, 55(4), 535?547. https://doi.org/https://doi.org/10.1177/0022343318784101 Pettersson, T., & ?berg, M. (2020). Organized violence, 1989?2019. Journal of Peace Research, 57(4), 597?613. https://doi.org/https://doi.org/10.1177/0022343320934986 211 Piazza, J. A. (2018). Transnational ethnic diasporas and the survival of terrorist organizations. Security Studies, 27(4), 607?632. https://doi.org/10.1080/09636412.2018.1483615 Pischedda, C. (2018). Wars within wars: Why windows of opportunity and vulnerability cause inter-rebel fighting in internal conflicts. International Security, 43(1), 138?176. https://doi.org/10.1162/isec_a_00322 Plank, F. (2017). When Peace Leads to Divorce: The Splintering of Rebel Groups in Powersharing Agreements. Civil Wars, 19 (2),176-197. Posen, B. R. (1993). Nationalism, the Mass Army, and Military Power. International Security, 18(2), 80?124. https://doi.org/10.2307/2539098 Posner, D. N. (2004). The political salience of cultural difference: Why Chewas and Tumbukas are allies in Zambia and adversaries in Malawi. American Political Science Review, 98(4), 529?545. https://doi.org/10.1017/S0003055404041334 Pressman, J. (2003). The Second Intifada: Background and Causes of the Israeli- Palestinian Conflict. Journal of Conflict Studies, 23 (2), 114-141. Prorok, A. K. (2016). Leader Incentives and Civil War Outcomes. American Journal of Political Science, 60(1), 70?84. https://doi.org/10.1111/ajps.12199 Pugel, J. (2010). Disaggregating the causal factors unique to child soldiering: The case of Liberia. . In S. Gates & S. Reich (Eds.), Child Soldiers in the Age of Fractured States (pp. 160?182). University of Pittsburgh Press. Quinn, P., & Shrader, K. (2005, June 30). Foreigners Responsible for Most Suicide Attacks in Iraq. Associated Press. 212 Rasgon, A. (2015). Fatah Central Committee Profiles, The Washington Institute for Near East Policy. Available from: https://www.washingtoninstitute.org/policy- analysis/fatah-central-committee-profiles. Rich, B., & Conduit, D. (2015). The impact of Jihadist foreign fighters on indigenous secular-nationalist causes: Contrasting Chechnya and Syria. Studies in Conflict & Terrorism, 38(2), 113?131. https://doi.org/10.1080/1057610X.2014.979605 Robinson, G. E. (1998). Authoritarianism with a Palestinian Face. Current History, 97, 13-18. Rodgers, D. (2017). Br?deres in arms: Gangs and the socialization of violence in Nicaragua. Journal of Peace Research, 54(5), 648?660. https://doi.org/10.1177/0022343317714299 Rodgers, D., & Muggah, R. (2009). Gangs as non-state armed groups: The Central American case. Contemporary Security Policy, 30(2), 301?317. https://doi.org/10.1080/13523260903059948 Rogoff, B., Paradise, R., Arauz, R. M., Correa-Ch?vez, M., & Angelillo, C. (2003). Firsthand learning through intent participation. Annual Review of Psychology, 54(1), 175?203. https://doi.org/10.1146/annurev.psych.54.101601.145118 Rohde, D. (2007, October 19). Foreign Fighters of Violent Bent Bolster Taliban. New York Times. Rotar, I. (1998). ?Wahhabism? in the Former Soviet Union: A Real Threat, or Just an Excuse to Settle Accounts with the Opposition? Prism, 4(16). Roth, A. (2015). The role of diasporas in conflict. Journal of International Affairs, 68(2), 289?301. 213 Rudloff, P., & Findley, M. G. (2016). The downstream effects of combatant fragmentation on civil war recurrence. Journal of Peace Research, 53(1), 19? 32. https://doi.org/10.1177/0022343315617067 Ruedy, J. D. (1992). Modern Algeria: The Origins and Development of a Nation. Bloomington: Indiana University Press. Salehyan, I. (2007). Transnational rebels: Neighboring states as sanctuary for rebel groups. World Politics, 59(2), 217?242. https://doi.org/10.1353/wp.2007.0024 Salehyan, I. (2008). No shelter here: Rebel sanctuaries and international conflict. The Journal of Politics, 70(1), 54?66. https://doi.org/10.1017/S0022381607080048 Salehyan, I. (2009). Rebels without Borders: Transnational Insurgencies in World Politics. Cornell University Press. Salehyan, I. (2010). The delegation of war to rebel organizations. Journal of Conflict Resolution, 54(3), 493?515. https://doi.org/10.1177/0022002709357890 Salehyan, I., Siroky, D., & Wood, R. M. (2014). External rebel sponsorship and civilian abuse: A principal-agent analysis of wartime atrocities. International Organization, 68(3), 633?661. https://doi.org/10.1017/S002081831400006X Salverda, N., & Otto, S. (2014). Inside Rebel Groups: The Impact of Vertical and Horizontal Relations on Rebel Group Splintering [Working Paper]. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-328460 San-Akca, B. (2016). States in Disguise: Causes of State Support for Rebel Groups. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780190250881.001.0001 214 San?n, F. G., & Wood, E. J. (2014). Ideology in civil war: Instrumental adoption and beyond. Journal of Peace Research, 51(2), 213?226. https://doi.org/10.1177/0022343313514073 Sayigh, Y. (1997). Armed Struggle and the Search for State: The Palestinian National Movement, 1949-1993. New York: Oxford University Press. Sayigh, Y. (2016). The Fatah Conference: From Liberating the Homeland to Institutionalizing Power, Malcolm H. Kerr Carnegie Middle East Center. Available from: https://carnegie-mec.org/2016/12/29/fateh-conference-from- liberating-homeland-to-institutionalizing-power-pub-66554. Schenker, D. (2000). Inside the Fatah Tanzim [online], The Washington Institute for Near East Policy. Available from: https://www.washingtoninstitute.org/policy- analysis/view/inside-the-fatah-tanzim-a-primer. Schiff, Z., & Ya?ari, E. (1984). Israel?s Lebanon War. New York: Simon & Schuster. Schubiger, L. I. (2015, February). One for all? State violence and insurgent cohesion. Annual Convention of the International Studies Association, New Orleans, LA. Schulhofer-Wohl, J. (2013). Fighting between allies and the civil war in Syria. The Political Science of Syria?s War, 42?44. Schulz, H. L. (2002). The ?Al-Aqsa Intifada? as a Result of Politics of Transition. Arab Studies Quarterly, 24 (4), 21-46. Seymour, L. J. M. (2014). Why Factions Switch Sides in Civil Wars: Rivalry, Patronage, and Realignment in Sudan. International Security, 39(2), 92?131. https://doi.org/10.1162/ISEC_a_00179 215 Seymour, L. J. M., Bakke, K. M., & Cunningham, K. G. (2016). E pluribus unum, ex uno plures: Competition, violence, and fragmentation in ethnopolitical movements. Journal of Peace Research, 53(1), 3?18. https://doi.org/10.1177/0022343315605571 Shain, Y. (2002). The role of diasporas in conflict perpetuation or resolution. SAIS Review (1989-2003), 22(2), 115?144. https://doi.org/10.1353/sais.2002.0052 Shapiro, J. N. (2013). The Terrorist?s Dilemma: Managing Violent Covert Organizations. New Jersey: Princeton University Press. Shapiro, J. N., & Siegel, D. A. (2007). Underfunding in Terrorist Organizations. International Studies Quarterly, 51 (2), 405-29. Shapiro, J. N., & Siegel, D. A. (2012). Moral Hazard, Discipline, and the Management of Terrorist Organizations. World Politics, 64(1), 39?78. https://doi.org/10.1017/S0043887111000293 Sheffer, G. (2003). Diaspora politics: At home abroad. Cambridge University Press. Shikaki, K. (2001). Old Guard, Young Guard: The Palestinian Authority and the Peace Process at Crossroads, Palestinian Center for Policy and Survey Research. Available from: https://www.pcpsr.org/en/node/267. Shikaki, K. (2002). Palestinians Divided. Foreign Affairs, 81 (1), 89-105. Shils, E. (1957). Primordial, Personal, Sacred and Civil Ties: Some Particular Observations on the Relationships of Sociological Research and Theory. The British Journal of Sociology, 8(2), 130?145. https://doi.org/10.2307/587365 216 Shils, E. A., & Janowitz, M. (1948). Cohesion and Disintegration in the Wehrmacht in World War II. Public Opinion Quarterly, 12(2), 280?315. https://doi.org/10.1086/265951 Siebold, G. L. (1999). The Evolution of the Measurement of Cohesion. Military Psychology, 11(1), 5?26. https://doi.org/10.1207/s15327876mp1101_2 Siebold, G. L. (2007). The Essence of Military Group Cohesion. Armed Forces & Society, 33(2), 286?295. https://doi.org/10.1177/0095327X06294173 Siebold, G. L. (2011). Key Questions and Challenges to the Standard Model of Military Group Cohesion. Armed Forces & Society, 37(3), 448?468. https://doi.org/10.1177/0095327X11398451 Siebold, G. L., & Lindsay, T. J. (1999). The Relation Between Demographic Descriptors and Soldier-Perceived Cohesion and Motivation. Military Psychology, 11(1), 109?128. https://doi.org/10.1207/s15327876mp1101_6 Sinno, A. H. (2011). Organizations at War in Afghanistan and Beyond. (1 ed.). Cornell University Press. Sly, L. (2013). Foreign Extremists Dominate Syria Fight. The Washington Post. Snyder, D., & Tilly, C. (1972). Hardship and Collective Violence in France, 1830 to 1960. American Sociological Review, 37 (5), 520-532. Sontag, D. (1998, December 13). There?s No Bossing a Democracy, Arafat Learns. The New York Times. Available from: https://www.nytimes.com/1998/12/13/world/there-s-no-bossing-a-democracy- arafat-learns.html. 217 Speckhard, A., & Ahkmedova, K. (2006). The making of a martyr: Chechen suicide terrorism. Studies in Conflict & Terrorism, 29(5), 429?492. https://doi.org/10.1080/10576100600698550 Staniland, P. (2012). Between a Rock and a Hard Place: Insurgent Fratricide, Ethnic Defection, and the Rise of Pro-State Paramilitaries. Journal of Conflict Resolution, 56 (1), 16-40. https://doi.org/10.1177/0022002711429681 Staniland, P. (2014). Networks of Rebellion: Explaining Insurgent Cohesion and Collapse. Cornell University Press. Stedman, S. J. (1997). Spoiler Problems in Peace Processes. International Security, 22(2), 5?53. https://doi.org/10.1162/isec.22.2.5 Stein, A. A. (1976). Conflict and Cohesion: A Review of the Literature. The Journal of Conflict Resolution, 20(1), 143?172. https://doi.org/10.1177/002200277602000106 Stepanova, E. A. (2008). Terrorism in Asymmetrical Conflict: Ideological and Structural Aspects. New York: Oxford University Press. Stern, M. (2022a). Factionalisation from Below: The Case of Palestinian Fatah. Civil Wars, 1?46. https://doi.org/10.1080/13698249.2022.2015194 Stern, M. (2022b). Rebel Socialization, Military Training, and Group Cohesion. [Unpublished manuscript] Stewart, F. (Ed.). (2008). Horizontal inequalities and conflict: Understanding group violence in multiethnic societies. New York: Palgrave Macmillan. Stewart, N. K. (1991). Mates & muchachos: Unit cohesion in the Falklands/Malvinas War. Potomac Books. 218 Stouffer, S. A., Suchman, E. A., Devinney, L. C., Star, S. A., & Williams Jr., R. M. (1949). The American soldier: Adjustment during army life. (Studies in social psychology in World War II). Princeton Univ. Press. Strachan, H. (2006). Training, Morale and Modern War. Journal of Contemporary History, 41(2), 211?227. https://doi.org/10.1177/0022009406062054 Strindberg, A. (2000). The Damascus-Based Alliance of Palestinian Forces: A Primer. Journal of Palestine Studies, 29 (3), 60-76. Tamm, H. (2016). Rebel Leaders, Internal Rivals, and External Resources: How State Sponsors Affect Insurgent Cohesion. International Studies Quarterly, 60 (4), 599?610. https://doi.org/10.1093/isq/sqw033 Tamm, H. (2019). In the balance: External troop support and rebel fragmentation in the Second Congo War. Journal of Strategic Studies, 1?28. https://doi.org/10.1080/01402390.2019.1701442 The Popular Resistance Committees: Portrait of the Terrorist Organisation Responsible for the Series of Combined Terrorist Attacks North of Eilat, Israel?s Southernmost City (2011). The Meir Amit Intelligence and Terrorism Information Center. Available from: https://www.terrorism- info.org.il/Data/pdf/PDF_11_182_2.pdf. Thomson, D. (1990). Europe since Napoleon. Penguin Books. UCDP - Uppsala Conflict Data Program. (n.d.). Retrieved May 20, 2022, from https://ucdp.uu.se/actor/507 Usher, G. (2000). Fatah?s Tanzim: Origins and Politics. Middle East Report, (217), 6- 7. 219 Van Epps, G. (2008). Relooking unit cohesion: A sensemaking approach. Military Review, 88(6), 102?110. Vogt, M., Bormann, N.-C., R?egger, S., Cederman, L.-E., Hunziker, P., & Girardin, L. (2015). Integrating Data on Ethnicity, Geography, and Conflict: The Ethnic Power Relations Data Set Family. Journal of Conflict Resolution, 59 (7), 1327?1342. Walsh, J. I., Conrad, J. M., Whitaker, B. E., & Hudak, K. M. (2018). Funding rebellion: The rebel contraband dataset. Journal of Peace Research, 55(5), 699?707. https://doi.org/10.1177/0022343317740621 Walter, B. F. (2017). The new new civil wars. Annual Review of Political Science, 20(1), 469?486. https://doi.org/10.1146/annurev-polisci-060415-093921 Walter, B. F. (2019). Explaining the Number of Rebel Groups in Civil Wars. International Interactions, 45 (1), 1-27. Walther, O. J., & Pedersen, P. S. (2020). Rebel fragmentation in Syria?s civil war. Small Wars & Insurgencies, 31(3), 445?474. https://doi.org/10.1080/09592318.2020.1726566 Warren, T. C., & Troy, K. (2015). Explaining Violent Intra-Ethnic Conflict: Group Fragmentation in the Shadow of State Power. Journal of Conflict Resolution, 59 (3), 484?509. https://doi.org/10.1177/0022002713515400 Weidmann, N. B. (2009). Geography as Motivation and Opportunity: Group Concentration and Ethnic Conflict. Journal of Conflict Resolution, 53 (4), 526- 543. https://doi.org/10.1177/0022002709336456 220 Weinstein, J. M., (2007). Inside Rebellion: The Politics of Insurgent Violence. New York: Cambridge University Press. https://doi.org/10.1017/CBO9780511808654 Wesbrook, S. D. (1980). The Potential for Military Disintegration. In S. C. Sarkesian (Ed.), Combat Effectiveness and Cohesion (pp. 244?278). Sage Publications. Woldemariam, M. (2016). Battlefield outcomes and rebel cohesion: Lessons from the Eritrean independence war. Terrorism and Political Violence, 28(1), 135?156. https://doi.org/10.1080/09546553.2014.886575 Woldemariam, M. (2018). Insurgent Fragmentation in the Horn of Africa: Rebellion and its Discontents. Cambridge: Cambridge University Press. https://doi.org/https://doi.org/10.1017/9781108525657 Wong, L. (2006). Combat Motivation in Today?s Soldiers: U. S. Army War College Strategic Studies Institute. Armed Forces & Society, 32(4), 659?663. https://doi.org/10.1177/0095327X06287884 Wong, L., Kolditz, T. A., Millen, R. A., & Porter, T. M. (2003). Why They Fight: Combat Motivation in the Iraq War. Strategic Studies Institute, U.S. Army War College. Wood, E. J. (2009). Armed Groups and Sexual Violence: When is Wartime Rape Rare? Politics & Society, 37(1), 131?161. https://doi.org/10.1177/0032329208329755 Wood, E. J. (2018). Rape as a Practice of War: Toward a Typology of Political Violence. Politics & Society, 46(4), 513?537. https://doi.org/10.1177/0032329218773710 221 Wood, E. J., & Jean, W. E. (2003). Insurgent collective action and civil war in El Salvador. Cambridge University Press. Wood, E. J., & Toppelberg, N. (2017). The persistence of sexual assault within the US military. Journal of Peace Research, 54(5), 620?633. https://doi.org/10.1177/0022343317720487 Wucherpfennig, J., Metternich, N. W., Cederman, L.-E., & Gleditsch, K. S. (2012). Ethnicity, the state, and the duration of civil war. World Politics, 64(1), 79? 115. https://doi.org/10.1017/S004388711100030X Yaghi, M. (2009). Fatah Congress: A Victory for Abbas, The Washington Institute for Near East Policy. Available from: https://www.washingtoninstitute.org/policy- analysis/view/fatah-congress-a-victory-for-abbas. Zaks, S. (2017). Relationships Among Rivals (RAR): A Framework for Analyzing Contending Hypotheses in Process Tracing. Political Analysis, 25 (3), 344- 362. 222